Agisoft Metashape
Agisoft Metashape => General => Topic started by: Dmitry Semyonov on October 06, 2018, 09:16:20 PM
-
Pre-release version of Agisoft Metashape 1.5.0 (former PhotoScan) is available for download.
Metashape inherits all the professional photogrammetric kit features from PhotoScan and offers some more to enrich your experience into the space of intelligent photogrammetry.
The release is planned for the end of 2018. Product update will be free and smooth as usual PhotoScan version updates.
This pre-release is considered as unstable and is not recommended for production use.
To try cloud processing support (requires Professional edition license) please register at account.agisoft.com (https://account.agisoft.com).
Please let us know if you have any problems using this version.
This topic will be removed when the final 1.5.0 version is released.
Note
Project files created with 1.5.0 version can't be opened with earlier versions.
EDIT:
Metashape 1.5.0 build 7492 was finally released. Thank you for useful feedback.
Please use the links on the main Metashape page for download:
Download Installers (http://www.agisoft.com/downloads/installer/)
-
I tried to register for Cloud Processing but I meet a problem in the register page.
I checked the "I agree with User's agreement" checkbox
I used Chrome and Internet Explorer, both of them shows the message asking me to agree although I have already checked the box.
I can do it with Microsoft Edge ^_^
-
Hello rexsham,
Thank you for reporting. We'll check what is the issue.
-
What's the meaning of "Refine Seamlines" ?
-
Is there a readme-File for the planed or already included changes, compared to the aktual version?
--> Done. I was blind. Sorry
-
@3DFranz
The changelog file tells you what has been included/changed
http://www.agisoft.com/pdf/metashape_changelog.pdf
Regards,
SAV
-
I think that in Animation the user should have the option to define the view height.
-
Any additional information on the Cloud Processing functionality?
Will this be a 'free' feature available to those possessing a Professional License or will it be paid add-on? Besides what are the anticipated advantages for user's to process in the cloud? Is it expected to shorten processing time or only for those user's that are running a low-end system locally themselves? How will it compare to processing locally using a medium to high-end workstation, will it still be faster then?
Also, what about user control over the processing workflow? Does one remain fully in control of all processing steps and subsequent parameters or is all of the processing done autonomously by the system itself based on default pre-defined settings?
-
Hello BobvdMeij,
Technically the cloud processing is similar to the network processing - you enable the Cloud in the Network preferences dialog, then open the project and configure the batch process or run some processing operation. After that (if the operation supports processing on a cloud, for example, the export can be performed locally only) you'll be asked, if the operation should be processed in the cloud. Providing that you accept that, the data will be first synchronized with the cloud and then processed with the user-defined settings. After the operation is completed the project will be synchronized with the cloud again.
You do not need to have Metashape GUI opened while the cloud processing is in progress, but in order to synchronize the project with the cloud you need to open the same project in GUI again.
In terms of the processing time - it is hard to say, as some time will be spent for the data synchronization and it could be quite long in case of very large projects. So mostly this feature would might be helpful for those who cannot process the data locally due to hardware limitations.
Currently the cloud is working in test mode and is free for the Professional license owners. However, after the testing is finished the cloud processing would be likely converted to paid service with the plans configurable in the user's account on the Agisoft Cloud page.
-
Dear Dmitry,
i am trying to do orthomosaic with option Refine Seamlines activated. After processing, I get error message and nothing is creted:
2018-10-08 10:31:21 Agisoft Metashape Professional Version: 1.5.0 build 6957 (64 bit)
2018-10-08 10:31:21 Platform: Windows
2018-10-08 10:31:21 CPU: Intel(R) Core(TM) i7-3630QM CPU @ 2.40GHz (laptop)
2018-10-08 10:31:21 CPU family: 6 model: 58 signature: 306A9h
2018-10-08 10:31:21 RAM: 16.0 GB
2018-10-08 10:31:22 OpenGL Vendor: NVIDIA Corporation
2018-10-08 10:31:22 OpenGL Renderer: GeForce GTX 670MX/PCIe/SSE2
2018-10-08 10:31:22 OpenGL Version: 4.6.0 NVIDIA 411.63
2018-10-08 10:31:22 Maximum Texture Size: 16384
2018-10-08 10:31:22 Quad Buffered Stereo: not enabled
2018-10-08 10:31:22 ARB_vertex_buffer_object: supported
2018-10-08 10:31:22 ARB_texture_non_power_of_two: supported
2018-10-08 10:31:22 Warning: PhotoScan module is deprecated, use Metashape module instead
2018-10-08 10:31:52 LoadProject: path = C:/Users/paul.pelletier/Documents/3_Deseos/3deseos15.psx
2018-10-08 10:31:52 Loading project...
2018-10-08 10:31:52 loaded project in 0.461 sec
2018-10-08 10:31:52 Finished processing in 0.465 sec (exit code 1)
2018-10-08 10:34:01 Finished processing in 0 sec (exit code 1)
2018-10-08 10:34:40 Finished processing in 0 sec (exit code 1)
2018-10-08 10:35:13 checking for missing images... done in 0 sec
2018-10-08 10:35:13 Finished processing in 0 sec (exit code 1)
2018-10-08 10:35:13 BuildOrthomosaic: projection = WGS 84 / UTM zone 14N + GGM2010 geoid height, surface = DEM, blending mode = Mosaic, refine seamlines = 1, pixel size = 0.07 x 0.07
2018-10-08 10:35:13 Generating orthomosaic...
2018-10-08 10:35:13 Analyzing DEM...
2018-10-08 10:35:13 estimating tile boundaries... done in 0.769 sec
2018-10-08 10:35:14 generating 37928x25257 orthomosaic (10 levels, 0 resolution)
2018-10-08 10:35:14 selected 265 cameras
2018-10-08 10:35:14 Orthorectifying images...
2018-10-08 10:35:14 Orthorectifying 265 images
2018-10-08 10:35:57 IMG_4237.JPG: 7314x9216 -> 689x350
2018-10-08 10:36:03 IMG_4238.JPG: 9216x9216 -> 956x925
2018-10-08 10:36:33 IMG_4278.JPG: 7314x8614 -> 701x1102
2018-10-08 10:36:35 IMG_4280.JPG: 7314x7314 -> 1894x1722
2018-10-08 10:36:35 IMG_4279.JPG: 7314x8614 -> 869x879
2018-10-08 10:36:39 IMG_4282.JPG: 9216x7314 -> 2352x1709
2018-10-08 10:36:40 IMG_4284.JPG: 9216x7314 -> 2670x2103
2018-10-08 10:36:40 IMG_4285.JPG: 7314x9216 -> 2819x2410
2018-10-08 10:36:40 IMG_4283.JPG: 9216x7314 -> 2569x2048
2018-10-08 10:36:40 IMG_4281.JPG: 9216x9216 -> 2143x1739
2018-10-08 10:36:46 IMG_4286.JPG: 9216x9216 -> 2224x1703
2018-10-08 10:36:49 IMG_4287.JPG: 9216x9216 -> 2470x1965
2018-10-08 10:36:49 IMG_4288.JPG: 9216x9216 -> 2303x1831
2018-10-08 10:36:51 IMG_4289.JPG: 7314x9216 -> 2546x1964
2018-10-08 10:36:51 IMG_4292.JPG: 9216x7314 -> 2791x2301
2018-10-08 10:36:52 IMG_4291.JPG: 9216x7314 -> 2244x1490
2018-10-08 10:36:52 IMG_4293.JPG: 9216x7314 -> 2556x2120
2018-10-08 10:36:53 IMG_4290.JPG: 9216x9216 -> 1939x1488
2018-10-08 10:36:56 IMG_4294.JPG: 7314x7314 -> 2463x1740
2018-10-08 10:36:59 IMG_4295.JPG: 7314x7314 -> 2266x1662
2018-10-08 10:37:01 IMG_4296.JPG: 7314x9216 -> 2576x2010
2018-10-08 10:37:02 IMG_4300.JPG: 7315x7314 -> 674x773
2018-10-08 10:37:03 IMG_4298.JPG: 7315x9216 -> 1539x943
2018-10-08 10:37:04 IMG_4299.JPG: 7315x9216 -> 1421x1230
2018-10-08 10:37:05 IMG_4297.JPG: 9216x9216 -> 1733x1440
2018-10-08 10:37:11 IMG_4335.JPG: 7315x7314 -> 391x789
2018-10-08 10:37:14 IMG_4336.JPG: 9216x7314 -> 1721x2107
2018-10-08 10:37:15 IMG_4337.JPG: 9216x7314 -> 2057x2238
2018-10-08 10:37:16 IMG_4339.JPG: 7314x9216 -> 2509x2321
2018-10-08 10:37:18 IMG_4338.JPG: 9216x9216 -> 2790x2369
2018-10-08 10:37:19 IMG_4341.JPG: 7314x7314 -> 2625x2318
2018-10-08 10:37:21 IMG_4340.JPG: 7314x9216 -> 3072x2634
2018-10-08 10:37:23 IMG_4342.JPG: 9216x7314 -> 3028x2546
2018-10-08 10:37:24 IMG_4344.JPG: 7314x7314 -> 3075x2719
2018-10-08 10:37:24 IMG_4345.JPG: 7314x7314 -> 2577x2283
2018-10-08 10:37:25 IMG_4343.JPG: 9216x7314 -> 2773x2549
2018-10-08 10:37:28 IMG_4347.JPG: 7314x7314 -> 2548x2351
2018-10-08 10:37:28 IMG_4346.JPG: 9216x7314 -> 2946x2630
2018-10-08 10:37:31 IMG_4348.JPG: 7314x9216 -> 2852x2472
2018-10-08 10:37:33 IMG_4349.JPG: 7314x9216 -> 3164x2695
2018-10-08 10:37:34 IMG_4351.JPG: 7314x7314 -> 2468x2397
2018-10-08 10:37:35 IMG_4353.JPG: 7314x7314 -> 2948x2978
2018-10-08 10:37:36 IMG_4350.JPG: 7314x9216 -> 3806x3220
2018-10-08 10:37:38 IMG_4355.JPG: 7314x7314 -> 2243x2367
2018-10-08 10:37:39 IMG_4352.JPG: 9216x9216 -> 3230x2756
2018-10-08 10:37:40 IMG_4354.JPG: 9216x7314 -> 2815x2396
2018-10-08 10:37:40 IMG_4356.JPG: 7314x7314 -> 1351x1723
2018-10-08 10:37:50 IMG_4365.JPG: 7314x7314 -> 2786x3386
2018-10-08 10:37:52 IMG_4364.JPG: 9216x9216 -> 2096x3265
2018-10-08 10:37:52 IMG_4366.JPG: 7314x9216 -> 3488x3752
2018-10-08 10:37:54 IMG_4367.JPG: 7314x9216 -> 3791x4110
2018-10-08 10:37:58 IMG_4368.JPG: 9216x9216 -> 3759x4132
2018-10-08 10:37:59 IMG_4369.JPG: 7314x9216 -> 3859x3924
2018-10-08 10:38:00 IMG_4371.JPG: 7314x7314 -> 4107x3668
2018-10-08 10:38:00 IMG_4370.JPG: 7314x9216 -> 3799x4153
2018-10-08 10:38:00 IMG_4372.JPG: 7314x7314 -> 3252x3916
2018-10-08 10:38:04 IMG_4373.JPG: 7314x9216 -> 4037x4241
2018-10-08 10:38:08 IMG_4374.JPG: 9216x9216 -> 4052x4302
2018-10-08 10:38:10 IMG_4375.JPG: 9216x9216 -> 4122x3911
2018-10-08 10:38:11 IMG_4376.JPG: 7314x9216 -> 3402x3988
2018-10-08 10:38:11 IMG_4378.JPG: 7314x7314 -> 3234x3945
2018-10-08 10:38:14 IMG_4380.JPG: 9216x7314 -> 3317x3940
2018-10-08 10:38:14 IMG_4377.JPG: 9216x9216 -> 4151x3717
2018-10-08 10:38:15 IMG_4381.JPG: 7314x7314 -> 3927x4192
2018-10-08 10:38:16 IMG_4379.JPG: 9216x9216 -> 4031x3904
2018-10-08 10:38:20 IMG_4382.JPG: 7314x9216 -> 3392x4032
2018-10-08 10:38:21 IMG_4383.JPG: 7314x7314 -> 3523x3950
2018-10-08 10:38:23 IMG_4384.JPG: 9216x7314 -> 2447x3308
2018-10-08 10:38:25 IMG_4385.JPG: 9216x9216 -> 2033x2950
2018-10-08 10:38:33 IMG_4412.JPG: 7315x9216 -> 491x988
2018-10-08 10:38:35 IMG_4414.JPG: 7314x9216 -> 2589x3527
2018-10-08 10:38:36 IMG_4416.JPG: 7314x7314 -> 3105x3780
2018-10-08 10:38:36 IMG_4417.JPG: 7314x7314 -> 3482x4032
2018-10-08 10:38:36 IMG_4415.JPG: 7314x9216 -> 3285x3793
2018-10-08 10:38:37 IMG_4413.JPG: 9216x9216 -> 2082x3291
2018-10-08 10:38:38 IMG_4418.JPG: 9216x7314 -> 3169x3860
2018-10-08 10:38:41 IMG_4419.JPG: 9216x7314 -> 3385x3988
2018-10-08 10:38:43 IMG_4420.JPG: 7314x7314 -> 3425x4031
2018-10-08 10:38:45 IMG_4421.JPG: 7314x7314 -> 3238x3827
2018-10-08 10:38:47 IMG_4425.JPG: 7314x7314 -> 3041x3811
2018-10-08 10:38:49 IMG_4423.JPG: 7314x9216 -> 3276x4051
2018-10-08 10:38:49 IMG_4424.JPG: 7314x9216 -> 3757x4041
2018-10-08 10:38:51 IMG_4422.JPG: 9216x9216 -> 3445x3936
2018-10-08 10:38:51 IMG_4426.JPG: 7314x9216 -> 3507x4024
2018-10-08 10:38:51 IMG_4427.JPG: 7314x7314 -> 3355x3946
2018-10-08 10:38:56 IMG_4429.JPG: 7314x7314 -> 3140x3889
2018-10-08 10:38:56 IMG_4428.JPG: 9216x7314 -> 3483x3939
2018-10-08 10:38:58 IMG_4430.JPG: 7314x7314 -> 3292x3933
2018-10-08 10:39:00 IMG_4434.JPG: 6072x9216 -> 1381x2673
2018-10-08 10:39:01 IMG_4431.JPG: 7314x9216 -> 3492x3865
2018-10-08 10:39:02 IMG_4432.JPG: 7314x9216 -> 2949x3467
2018-10-08 10:39:03 IMG_4433.JPG: 7314x9216 -> 2084x3170
2018-10-08 10:39:10 IMG_4442.JPG: 9216x7314 -> 2002x2925
2018-10-08 10:39:13 IMG_4443.JPG: 9216x7314 -> 2720x3466
2018-10-08 10:39:14 IMG_4446.JPG: 7314x7314 -> 3665x4048
2018-10-08 10:39:14 IMG_4445.JPG: 9216x7314 -> 3834x4060
2018-10-08 10:39:15 IMG_4444.JPG: 9216x9216 -> 3208x4228
2018-10-08 10:39:17 IMG_4447.JPG: 7314x9216 -> 3770x4131
2018-10-08 10:39:21 IMG_4448.JPG: 9216x9216 -> 4003x4355
2018-10-08 10:39:22 IMG_4449.JPG: 9216x9216 -> 3982x4096
2018-10-08 10:39:23 IMG_4450.JPG: 7314x9216 -> 3750x4122
2018-10-08 10:39:23 IMG_4451.JPG: 7314x7314 -> 3970x4196
2018-10-08 10:39:24 IMG_4452.JPG: 7314x7314 -> 3717x4052
2018-10-08 10:39:27 IMG_4453.JPG: 9216x7314 -> 3928x4130
2018-10-08 10:39:28 IMG_4454.JPG: 9216x7314 -> 3699x4042
2018-10-08 10:39:32 IMG_4455.JPG: 9216x9216 -> 3899x4198
2018-10-08 10:39:34 IMG_4456.JPG: 9216x7314 -> 3904x4091
2018-10-08 10:39:35 IMG_4457.JPG: 7314x9216 -> 3991x4137
2018-10-08 10:39:36 IMG_4458.JPG: 7314x9216 -> 3729x4159
2018-10-08 10:39:37 IMG_4462.JPG: 7314x6971 -> 3148x3281
2018-10-08 10:39:38 IMG_4459.JPG: 9216x9216 -> 3943x4099
2018-10-08 10:39:39 IMG_4461.JPG: 7314x9216 -> 3940x3945
2018-10-08 10:39:39 IMG_4460.JPG: 9216x9216 -> 3773x4049
2018-10-08 10:39:42 IMG_4463.JPG: 7314x6971 -> 2739x2899
2018-10-08 10:39:45 IMG_4464.JPG: 9216x6972 -> 1241x1940
2018-10-08 10:39:47 IMG_4484.JPG: 7314x6971 -> 1281x1386
2018-10-08 10:39:47 IMG_4483.JPG: 7314x6971 -> 334x683
2018-10-08 10:39:49 IMG_4485.JPG: 7314x6971 -> 2546x3286
2018-10-08 10:39:52 IMG_4486.JPG: 7314x6971 -> 3165x3483
2018-10-08 10:39:52 IMG_4487.JPG: 7314x6971 -> 2899x3657
2018-10-08 10:39:54 IMG_4489.JPG: 7314x6971 -> 3323x3839
2018-10-08 10:39:55 IMG_4488.JPG: 9216x6972 -> 3582x4014
2018-10-08 10:39:58 IMG_4490.JPG: 7314x9216 -> 3417x3902
2018-10-08 10:39:59 IMG_4491.JPG: 7314x9216 -> 3311x3916
2018-10-08 10:40:00 IMG_4493.JPG: 7314x7314 -> 3079x3761
2018-10-08 10:40:01 IMG_4492.JPG: 9216x9216 -> 3616x3985
2018-10-08 10:40:03 IMG_4494.JPG: 7314x7314 -> 3500x4002
2018-10-08 10:40:03 IMG_4495.JPG: 7314x7314 -> 3378x3918
2018-10-08 10:40:05 IMG_4496.JPG: 7314x7314 -> 3444x3915
2018-10-08 10:40:06 IMG_4497.JPG: 7314x7314 -> 3147x3864
2018-10-08 10:40:10 IMG_4498.JPG: 9216x7314 -> 3565x3991
2018-10-08 10:40:11 IMG_4499.JPG: 7314x9216 -> 2964x3692
2018-10-08 10:40:12 IMG_4500.JPG: 7314x9216 -> 3408x3850
2018-10-08 10:40:14 IMG_4501.JPG: 7314x9216 -> 3039x3707
2018-10-08 10:40:14 IMG_4502.JPG: 7314x7314 -> 3371x3771
2018-10-08 10:40:14 IMG_4504.JPG: 7314x7314 -> 2246x3222
2018-10-08 10:40:15 IMG_4505.JPG: 7314x7314 -> 1272x2442
2018-10-08 10:40:15 IMG_4503.JPG: 9216x7314 -> 2931x3421
2018-10-08 10:40:22 IMG_4515.JPG: 7314x6029 -> 2827x3429
2018-10-08 10:40:23 IMG_4513.JPG: 7314x7771 -> 1281x1721
2018-10-08 10:40:24 IMG_4514.JPG: 7314x9216 -> 2603x3438
2018-10-08 10:40:28 IMG_4516.JPG: 9216x9216 -> 3695x3983
2018-10-08 10:40:28 IMG_4518.JPG: 7314x7314 -> 3620x4001
2018-10-08 10:40:28 IMG_4517.JPG: 9216x6029 -> 4134x3814
2018-10-08 10:40:30 IMG_4519.JPG: 7314x7314 -> 3941x3854
2018-10-08 10:40:31 IMG_4520.JPG: 7314x7314 -> 3838x4031
2018-10-08 10:40:31 IMG_4521.JPG: 7314x7314 -> 3958x3918
2018-10-08 10:40:35 IMG_4522.JPG: 9216x7314 -> 3871x4076
2018-10-08 10:40:35 IMG_4523.JPG: 7314x8057 -> 3745x3947
2018-10-08 10:40:39 IMG_4526.JPG: 9216x6300 -> 3917x3905
2018-10-08 10:40:40 IMG_4524.JPG: 7314x9216 -> 3942x4034
2018-10-08 10:40:41 IMG_4528.JPG: 7314x6971 -> 4050x3762
2018-10-08 10:40:42 IMG_4529.JPG: 9216x6972 -> 3824x4076
2018-10-08 10:40:42 IMG_4527.JPG: 7314x9216 -> 3867x4110
2018-10-08 10:40:42 IMG_4525.JPG: 9216x9216 -> 3904x4111
2018-10-08 10:40:44 IMG_4530.JPG: 7314x6971 -> 3964x3870
2018-10-08 10:40:45 IMG_4531.JPG: 7314x6971 -> 3668x4102
2018-10-08 10:40:47 IMG_4536.JPG: 6671x3314 -> 2618x2501
2018-10-08 10:40:48 IMG_4533.JPG: 7314x6971 -> 2448x3345
2018-10-08 10:40:49 IMG_4534.JPG: 7314x6971 -> 2206x2577
2018-10-08 10:40:49 IMG_4532.JPG: 9216x6971 -> 3095x3784
2018-10-08 10:40:51 IMG_4535.JPG: 7314x6971 -> 2387x3103
2018-10-08 10:40:52 IMG_4537.JPG: 7314x6971 -> 2971x3461
2018-10-08 10:40:53 IMG_4540.JPG: 3657x6300 -> 2747x3045
2018-10-08 10:40:54 IMG_4539.JPG: 6671x6971 -> 3035x3046
2018-10-08 10:40:54 IMG_4538.JPG: 7314x6971 -> 2798x3219
2018-10-08 10:40:56 IMG_4541.JPG: 7314x6300 -> 2930x3495
2018-10-08 10:40:56 IMG_4542.JPG: 7314x6300 -> 2998x3321
2018-10-08 10:40:57 IMG_4543.JPG: 7314x6300 -> 2939x3266
2018-10-08 10:40:58 IMG_4546.JPG: 7314x3657 -> 2816x3051
2018-10-08 10:40:58 IMG_4544.JPG: 7314x4400 -> 2923x3379
2018-10-08 10:40:59 IMG_4545.JPG: 7314x4400 -> 2847x3213
2018-10-08 10:41:03 IMG_4547.JPG: 7314x7314 -> 2867x3107
2018-10-08 10:41:03 IMG_4548.JPG: 7314x7314 -> 2791x3218
2018-10-08 10:41:04 IMG_4554.JPG: 7314x4114 -> 591x1209
2018-10-08 10:41:05 IMG_4552.JPG: 7314x6029 -> 2105x2611
2018-10-08 10:41:05 IMG_4550.JPG: 7314x6029 -> 2888x3243
2018-10-08 10:41:05 IMG_4549.JPG: 7314x7314 -> 2911x3010
2018-10-08 10:41:07 IMG_4551.JPG: 9216x6029 -> 3073x3017
2018-10-08 10:41:08 IMG_4553.JPG: 7314x9216 -> 1832x2658
2018-10-08 10:41:11 IMG_4585.JPG: 7314x6971 -> 832x1121
2018-10-08 10:41:11 Finished orthorectification in 357.137 sec
2018-10-08 10:41:11 selected 21 tiles
2018-10-08 10:41:11 selected 21 tiles
2018-10-08 10:41:11 Updating partition...
2018-10-08 10:41:11 8 of 8 processed in 0.055 sec
2018-10-08 10:41:12 Finished processing in 358.84 sec (exit code 0)
2018-10-08 10:41:12 Error: Can't create file: El sistema no puede encontrar la ruta especificada (3): /seamlines/tile-0-3-p.png
I guess my problem is that the Metashape Python 3 module is not installed. How do we install such a module after downloading the file Metashape-1.5.0-cp35.cp36.cp37-none-win_amd64.man for Windows 64?
-
Hello Paul,
Does the orthomosaic generation works properly on the same project with the "Refine Seamlines" option disabled? Also please check if the processing is running properly, if you enable Fine-level Task distribution for orthomosaic generation in the Network Preferences tab.
And you do not need to install Metashape Python module, it's only necessary for integration of Metashape functionality to some custom stand-alone Python scripts in order to work without a need of running Metashape application (neither in GUI, nor in headless mode).
-
Dear Alexey,
Yes I do get same error without Refine Seamlines option,,,see attached log file
-
Hello Paulo,
OK, it seems I've managed to reproduce the problem. Meanwhile I suggest to enable Fine-level task distribution for the orthomosaic generation in the Network Preferences tab and see, if it helps.
-
Thanks Alexey,
ye I did try with fine level task for ortomosaic and got a result using Refine Seamlines option. Of course it takes a longer time as it does many iterations to refine the seam lines…
However the result is not great as I would expect it to avoid building structures in seam lines path but in attached screen copy, it shows that it passes thru a lot of buildings….
-
Hello Paul,
Refine Seamlines feature in general case assumes that the orthorectification is based on the DTM. Can you post the screenshot for the same area of the DEM view?
-
YES Alexey,
I used a DTM as base for orthorectifying...here is screen shot
-
Can we have more info about the Metashape Python 3 module? What's its use, how can we install it if it must check the license?
I tried to install it in a virtualenv (Python 3.6.6) under Ubuntu 16.04 (without Metashape) but maybe dependencies and/or environment variables are missing:
In [1]: import Metashape
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
<ipython-input-1-ec7fe42b313a> in <module>()
----> 1 import Metashape
~/.virtualenvs/metashape/lib/python3.6/site-packages/Metashape/__init__.py in <module>()
1 import importlib
2
----> 3 from .Metashape import *
4
5 # copy private variables
ImportError: ~/.virtualenvs/metashape/lib/python3.6/site-packages/Metashape/Metashape.so: undefined symbol: glColor3dv
-
I signed up for an account and have started processing but I cannot find any status on the project, other than that it is processing. At https://account.agisoft.com nothing happens when pressing General, Projects or Payments, I can only change my name and password.
-
I tried Metashape, and first realized that it removed Photoscan (1.4.4). I thought it could be installed parallel.
Anyway, I realized you cannot export Orthomosaics defined by 3-markers-coordinatesystem anymore?
I ended up deinstalling Metashape and reinstalling Photoscan 1.4.4.
But now I cannot export Orthomosaic in coordinate-systems (or self-defined via markers) anymore, even in Photoscan??
Am I doing sth wrong? I cannot choose anything under "coordinate system" at the top...
edit: Nevermind, I got it sorted.. Realized the definition has to go in Orthomosaic-Creation-Wizard (and not at the export)
-
I signed up for an account and have started processing but I cannot find any status on the project, other than that it is processing. At https://account.agisoft.com nothing happens when pressing General, Projects or Payments, I can only change my name and password.
Hello Kjellis85,
Currently Cloud is working in the test mode, so other personal account page tabs are not yet accessible.
-
Can we have more info about the Metashape Python 3 module? What's its use, how can we install it if it must check the license?
Hello Gall,
The issue with Metashape module installation will be fixed in the next pre-release update.
The module itself requires to have Metashape Pro (or PhotoScan Pro) activated on the same computer, as it uses the same license.
-
Great to finally have a new pre-release version to test! But sad that there doesn't seem to be any new performance related features.
Cheers
Mak
-
I have tested the cloud computing and there seems to be some bug in the pipeline. Not sure if it has finished processing yet, because it is less than a day since I started processing.
-
Orthomosaic : Refine Seamlines.
I also have a problem when building Orthomosaic with Refine Seamlines enabled.
I always get : "Bad Allocation".
...
...
11:08:44 partition updated in 221.71 sec
2018-10-10 11:08:44 loaded camera partition in 0 sec
2018-10-10 11:08:44 Refining seamlines...
2018-10-10 11:08:44 Building topology...
2018-10-10 11:24:43 Finished processing in 2090.34 sec (exit code 0)
2018-10-10 11:24:43 Error: bad allocation
-
Hello Kjellis85,
It seems that the issue appeared during the internal synchronization of the already processed project (took almost one day to complete), we are fixing the problem now.
-
Orthomosaic : Refine Seamlines.
I also have a problem when building Orthomosaic with Refine Seamlines enabled.
I always get : "Bad Allocation".
...
...
11:08:44 partition updated in 221.71 sec
2018-10-10 11:08:44 loaded camera partition in 0 sec
2018-10-10 11:08:44 Refining seamlines...
2018-10-10 11:08:44 Building topology...
2018-10-10 11:24:43 Finished processing in 2090.34 sec (exit code 0)
2018-10-10 11:24:43 Error: bad allocation
Hello 3DFranz,
Can you provide the complete log (as text file) of the Build Orthomosaic operation and also specify the project properties: number and resolution of images, effective overlap (average tie point multiplicity), orthomosaic and source DEM resolution used. Also check that you have 64-bit version of Metashape Pro installed and specify the system configuration (CPU, GPU, RAM, OS version).
-
Tried another run on the cloud with an unprocessed project, but it returned the attatched error message.
-
In new version, there are some new interesting features regarding variance information:
1. in point cloud model display, there is an option to show Point cloud variance. In model view, the point cloud is colored according to each tie point variance. How is this variance calculated and what does it represent?
2. in Reference pane, there is a new button to View variance of camera positions and orientations. Same question as 1. how are these variances calculated and what do they represent?
Some more insight in these features would be greatly appreciated,
-
Hello Paulo,
The variance for the tie points is calculated from the covariance matrix obtained at the end of the bundle adjustment procedure. It represents the uncertainty measure of estimated tie point coordinates.
The variance for the camera locations and orientation angles is calculated in the same way as described above.
-
Tried another run on the cloud with an unprocessed project, but it returned the attatched error message.
Hello Kjellis85,
We will try to implemented some fixes on the cloud side shortly, so hopefully the issue will no longer appear. The problem was related to the large size of the project files.
-
Dear Alexey,
with latest release, I retried orthorectifying with refine Seamlines options. It works now without setting the Fine level in Network options....
However it still does not give satisfying results with seam lines passing thru buildings (marked in red on screen shot).
Also, the OrthoMosaic was done on a small region (shown in model view).
However, the ortho shows an extent of 37928 by 25257 pix at 7cm/pix which does not correspond to the extent of region … It corresponds to full extent of model...
Attached is the log of processing. Note that
2018-10-14 07:58:31 Error: Connection closed (status code 0)
corresponds to attempt at sending Ortho task to Cloud processing after enabling it in Preferences....
-
Hi Alexey,
Thank you for the new update, Is there any improvement in the process time?
Also what does ''Estimate tie point covariance'' mean?
Thanks
Mohammed
-
Most times the "Refine Seamlines" option leads to holes in the orthomosaic while no holes appear if the option is switched off. Analysing seamlines from both options, it's notorious that a much better result is produced with this new option. Mosaic tends to be a lot better without too much jagged limits on objects. However if holes are produced, it will be a lot less interesting to use.
PS: Alexey and Dmitry, excuse my curiosity in advance :). Photoscan is now a mature Brand and a perfectly well known name among photogrammetry users. In every single congress or meeting, when we talk about Photoscan, everyone knows the product and the outstanding final results and features that it offers. Why this change for Metashape? I like this new brand a lot also but Photoscan was simply "familiar".
Cheers
-
PS: Alexey and Dmitry, excuse my curiosity in advance :). Photoscan is now a mature Brand and a perfectly well known name among photogrammetry users. In every single congress or meeting, when we talk about Photoscan, everyone knows the product and the outstanding final results and features that it offers. Why this change for Metashape? I like this new brand a lot also but Photoscan was simply "familiar".
Cheers
Agree with you, photscan is very familiar to us.
-
PS: Alexey and Dmitry, excuse my curiosity in advance :). Photoscan is now a mature Brand and a perfectly well known name among photogrammetry users. In every single congress or meeting, when we talk about Photoscan, everyone knows the product and the outstanding final results and features that it offers. Why this change for Metashape? I like this new brand a lot also but Photoscan was simply "familiar".
Cheers
I really do not agree, the real name of the Brand is AGISOFT, not PhotoScan, so welcome the change of name, the important thing is the AGISOFT Brand, in fact google also uses the name PhotoScan https://www.google.com/photos/scan/ (https://www.google.com/photos/scan/)
Welcome the change :)
-
Hi Alexey,
How can i activate ''Point cloud variance''?
Thanks
-
You have to first calculate it by checking it in the optimizing dialog then it becomes available as a render mode for the sparse cloud.
Alexey > the refine seamline option doesn't seem to be multithreaded or using the gpu, is there a future evolution possible on this front ? I see a 10x increase in processing time when it is used.
-
Can we get some information about the auto classification being used? I ran a quick dataset through it and recieved very poor results. It seems strange to me that I am getting these results when I am using a very clean dataset.
Photos have 90% overlap shot at 100m agl as 20mp. I classified the dataset using the old tools in version 1.3 and was able to get very good results in a short period of time. I was hoping that the autoclassification would at least get close, but my results are extremely bad. Could I be missing something? I should add that I am running the 1.5 pro beta as a demo right now.
-
with latest release, I retried orthorectifying with refine Seamlines options. It works now without setting the Fine level in Network options....
However it still does not give satisfying results with seam lines passing thru buildings (marked in red on screen shot).
Also, the OrthoMosaic was done on a small region (shown in model view).
However, the ortho shows an extent of 37928 by 25257 pix at 7cm/pix which does not correspond to the extent of region … It corresponds to full extent of model...
Hello Paulo,
Would it be possible to send the project files (alignment results + DTM + source images) that we can use for testing the seamlines refinement issue and extent problems? If so, please send the download link to support@agisoft.com and mention this thread in the message body.
As for the connection to the cloud problems, then is there a proxy server used on your side?
-
Most times the "Refine Seamlines" option leads to holes in the orthomosaic while no holes appear if the option is switched off. Analysing seamlines from both options, it's notorious that a much better result is produced with this new option. Mosaic tends to be a lot better without too much jagged limits on objects. However if holes are produced, it will be a lot less interesting to use.
Hello RHenriques,
If there's any sample set of images that can be used to reproduce the problems with the orthomosaic stitching (e.g. holes in the mosaic), please also send the data to support@agisoft.com, so that we could analyze the source of the issue and fix it.
-
Can we get some information about the auto classification being used? I ran a quick dataset through it and recieved very poor results. It seems strange to me that I am getting these results when I am using a very clean dataset.
Photos have 90% overlap shot at 100m agl as 20mp. I classified the dataset using the old tools in version 1.3 and was able to get very good results in a short period of time. I was hoping that the autoclassification would at least get close, but my results are extremely bad. Could I be missing something? I should add that I am running the 1.5 pro beta as a demo right now.
Hello SuperNewb,
The new classification option is based on the machine learning, so if you are observing unsatisfying results, we would like to ask you to send the point cloud that you are working on to support@agisoft.com (just point cloud exported from the project will be fine, just keep the source coordinate system and point colors).
-
Can we get some information about the auto classification being used? I ran a quick dataset through it and recieved very poor results. It seems strange to me that I am getting these results when I am using a very clean dataset.
Photos have 90% overlap shot at 100m agl as 20mp. I classified the dataset using the old tools in version 1.3 and was able to get very good results in a short period of time. I was hoping that the autoclassification would at least get close, but my results are extremely bad. Could I be missing something? I should add that I am running the 1.5 pro beta as a demo right now.
Hello SuperNewb,
The new classification option is based on the machine learning, so if you are observing unsatisfying results, we would like to ask you to send the point cloud that you are working on to support@agisoft.com (just point cloud exported from the project will be fine, just keep the source coordinate system and point colors).
Just sent the email with a link to download the dense point surface.
Thanks
-
When running in network mode "Refine Seamlines" consume 100% of the memory of each node and crash on "Bad Allocation". it doesn't happen on the same project using a single computer.
-
Most times the "Refine Seamlines" option leads to holes in the orthomosaic while no holes appear if the option is switched off. Analysing seamlines from both options, it's notorious that a much better result is produced with this new option. Mosaic tends to be a lot better without too much jagged limits on objects. However if holes are produced, it will be a lot less interesting to use.
Hello RHenriques,
If there's any sample set of images that can be used to reproduce the problems with the orthomosaic stitching (e.g. holes in the mosaic), please also send the data to support@agisoft.com, so that we could analyze the source of the issue and fix it.
Later today I'll send you one of the sets where I've identified this problem. Are the pictures enough?
Cheers
-
Please make: "Reuse already created Orthorectified images" in the Orthomosaic menu in case that it fails the first time.
-
How to install along side existing installation?
If I remember correctly, in the past the current installation could be moved to another location prior to installing the pre-release to have both installed. Is this still the case?
-
Most times the "Refine Seamlines" option leads to holes in the orthomosaic while no holes appear if the option is switched off. Analysing seamlines from both options, it's notorious that a much better result is produced with this new option. Mosaic tends to be a lot better without too much jagged limits on objects. However if holes are produced, it will be a lot less interesting to use.
Hello RHenriques,
If there's any sample set of images that can be used to reproduce the problems with the orthomosaic stitching (e.g. holes in the mosaic), please also send the data to support@agisoft.com, so that we could analyze the source of the issue and fix it.
Hi Alexey
Just send the data.
Best Regards
-
I would like to see a different font style to highlight the active chunk while using the "dark" theme interface option. In the "Classic" and "Light" options, the Bold black font is clearly visible in the white background. If dark background is used, it's hard to see the white bold highlighted chunk. Probably a bold green, as the bars in the ongoing processes in this interface theme, should be better to see.
Best Regards
-
I submitted a crash report but wanted to provide additional information.
After optimizing and checking Estimate tie point covariance then switching to view Point cloud variance in the Model view - then selecting a camera in the Photos pane and Choosing Look Through will crash Metashape every time
Tom
-
Hi Alexey
please test the data I've sent to support@agisoft.com, with highest settings in alignment and dense cloud both in 1.4.4 and 1.5.0 . With version 1.4.4 the dense point cloud gets a lot better when compared to the 1.5.0 version. It's noticeable in areas where there are higher trees for instance. Where the previous version could get a good dense cloud, Metashape completely erases large areas, living big holes in data. I've just noticed now because I've started this same dataset from ground up in Metashape. To test the orthomosaic generation, from my last messages, I was using a project started in version 1.4.4.
Cheers
-
How to install along side existing installation?
If I remember correctly, in the past the current installation could be moved to another location prior to installing the pre-release to have both installed. Is this still the case?
Hello Bill,
You can copy existing PhotoScan Pro installation to another directory and then install Metashape to the old location over the existing installation. Then to use old version run PhotoScan from the folder where to you have copied it previously.
The existing license will be valid both for PhotoScan 1.4 and Metashape 1.5.
-
I submitted a crash report but wanted to provide additional information.
After optimizing and checking Estimate tie point covariance then switching to view Point cloud variance in the Model view - then selecting a camera in the Photos pane and Choosing Look Through will crash Metashape every time
Hello Tom,
The issue with the Look Through will be fixed in the next pre-release update, thanks for reporting.
-
Thanks Alexey
-
Can Metashape classify clouds in the sky and then not select that class to reconstruct a dense point cloud?
-
Dear Alexey,
Did you make any changes to Rolling Shutter compensation?
I am asking because I have data set from fixed wing UAV (flying speed 20 m/s) with Sony a6000 and PPK receiver.
Previously I process this data with Agisoft 1.4.4 and get very good results (without GCP, with Rolling Shutter compensation ON).
With Metashape 1.5.0 I get unreasonable results after optimizing....
Regards,
Paweł
-
Can Metashape classify clouds in the sky and then not select that class to reconstruct a dense point cloud?
Hello engelbeck,
Currently there's no special functionality that allows to detect and classify the noise in the air (for example, related to the clouds) in the dense point cloud.
-
Did you make any changes to Rolling Shutter compensation?
I am asking because I have data set from fixed wing UAV (flying speed 20 m/s) with Sony a6000 and PPK receiver.
Previously I process this data with Agisoft 1.4.4 and get very good results (without GCP, with Rolling Shutter compensation ON).
With Metashape 1.5.0 I get unreasonable results after optimizing....
Hello Paweł,
Can you send both projects saved after or prior to the optimization in 1.4.4 and 1.5.0 versions to support@agisoft.com?
Also please specify, if in both cases you are aligning the cameras without rolling-shutter compensation and are enabling it only prior to the optimization step.
-
Hi Alexey,
I have been trying to conduct some tests with cloud processing. I put my laptop to sleep whilst processing of a dense cloud was taking place. Should one always disconnect in the dialog box when cloud processing is is progress?
When I opened it again some time later, I received an error message saying "timeout" (sorry, no screenshot). Then I closed the reopened Metashape, opened the project. Then i tried to run cloud processing of the dense cloud again. Then I receive the error message that "Project is busy in another processing". Can one reconnect manually using the console? Or even cancel the job? Presumably it is possible. Does one need to install the Python module provided first (NB operating on MacOS Sierra)?
Update: now I seem to be able to start the dense cloud processing again. However was the problem caused by operation, or by something to do with the syncing on Amazon EC2 system (reading the terms and conditions, and making some deductions, I assume that processing is occurring on their Elastic GPUs?).
Despite the teething issues, thanks for letting us test drive the new system - and also allowing us to continue using PhotoScan Pro in case we find we need stability.
Kind regards,
James
-
Hi,
Me again... Has anyone come across the following problem during export of the model:
When I try to set the co-ordinate system from local to any other projection, Metashape becomes unresponsive and I have to force quit.
Regards,
James
-
Hi,
While using the python module in standalone mode I have the following error:
Python 3.6.6 (default, Sep 12 2018, 18:26:19)
[GCC 8.0.1 20180414 (experimental) [trunk revision 259383]] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import Metashape
>>> doc = Metashape.app.document
>>> doc.open("benchmark2.psz")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
AttributeError: 'NoneType' object has no attribute 'open'
>>> print(doc)
None
Basically The doc object is a None type and hence i am unable to do anything with this doc object.
Regards,
khedar
-
Hello Khedar,
Metashape.app.document is related to the active document. Without GUI you do not have active document, so please try the following instead:
import Metashape
doc = Metashape.Document()
doc.open("benchmark2.psz")
-
Me again... Has anyone come across the following problem during export of the model:
When I try to set the co-ordinate system from local to any other projection, Metashape becomes unresponsive and I have to force quit.
Hello James,
Please specify, in which coordinate system is your chunk referenced - which options are selected in the Reference pane settings dialog?
If you have the model referenced in the Local coordinates, then you wouldn't be able to export it in geographic/projected coordinate systems, but at which point the application is freezing - when you click on "more..." line in the coordinate system selection dialog?
-
Hi Alexey,
EPSG::25833. Screenshot of the reference pane settings dialogue box. Following the alignment, i allocate false co-ordinates, purely so that one can measure from the model. The co-ordinate system is then set. Following this, I have run batch processing for the remaining processes. I have successfully created and exported the DEM and orthophotos.
>File>Export>Model ...all ok
... Select destination folder and file name (no silly characters or spaces)
Then the box comes up, and you have to select from the drop down menu the co-ordinate system desired. When I click on the co-ordinate system I desire (untitled 3)... then I see the daisy wheel spinning around and have to force quit Metashape.
And sorry to be the bearer of more issues... can one only have one project processing in the cloud at a time. I tried another pair of large surfaces.... also rather large rocks. Following processing being apparently completed (at some point in the night, as I have to leave it to run 12+ hours), both projects returned the same failure message (NB I have just opened it again to reproduce out the error message):
2018-11-01 23:41:07 LoadProject: path = /Users/au575462/Documents/bmr2262_72/bmr2262_72_metash.psx
2018-11-01 23:41:07 Loading project...
2018-11-01 23:41:07 loaded project in 0.163842 sec
2018-11-01 23:41:07 Finished processing in 0.165855 sec (exit code 1)
2018-11-01 23:41:17 Syncing project with cloud...
2018-11-01 23:41:45 Downloading project files (2.95 GB)...
2018-11-01 23:41:45 Error: Error transferring https://api.agisoft.com:8086/v1.4/projects/ade9da41-3a33-4fcb-938a-dbaea198f428/files/bmr3020_1_hq_extra_up_metash.files%2F0%2F0%2Fdense_cloud%2Fdense_cloud.zip - server replied: (status code 403)
Is it due to the length of time processing? (is there a maximum time limit on jobs?). Also could the problem have arisen by having two projects processing simultaneously in the cloud?
Regards,
James
-
And sorry to be the bearer of more issues... can one only have one project processing in the cloud at a time. I tried another pair of large surfaces.... also rather large rocks. Following processing being apparently completed (at some point in the night, as I have to leave it to run 12+ hours), both projects returned the same failure message (NB I have just opened it again to reproduce out the error message):
2018-11-01 23:41:07 LoadProject: path = /Users/au575462/Documents/bmr2262_72/bmr2262_72_metash.psx
2018-11-01 23:41:07 Loading project...
2018-11-01 23:41:07 loaded project in 0.163842 sec
2018-11-01 23:41:07 Finished processing in 0.165855 sec (exit code 1)
2018-11-01 23:41:17 Syncing project with cloud...
2018-11-01 23:41:45 Downloading project files (2.95 GB)...
2018-11-01 23:41:45 Error: Error transferring https://api.agisoft.com:8086/v1.4/projects/ade9da41-3a33-4fcb-938a-dbaea198f428/files/bmr3020_1_hq_extra_up_metash.files%2F0%2F0%2Fdense_cloud%2Fdense_cloud.zip - server replied: (status code 403)
Is it due to the length of time processing? (is there a maximum time limit on jobs?). Also could the problem have arisen by having two projects processing simultaneously in the cloud?
Hello James,
Thank you for the detailed feedback, it is really useful for us.
It is possible to process several projects in Cloud at the same time, but it's impossible to have several processes of the same project at the same time.
As for "status code 403", it seems that you have exceeded your Cloud account storage space. The storage is limited to 50GB while beta-testing and It is forbidden to upload, download and process projects in Cloud after limit exceeded. We are going to remove download restriction, but upload and process restrictions will still remain.
Please visit your account page on account.agisoft.com, you can check your projects statuses and storage usage here. You can also delete old projects to free up your storage space.
-
Hi Zakhar,
Many thanks for the reply. Glad the feedback is useful. I can see how the need to manage the files was the problem. Thanks for opening that up. Providing one has the space, I can see now that having two or more instances at the same time is not a problem.
I am sure you will be developing it, but having been used to running cluster processing using Terminal to create Turbo VNC instances, having some cribs on how to get information / control things in the same way using either Python, or using unix in Terminal would be useful. But I appreciate the system needs to walk before it can run, so to speak.
I also conduct processing in the field over the mobile phone network, so in order to save on data costs, the ability to control the syncing and the ability to export the model without pulling down the whole model would be great.
Regards,
James
-
have try the split script from previous photoscan but not work.. :)
have try to modify them but lol dont work for now someone can help :)
-
have try the split script from previous photoscan but not work.. :)
have try to modify them but lol dont work for now someone can help :)
Hello Lucky555,
Which error message is displyed in the Console after you run the script?
If you are using the latest version from GitHub repository (https://github.com/agisoft-llc/photoscan-scripts/tree/master/src), then you may need to alter the version in the version check block in the start of the script.
Also you can use "import Metashape as PhotoScan" instead of "import PhotoScan" line.
-
incompactible photoscan version 1.5!= 1.4
-
incompactible photoscan version 1.5!= 1.4
Right in the OP of this thread:
Note
Project files created with 1.5.0 version can't be opened with earlier versions.
-
its not that.. fresh new project... just because in the script photoscan 1.4 is writed after mod to 1.5 the script start but receive
2018-11-02 14:34:01 Script started...
2018-11-02 14:34:01 Traceback (most recent call last):
2018-11-02 14:34:01 File "C:/Users/pluc2/Desktop/script (2).py", line 152, in <lambda>
2018-11-02 14:34:01 proc_split = lambda: self.splitChunks()
2018-11-02 14:34:01 File "C:/Users/pluc2/Desktop/script (2).py", line 219, in splitChunks
2018-11-02 14:34:01 if not chunk.transform.translation.norm():
2018-11-02 14:34:01 AttributeError: 'NoneType' object has no attribute 'norm'
>>>
his error
but look to work now... be back after some other test
-
Hello,
I am trying to Build Tiled Model from depth maps on a 197 20 MP image data set and get a Bad allocation error after 8311 sec of processing….
Attached is log file reduced in Depth maps creation part to respect size attachment limitations...
-
its not that.. fresh new project... just because in the script photoscan 1.4 is writed after mod to 1.5 the script start but receive
2018-11-02 14:34:01 Script started...
2018-11-02 14:34:01 Traceback (most recent call last):
2018-11-02 14:34:01 File "C:/Users/pluc2/Desktop/script (2).py", line 152, in <lambda>
2018-11-02 14:34:01 proc_split = lambda: self.splitChunks()
2018-11-02 14:34:01 File "C:/Users/pluc2/Desktop/script (2).py", line 219, in splitChunks
2018-11-02 14:34:01 if not chunk.transform.translation.norm():
2018-11-02 14:34:01 AttributeError: 'NoneType' object has no attribute 'norm'
>>>
his error
but look to work now... be back after some other test
I think you can modify line #219 as following:
if not chunk.transform.matrix.translation().norm():
-
Hello,
I am trying to Build Tiled Model from depth maps on a 197 20 MP image data set and get a Bad allocation error after 8311 sec of processing….
Attached is log file reduced in Depth maps creation part to respect size attachment limitations...
Hello Paulo,
Does the problem persist if you enable Fine-Level Task Distribution for the tiled model generation in the Network preferences tab and with the smaller tile block size?
-
Hi Alexey,
yes I tried on a reduced region with Fine Level Task Distribution option for Build Tiled Model checked and it worked….
see attached screen capture and reduced log file….
Just tried to redo Tiled Model on full data set from depth Maps but with Tile size of 512 and get Error : No samples in region
See attached log_TiledModel_tilesize.txt
What is this error meant?
-
Using latest radeon drivers works now,thanks.
A bunch of problems came up.
- Playing a youtube video simultaneous with align photos. Alignment stops in the moment of starting the youtube video. Error messge is something like cl device memory...
- Playing a youtube video simultaneous with Build Densecloud. System freezes, black screen after a while, screen comes back, metashape has a blank white screen.
regards
-
Using latest radeon drivers works now,thanks.
A bunch of problems came up.
- Playing a youtube video simultaneous with align photos. Alignment stops in the moment of starting the youtube video. Error messge is something like cl device memory...
- Playing a youtube video simultaneous with Build Densecloud. System freezes, black screen after a while, screen comes back, metashape has a blank white screen.
regards
Yes streaming videos in the browser seems to crash the GPU drivers when a high compute load in being processed.
Mak
-
yes I tried on a reduced region with Fine Level Task Distribution option for Build Tiled Model checked and it worked….
see attached screen capture and reduced log file….
Just tried to redo Tiled Model on full data set from depth Maps but with Tile size of 512 and get Error : No samples in region
See attached log_TiledModel_tilesize.txt
What is this error meant?
Hello Paulo,
Thank you for additional information. Seems that we were able to reproduce the issue on some similar datasets, so hopefully the fix will be included in the next pre-release update. As a workaround I can suggest to try building the complete mesh from the depth maps and then generate the tiled model from the mesh.
-
A bunch of problems came up.
- Playing a youtube video simultaneous with align photos. Alignment stops in the moment of starting the youtube video. Error messge is something like cl device memory...
- Playing a youtube video simultaneous with Build Densecloud. System freezes, black screen after a while, screen comes back, metashape has a blank white screen.
Hello RouvenMiller,
Which Internet browser are you using and whether the hardware GPU acceleration is enabled in the browser settings?
-
A bunch of problems came up.
- Playing a youtube video simultaneous with align photos. Alignment stops in the moment of starting the youtube video. Error messge is something like cl device memory...
- Playing a youtube video simultaneous with Build Densecloud. System freezes, black screen after a while, screen comes back, metashape has a blank white screen.
Hello RouvenMiller,
Which Internet browser are you using and whether the hardware GPU acceleration is enabled in the browser settings?
In my case using Edge (which is GPU accelerated) and only happens randomly when there's a super high GPU load. (GPU is Radeon Fury X). But frankly this isn't that big of an issue. Seems like lowering the GPU clock may help.
-
I use Opera which has hardware acceleration on by default.
-
can one only have one project processing in the cloud at a time.
Hello James,
Thank you for the detailed feedback, it is really useful for us.
It is possible to process several projects in Cloud at the same time, but it's impossible to have several processes of the same project at the same time.
Hi,
Further to the quoted sections above, I believe I may have encountered another bug that needs some attention. I note Zakhar's previous reply and can successfully process more than one project in the cloud. However, when one has more than 1 project processing in the cloud, one cannot download any projects that have finished processing until all have finished. Irrespective of whether you open one or the other project, the progress bar for the last process started is always the one displayed. I knew and could see that the other project had finished processing by looking in my agisoft cloud account. Once all processing had finished, I could download all the files by opening them in Metashape in the expected manner. Perhaps by running a line of code in the python console, one might be able to force this to happen, but I haven't tried that yet.
Allowing things to take their course appears to be important due to the syncing. If syncing or upload interrupted (can also occur due to network failures), or conflicting commands are sent (starting another process, cancelling a process, because one hasn't realised the above issue when processing multiple projects, etc), I have experienced the situation on more than one occasion that you loose all your work on the project. Saving things like masks and markers has therefore become more necessary at present than was the case before.
Regards,
James
-
can one only have one project processing in the cloud at a time.
Hello James,
Thank you for the detailed feedback, it is really useful for us.
It is possible to process several projects in Cloud at the same time, but it's impossible to have several processes of the same project at the same time.
Further to the quoted sections above, I believe I may have encountered another bug that needs some attention. I note Zakhar's previous reply and can successfully process more than one project in the cloud. However, when one has more than 1 project processing in the cloud, one cannot download any projects that have finished processing until all have finished. Irrespective of whether you open one or the other project, the progress bar for the last process started is always the one displayed. I knew and could see that the other project had finished processing by looking in my agisoft cloud account. Once all processing had finished, I could download all the files by opening them in Metashape in the expected manner. Perhaps by running a line of code in the python console, one might be able to force this to happen, but I haven't tried that yet.
Hello James,
Thank you for the information.
We have investigated this issue and found out the cause of the problem. The fix will be included in the next pre-release update.
-
Hi Zakhar,
Super :). Many thanks!!
-
Hi Zakhar
I wonder if there might also be an issue with connecting to the project when switching between wireless and the wired ethernet when these mean there is a switch of IP address?
If I: start a process using the ethernet, then disconnect and close Metashape and then, at a later point in time, try to connect over the wireless network, I find that the connection times out. The project then reverts to whatever is stored locally. One is then unable to connect to the cloud process and sync to the file located there once it has finished. The only option left is then to delete the project and start again, and then check back online on my account page to see if the process has completed. When it is completed, one can connect the ethernet and everything syncs as it should.
Much of the traffic I generate originates from the Danish e-Infrastructure Collaboration 'researcher network'. The wireless connection is over eduroam. Could the problem be something to do with a change in the IP address of the instance? (Amazon specify that regions must be set for instances, but I am not sure if the system would discriminate between eduroam and ethernet traffic from the same institution. I have not systematically tested the issue enough to see if the same fault occurs when changing from working from home and / or another business premises.
And another thing....the issue has already been mentioned here of 'status bad_alloc'. I appreciate that the models with a large number of cameras and with strong geometry pose the greatest demands on the memory in the model creation stage. In the forthcoming payment plans, will be be able to run on the higher capacity G3 instances.
An alternative to this issue would be to able to move the light source in the dense cloud classes mode. Agisoft oc3 format can only be opened in a limited number of applications, but the speed of Metashape's ability to handle very large point clouds is admirable. It is a pity that we cannot harness that potential a little more. With the ability to create bigger and bigger models, this will become a more important issue over time.
Regards,
James
-
I've registered but there are no instructions can anyone tell me how I get started?
-
I wonder if there might also be an issue with connecting to the project when switching between wireless and the wired ethernet when these mean there is a switch of IP address?
If I: start a process using the ethernet, then disconnect and close Metashape and then, at a later point in time, try to connect over the wireless network, I find that the connection times out. The project then reverts to whatever is stored locally. One is then unable to connect to the cloud process and sync to the file located there once it has finished. The only option left is then to delete the project and start again, and then check back online on my account page to see if the process has completed. When it is completed, one can connect the ethernet and everything syncs as it should.
Much of the traffic I generate originates from the Danish e-Infrastructure Collaboration 'researcher network'. The wireless connection is over eduroam. Could the problem be something to do with a change in the IP address of the instance? (Amazon specify that regions must be set for instances, but I am not sure if the system would discriminate between eduroam and ethernet traffic from the same institution. I have not systematically tested the issue enough to see if the same fault occurs when changing from working from home and / or another business premises.
And another thing....the issue has already been mentioned here of 'status bad_alloc'. I appreciate that the models with a large number of cameras and with strong geometry pose the greatest demands on the memory in the model creation stage. In the forthcoming payment plans, will be be able to run on the higher capacity G3 instances.
Hello James,
Cloud API has a static public IP so it doesn't change. Unfortunately I'm not familiar with eduroam, but it seems there is no access to Cloud IP address over the eduroam wireless network.
Changing a network interface can interrupt current connection and it can be a problem for Metashape while cloud processing. To restore the connection after network errors occur, you can close and reopen the project (it's not a user-friendly solution but we are working on it).
As for bad_alloc issue, we are going to increase the amount of RAM on our processing instances.
-
I've registered but there are no instructions can anyone tell me how I get started?
Hello John,
Unfortunately there is no tutorial for cloud processing yet.
There is an Alexey's post about Cloud in this topic:
Technically the cloud processing is similar to the network processing - you enable the Cloud in the Network preferences dialog, then open the project and configure the batch process or run some processing operation. After that (if the operation supports processing on a cloud, for example, the export can be performed locally only) you'll be asked, if the operation should be processed in the cloud. Providing that you accept that, the data will be first synchronized with the cloud and then processed with the user-defined settings. After the operation is completed the project will be synchronized with the cloud again.
You do not need to have Metashape GUI opened while the cloud processing is in progress, but in order to synchronize the project with the cloud you need to open the same project in GUI again.
In terms of the processing time - it is hard to say, as some time will be spent for the data synchronization and it could be quite long in case of very large projects. So mostly this feature would might be helpful for those who cannot process the data locally due to hardware limitations.
Currently the cloud is working in test mode and is free for the Professional license owners. However, after the testing is finished the cloud processing would be likely converted to paid service with the plans configurable in the user's account on the Agisoft Cloud page.
You can set it up using Metashape menu Tools -> Preferences. In the bottom of Network tab you should enable cloud processing and fill Email and Password fields.
-
Quick comment regarding this note in the change log of 1.5.0 7125
" Optimized image matching performance with Generic preselection in case of large overlap."
What exactly happened ? My alignement times are now literally 50%-85% SHORTER than before! Let's say that I'm really pleased and surprised (i7 6700K with a Radeon Fury X).
The longest one went from 40 minutes to 7 minutes now.
Good job
Mak
-
Quick question: I have been playing with the Metashape pre-release and it seems like the
Camera metadata {'Image/Quality': '1.11526'} attribute is no longer stored as a float data format, is this correct?
-
Quick question: I have been playing with the Metashape pre-release and it seems like the Camera metadata {'Image/Quality': '1.11526'} attribute is no longer stored as a float data format, is this correct?
Nope, sorry my bad - I was executing some old code because I screwed up in Git :-p
-
Hi Zakhar,
Another small problem. See attached screenshot of error message. The model has finished processing according to the agisoft cloud account page, but when trying to download the completed product, all I get so far is the attached error message.
How does one go about retrieving the project in such a circumstance? (though I note your previous comment that Agisoft are working on the interface).
Update: after a couple of hours, it has finally worked. Is there a manual way of forcing the action, or is one simply required to wait patiently.?
I note that Agisoft have transferred me over to more powerful server instances. Great! :) Thank you. The difference of additional RAM and another GPU is already noticeable.
Kind regards,
James
-
Hi Zakhar,
Another small problem. See attached screenshot of error message. The model has finished processing according to the agisoft cloud account page, but when trying to download the completed product, all I get so far is the attached error message.
How does one go about retrieving the project in such a circumstance? (though I note your previous comment that Agisoft are working on the interface).
Update: after a couple of hours, it has finally worked. Is there a manual way of forcing the action, or is one simply required to wait patiently.?
James
Hi James,
Thank you for your feedback. We have just fixed this issue.
-
Generating mesh from depth maps always end up with error. Sometimes, if we use low settings, it comes up with "Empty mesh". If higher options are used, for instance Quality: Medium; Face Count: Medium, the following error appears "Can't allocate 922013696 bytes, because max allocation size is 536870912!".
Cheers
-
Hi there,
Is there a change log for build 1.5.0 build 7156 ?
Regards
Mak
-
Generating mesh from depth maps always end up with error. Sometimes, if we use low settings, it comes up with "Empty mesh". If higher options are used, for instance Quality: Medium; Face Count: Medium, the following error appears "Can't allocate 922013696 bytes, because max allocation size is 536870912!".
Hello RHenriques,
Are you using the latest build? Also if possible, please provide the processing log related to the operation and describe the project type you are working on.
-
Hi there,
Is there a change log for build 1.5.0 build 7156 ?
Hello Mak,
Mostly bug fixes, some of them related to cloud processing, as per James report on the previous pages.
-
Hi there,
Is there a change log for build 1.5.0 build 7156 ?
Hello Mak,
Mostly bug fixes, some of them related to cloud processing, as per James report on the previous pages.
Alexey,
Thanks for the info!
Hopefully you guys are able to figure out the issues I reported alongside the .pz projects I provided to you (texturing issue & Metashape crash).
Cheers
Mak
-
Excuse me, who is the designer or user who feels like having menu buttons with no colors a better way of doing things? So that you cant even easily discern is the option grayed out or not selected?
Could you send that designer in here to explain to everyone, why is it a good idea to not use our ability to discern color to speed have better and faster understanding and reference to remember items and buttons?
Windows 10 did this crap too, go to settings and see ALL icons being same blue.. does that improve the user experience for anyone? I do not think so.
Please introduce colors back to the menus and use the gray when the item is not selectable, thank you.
-
Hi Alexey & Zakhar,
Just downloaded the new installer for Metashape... including the bug fixes :) :).
Tried opening up to download a completed project in the cloud... and encountered the error message attached. Is it correct that Metashape needs to use OpenSSL for cloud processing?
I see that Apple have stopped shipping OpenSSL around 2015. Therefore, I have tried installing it manually. The correct results are returned in Terminal, but the same error message is returned in Metashape. I tried deactivating Metashape, deleting the old copy and emptying the trash, followed by restarting, downloading Metashape and reactivating it. I was not successful. So before I try again installing Open SSL, I thought I should check back with you guys.
Like an idiot, I didn't take a copy of the previous installer, in order to carry on working. But I guessed any problems arising will be solved, given how admirably quick your response times are.
Regards,
James
Update: According to terminal, I have managed to install and can run OpenSSL 1.0.2. Could I hazard a guess that Metashape is not searching the "usual" places when it checks for version?
-
Hi,
I also have an issue with cloud processing on the new version : when I tried to upload a new project, terminal returned a "can't write" message.
-
Tried opening up to download a completed project in the cloud... and encountered the error message attached. Is it correct that Metashape needs to use OpenSSL for cloud processing?
Hello James,
Than you for reporting.
We will publish the pre-release update in a few days that would properly work with default Mac OS X installation without a need of OpenSSL. Sorry for the inconvenience.
-
Hi,
I also have an issue with cloud processing on the new version : when I tried to upload a new project, terminal returned a "can't write" message.
Hello jnb,
Thank you for reporting this issue. But unfortunately we couldn't reproduce it. The cause of this problem can be an unreliable Internet connection. Nevertheless we have increased some timeouts on server-side, so it could help.
Could you please try again?
-
Hi Zakhar,
Working now ! Nothing has changed on my side, so maybe increasing the timeouts did the trick.
-
Dear Agisoft team,
Can i ask where can i found Use strict volumetric masking in the new version 1.5.0? it's not exist in the workflow--build mesh
Best
Mohammed
-
Hello Mohammed,
New meshing method will include strict volumetric masks in the final release as well as vertex colors option.
-
Dear Alexey,
Is it possible to add a tutorial in your blog about how to use build Orthomosaic (cylinder projection) for semi cylinder or quarter cylinder objects/ceiling it's still bit difficult for me to understand, i can provide you a data set if you need.
Thanks
Mohammed
-
Hi,
What Optimize coverage does in the tool menu?
-
Is it possible to add a tutorial in your blog about how to use build Orthomosaic (cylinder projection) for semi cylinder or quarter cylinder objects/ceiling it's still bit difficult for me to understand, i can provide you a data set if you need.
Hello Mohammed,
If you can send the sample image set that can be used to create the text tutorial, please provide the link to download to support@agisoft.com.
-
What Optimize coverage does in the tool menu?
Optimize coverage is a new experimental feature that allows to disable the images that result in excessive overlap. Currently it requires to have the alignment and at least rough mesh model. The input value defines the target number of cameras that should be left enabled from the complete set.
-
Alexey,
regarding Optimize Coverage command, i tried on a data set. I first created a mesh from sparse point cloud and then ran Optimize Coverage.
It ran and added 182 cameras out of 364 but then gave error No texture… see attached log file...and screen capture
Why should the texture be present for command to run?
-
Hi Alexey,
Thanks for the great work, I just tried the auto-classification function, and I have two questions to report:
(1) After the autoclassification, I can't reset the classification. However if I continue used "Classify Ground Points" the point clouds will be classified based on the "ground point" classes.
(2) Some ground in "Classify points" were clasified into building, since it is based on machine learning, would it be possible to adjust any parameters or it is all default set? Also the time is much longer than previous tool, I got 52.7595 min vs 6.750816667 mins.
Thanks,
Kind regards
-
Tried opening up to download a completed project in the cloud... and encountered the error message attached. Is it correct that Metashape needs to use OpenSSL for cloud processing?
Hello James,
Than you for reporting.
We will publish the pre-release update in a few days that would properly work with default Mac OS X installation without a need of OpenSSL. Sorry for the inconvenience.
Hi Alexey,
Whilst I do not wish to sound impatient... I was wondering if you could give an update on the progress of the forthcoming update to fix this particular issue?
Kind regards,
James
-
Alexey,
regarding Optimize Coverage command, i tried on a data set. I first created a mesh from sparse point cloud and then ran Optimize Coverage.
It ran and added 182 cameras out of 364 but then gave error No texture… see attached log file...and screen capture
Why should the texture be present for command to run?
Same thing happened to me. I ran it with automatic for the Max. cameras option.
I textured the model and ran it again, and it worked.
-
Metashape tends to give, consistently, poor results both in Dense Point Cloud and Mesh. Things get worst when perfect conditions in photographies are not possible. Here are the results from a table terrain model with photos taken in low light conditions with an iPhone 7 Plus. Comparison between Photoscan (Latest version) and Metashape (latest beta). Diferences are impressive. The parameters are exactly the same in both packages.
Cheers
-
The issue with the "no texture" message should be fixed for Optimize Coverage command in the latest pre-release update (build 7205). No texture is required, only mesh model.
The problem with the OpenSSL message on Mac OS X related to cloud computing should be fixed in this version as well.
Also the possible crashing of the new meshing method reported by Mak11 should be fixed in the latest update.
-
I just tried the auto-classification function, and I have two questions to report:
(1) After the autoclassification, I can't reset the classification. However if I continue used "Classify Ground Points" the point clouds will be classified based on the "ground point" classes.
(2) Some ground in "Classify points" were clasified into building, since it is based on machine learning, would it be possible to adjust any parameters or it is all default set? Also the time is much longer than previous tool, I got 52.7595 min vs 6.750816667 mins.
Hello Phogi,
Can you send the dense cloud in one of the formats that are supported for import in Metashape 1.5.0 to support@agisoft.com?
Currently there are no parameters for the new Classify Points feature. We can, however, improve the classificator by extending the source set of the dense clouds. Ideally we would accept already manually classified dense clouds for the common projects and would include them to the learning procedure. But also some examples of unacceptable results could be provided for our consideration.
-
Metashape tends to give, consistently, poor results both in Dense Point Cloud and Mesh. Things get worst when perfect conditions in photographies are not possible. Here are the results from a table terrain model with photos taken in low light conditions with an iPhone 7 Plus. Comparison between Photoscan (Latest version) and Metashape (latest beta). Diferences are impressive. The parameters are exactly the same in both packages.
Hello RHenriques,
Can you provide the reports from these two projects and, if possible, the processing logs?
I also see that there is different numbers of tie points in the projects, which means that the dense cloud reconstruction is based on different alignment.
-
Metashape tends to give, consistently, poor results both in Dense Point Cloud and Mesh. Things get worst when perfect conditions in photographies are not possible. Here are the results from a table terrain model with photos taken in low light conditions with an iPhone 7 Plus. Comparison between Photoscan (Latest version) and Metashape (latest beta). Diferences are impressive. The parameters are exactly the same in both packages.
Cheers
I've encountered exactly the same issue with one of my data sets but can't remember which one right now. If you look at the depth maps in MetaShape you notice right away that there's a problem. There seems to be a bug messing up the depth maps generation process.
Mak
-
If there's any sample image set that can be used to reproduce the depth maps / dense cloud difference between the versions 1.4.4 and 1.5.0, please send them to support@agisoft.com.
-
If there's any sample image set that can be used to reproduce the depth maps / dense cloud difference between the versions 1.4.4 and 1.5.0, please send them to support@agisoft.com.
Now I remember which one it was. Wasn't mine but it's a super crappy set of photos with major noise because of high ISO value. Doesn't seem to be a problem in Photoscan but Metashep simply couldn't correctly generate the depth maps. So maybe look into datasets with high ISO value to try to replicate the bug? I won't be able to send you the dataset until Thursday.
Cheers
Mak
-
Dear support,
thanks for the update to pre release….
I would like to ask what is the meaning and use of setting photo invariant parameters in Camera Calibration?
For example in new Phantom 4 RTK camera, the exif data specifies calibrated focal length and principal point (optical center):
---- XMP-drone-dji ----
Absolute Altitude : +683.68
Relative Altitude : +40.00
Gps Latitude : 40.14186282
Gps Longtitude : -3.82089900
Gimbal Roll Degree : +0.00
Gimbal Yaw Degree : -83.90
Gimbal Pitch Degree : -90.00
Flight Roll Degree : -1.40
Flight Yaw Degree : -79.90
Flight Pitch Degree : -18.80
Flight X Speed : +0.00
Flight Y Speed : -0.30
Flight Z Speed : +0.00
Cam Reverse : 0
Gimbal Reverse : 0
Self Data : Undefined
Calibrated Focal Length : 3666.666504
Calibrated Optical Center X : 2736.000000
Calibrated Optical Center Y : 1824.000000
Rtk Flag : 50
Rtk Std Lon : 0.03910
Rtk Std Lat : 0.03385
Rtk Std Hgt : 0.06432
Dewarp Flag : 1
Should we then define f, cx and cy as photo invariant in Camera Calibration as well as set f = 3666.666504, cx = cy =0 ?
Any info on this question? anyone ?
-
Hi Alexey / Zakhar,
Thanks for all the fixes in the new release! All seems to be working!
Kind regards,
James
-
Metashape tends to give, consistently, poor results both in Dense Point Cloud and Mesh. Things get worst when perfect conditions in photographies are not possible. Here are the results from a table terrain model with photos taken in low light conditions with an iPhone 7 Plus. Comparison between Photoscan (Latest version) and Metashape (latest beta). Diferences are impressive. The parameters are exactly the same in both packages.
Hello RHenriques,
Thank you for providing the sample image set.
Actually, if processed using Mild filtering option the difference is not so considerable, while aggressive filters out the unreliable areas that are too noisy and lack actual information. If you look on the source images - the surface details are almost lost due to image compression and high noise. If you look on the areas that are missing in 1.5.0 in the dense cloud from the older version, you would see that the point cloud there is very noisy, meaning that it is inaccurate and also unreliable.
Screenshot below demonstrates the results of processing in the version 1.5 (left side) and 1.4 (right side). High quality and Mild filtering have been used.
-
Thank you Alexey! Usually I use the aggressive filter precisely to avoid High Noisy Clouds. I'll have to change a little bit my habits to adapt to Metashape. It's good to know that this issue is rather easy to solve. I'll try later the same settings to see if I can get the same results.
Best Regards
-
Hello,
I don't know if this is related to the beta version, but I have an error trying to import chunk in an existing project (missing point_cloud/points2.zip).
But I have no problem opening or working on this specific chunk.
Regards
-
Hi Alexey / Zakhar,
I have tried to set in motion a batch job consisting of a dense cloud on ultra high quality and a DEM. It will take perhaps 20 hours to run. Fine. However, after having disconnected from the service to leave it to get on with it, when I try check back on the progress of the result, the following error is returned. I left the same process to run overnight and the same error message was returned this morning. The account status page said the job was still in progress. Perhaps I should leave it until the job has finished / a long period of time has elapsed (at least a day and a half). Or perhaps it is another bug?
Update: the processes successfully completed and the model is now downloading... so whatever the error message refers to, the system is robust enough to survive. However, for a model at dense cloud + DEM of 46 GB, I am going to need some more memory to go to the next level. Files tend to become 3 times as large after a mesh is created. The model is of one of Denmark's most significant rock art sites that was recently discovered on Bornholm.
Kind regards,
James
-
Hi again,
I may have found a new bug : when generating a MNE while orthomaisaics alreeady exist, orthomosaics won't display anymore.
Regards
-
The new Build Tiled Model produces good meshes. However, if we align photos and directly produce a tiled model, it's impossible to build a DEM or an Orthomosaic. The tiled model seems to be, at least, as good (if not better) as the meshes that are obtained with the classical steps. Considering that it's possible to obtain a DEM from the mesh using the classical approach, is it not also possible to obtain it from a mesh obtained with the "tiled model" reconstruction method? Is this a real limitation that makes this feature impossible or an ongoing development? Not being able to extract a DEM from a tiled model (and an Ortho) is a limitation for all those that are using Metashape to extract mapping data from aerial surveys. Of course, there's always the classical approach that works really well.
Cheers
-
Hello RHenriques,
We are planning to add the DEM generation from the tiled model in the version 1.5.0 release. If dense cloud stage needs to be skipped.
-
I may have found a new bug : when generating a MNE while orthomaisaics alreeady exist, orthomosaics won't display anymore.
Hello jnb,
I'm not observing this issue with the current pre-release build. Is it possible that you are gnerating DEM in a different coordinate system than orthomosaic has been generated in? Does "reset view" button works for orthomosaic in the Ortho view to bring it back?
-
Indeed, reset view button was the solution, sorry for the false alarm and thank you very much for your answer.
-
I have tried to set in motion a batch job consisting of a dense cloud on ultra high quality and a DEM. It will take perhaps 20 hours to run. Fine. However, after having disconnected from the service to leave it to get on with it, when I try check back on the progress of the result, the following error is returned. I left the same process to run overnight and the same error message was returned this morning. The account status page said the job was still in progress. Perhaps I should leave it until the job has finished / a long period of time has elapsed (at least a day and a half). Or perhaps it is another bug?
Update: the processes successfully completed and the model is now downloading... so whatever the error message refers to, the system is robust enough to survive. However, for a model at dense cloud + DEM of 46 GB, I am going to need some more memory to go to the next level. Files tend to become 3 times as large after a mesh is created. The model is of one of Denmark's most significant rock art sites that was recently discovered on Bornholm.
Hello James,
Seems to be a bug, although it doesn't affect the processing. We'll fix it in the next pre-release update.
-
I had some time to test a few datasets, mainly captures by UAV's and Metashape really has a problem in dense clouds. Usually they are a lot poor than the ones produced in version 1.44. Even with the filtering disabled, lots of areas disappear from the cloud leaving huge holes. This problem is common mainly in areas with vegetation or more complex terrain. The last version usually solves these areas quite well. Metashape tends to eliminate them, apparently with excessive filtering. It's not a case of a single dataset or special cases. Its consistent whenever there's terrain and vegetation. The tied model improves things a bit because these areas get solved. However there are a few things that still need a dense cloud to be generated (for instance point classification). Meshes from dense clouds are also more detailed, at the present stage, than meshes produced using tiled model. Probably things will improve if some bugs in high quality tiled model generation are solved and a DEM can be generated from it.
Cheers
PS: Tried with filtering disabled and the cloud, as expected, does not have any holes. It seems that the thresholds for the filtering limits have been tighten, maybe too much. I would say that the depth Mild filtering in now equivalent to the Moderate option in PS 1.4.4 and "Disabled" from MS is almost equivalent and Mild in PS 1.44. It's more or less what James_AU is also saying in the next post. In UAV surveys, we might need the data that is "disappearing" now in Metashape if Mild Filter is used but using disabled is a little bit excessive also. Maybe we could get a few more options in the middle to get the best of both versions.
-
I had some time to test a few datasets, mainly captures by UAV's and Metashape really has a problem in dense clouds. Usually they are a lot poor than the ones produced in version 1.44. Even with the filtering disabled, lots of areas disappear from the cloud leaving huge holes. This problem is common mainly in areas with vegetation or more complex terrain. The last version usually solves these areas quite well. Metashape tends to eliminate them, apparently with excessive filtering. It's not a case of a single dataset or special cases. Its consistent whenever there's terrain and vegetation. The tied model improves things a bit because these areas get solved. However there are a few things that still need a dense cloud to be generated (for instance point classification). Meshes from dense clouds are also more detailed, at the present stage, than meshes produced using tiled model. Probably things will improve if some bugs in high quality tiled model generation are solved and a DEM can be generated from it.
Cheers
Hi RHenriques & Alexey
Interesting. I have been rereading the post on this issue as well as the replies. I have also noticed more holes in the final result. However, for me these emerge at a scale that is less problematic for my purposes. However, I think Alexey's observation is very pertinent: there was more noise in those areas. This lowers the statistical confidence of the matches, resulting in rejection during the alignment stage. The lack of sufficient key and tie points is then carried through into the dense cloud
Agisoft have stated in another post (http://www.agisoft.com/forum/index.php?topic=89.0), the workflow is similar to SIFT, although there have been changes since 2011. However, SIFT only considers matches to be true if they pass a very high level of confidence. I suspect that, as the software has developed, the threshold has been successively tightened (and why not) and further improvements have been implemented to improve precision. Personally, working with depth differences in rock art regularly as little as 1mm, I value that focus on precision. If we are not as sure as we thought we before: ok. Then we need to reevaluate the quality of our pictures and/or look critically on why this might be the case, particularly surfaces with a very uniform texture.
Regards
James
-
Hello RHenriques,
We are planning to add the DEM generation from the tiled model in the version 1.5.0 release. If dense cloud stage needs to be skipped.
Excellent!
-
What Optimize coverage does in the tool menu?
Optimize coverage is a new experimental feature that allows to disable the images that result in excessive overlap. Currently it requires to have the alignment and at least rough mesh model. The input value defines the target number of cameras that should be left enabled from the complete set.
This will be very useful for Kite Aerial Photography (KAP)! :)
-
Hello,
i don't know, if it is intended to work in this way, but when i draw a polygon in my DEM an define it as outer boundary, i calculate a volume A.
(https://i.ibb.co/ZgtwyKj/2.jpg) (https://ibb.co/rpTLnMj)
Now i draw another polygon inside my outer boundary and define it as inner boundary, to calculate a volume B. I'd think, that Metashape ignores everything inside the inner boundary and triangulates everythink between the outer boundary and the inner boundary plus the "flat" inner boundary (big triangles from inner boundary to the other side of the same boundary while ignoring everything inside. But the result seems to be exactly the same:
(https://i.ibb.co/PFHtbnt/1.jpg) (https://ibb.co/vVnDrGD)
The way i'd expect the calculation to work is something like this:
(https://i.ibb.co/sFwqNxJ/3.jpg) (https://ibb.co/GV2R8Y3)
Best regards
Thomas
-
Hello Thomas,
Inner and Outer boundary type is meant to be applied only for the export operations. These types are not considered for the area and volume calculation. If the measurements are applied to the group of selected shapes, in the Measure dialog the sum across all the shapes will be shown for perimeter, area and volume.
-
Hello Alexey,
in this case, please see my post as a feature request!
Maybe you want to remove parking cars, trees and bushes or some containers, before calculating the volume. If i do this with a mesh, i have to create one first, then delete all the unwanted objects and in the last step, i have to close the holes again, before being able to calculate the volume.
Using vector shapes, this would be much easier and faster!
Best regards
Thomas
-
Hello Alexey,
in this case, please see my post as a feature request!
Maybe you want to remove parking cars, trees and bushes or some containers, before calculating the volume. If i do this with a mesh, i have to create one first, then delete all the unwanted objects and in the last step, i have to close the holes again, before being able to calculate the volume.
Using vector shapes, this would be much easier and faster!
Best regards
Thomas
Have you considered classifying the dense cloud prior to generation of a DEM/Mesh? You can either classify the densecloud manually by means of extracting desired points using polygon drawing and subsequently reclassifying the said selection to a desired non-ground class. Alternatively you can play around with the Automatic Classification parameters found under Tools and let the software come up with a reclassified pointcloud with respect to ground and non-ground points.
Once you're happy with the resultant classification you can follow-up with generating of the DEM or Mesh. In the settings screen for both you can define Dense Cloud as a source surface, besides allowing you to define the pointcloud classes on which you would like to products to be generated. When only activating the Ground class you should end up with a DEM/Mesh that closely approximates the actual ground surface of the terrain, provided your classification was done correctly.
-
Yes I have. Areas that I definitely do not need, I mask even before I align the cameras (areas with people, parking cars, ...), so that they do not even appear in the 3D model. But there are also areas that I have to "hide" (with a polygon) temporarily (trees for a mass calculation for example) and then "unhide" (deleting the polygon) in later to export a DEM (with the trees), which serves as the basis for a more detailed overflight (mission planning). There, i want to have all obstacles to be exported. Of course, your suggestion would work, but it would cost a lot more time than drawing just a polygon around the spot.
-
Hello,
I would like to ask here for a feature request.
We have no ability to export the depthmaps -> normal maps via Batch process. I would really welcome that feature.
regards
-
Has anyone experimented with using the standalone python version in a docker / containerized environment?
-
What is exactly "Added an option to fix a subset of calibration parameters.".
-
Updated to version 1.5.0 build 7336
Whats the changelog for this version?
-
Hi FOSI. Change log is in the downloads page.
After testing this beta with two projects, it's clear that the quality of the dense cloud is again, at least, on pair with PS 1.4.4. In two projects that I've compared the overall quality is quite good for the chosen filtering parameters.
The only drawback that I've noticed is the Optimize Coverage feature. I've tested in a 750 cameras project, from a UAV survey, and it seems to "over optimize" a little bit. The model was based on 375 photos and ended up with lots of holes, clearly related with lack of coverage.
-
There's a bug in Build 7336 when using the new Mesh generation method and the calculate Vertex Colors toggle (which was introduced in this build).
During the "Extracting Surface" and vertex colors generation phase the following error messages appear in the console:
"Error: Image and depth map size mismatch! 4000x4000 and 4000x6000". <---which is obviously false btw.
It doesn't prevent the process from finishing but the vertex colors are not generated (the model is black).
Regards
Mak
-
Hello Mak,
Thank you for reporting.
I was able to reproduce the problem with the vertex colors similar to yours by working with the dataset where the image width is shorter than image height, hope to find the source of the issue as soon as possible and update the pre-release. I assume that working with the datasets with landscape image proportions wouldn't rise this error.
-
What is exactly "Added an option to fix a subset of calibration parameters.".
I would like to reiterate that more clarity, explanation will be appreciated on the use and difference in Fixed parameters and photo invariant parameters in latest release…
I can understand that you may have a calibrated lens where you want to fix f, cx and cy and leave other parameters free to be adjusted by software...
But what would using photo invariant parameters do?
-
Hello Paul,
Defined photo-invariant parameters means that these parameters will be individual for each image in the calibration group, while defined fixed parameters means that the selected parameters will be fixed to initial values for all the images in the calibration group. Previously it was possible only to fix all the Initial parameters at once and now it is possible to fix only certain parameters.
If you do not fix the parameters and do not use photo-invariant options, then all the parameters will be adjusted during the alignment operation and the adjusted values will be the same for all the images in the calibration group.
-
Updated to version 1.5.0 build 7336
Whats the changelog for this version?
Hello FOSI,
If this document still shows the older version, then use CTRL+F5 in your browser to force the update: http://www.agisoft.com/pdf/metashape_changelog.pdf
-
Hi Alexey,
thank you for the clarification.....
So For Photo-invariant parameters option would it not be better to use instead Photo-variant parameters as in this case a different parameter will be estimated for each photo?
Using this option, where can we find the adjusted Photo Invariant (Variant) parameter for each image?
-
Hello Paul,
You can see the values of "photo-invariant parameters" for each photo in the bottom frame with the images list in the Camera Calibration dialog.
-
Hi !
I'm trying to generate a texture wih the new version, and I get this error : "Unexpected channel count".
Regards
-
Alexey, I reinforce my last message about dense clouds. They are now better than in version PS1.4.4 by using max parameters and the Mild filter. It's not noticeable by direct comparision but it's very noticeable when we generate a DEM from it. The DEM is just perfect! Please do not "fine tune" more these parameters :)
Optimize Coverage does need some tuning, perhaps make it a little more tolerant to allow a few more overlap. I guess that this can be related to the fact that some pictures can be slightly oblique and will not ensure enough overlap if some neighbor ones are eliminated from calculation.
-
I'm trying to generate a texture wih the new version, and I get this error : "Unexpected channel count".
Hello jnb,
Can you please provide the log from the Console pane related to the failed texturing operation and also the information regarding the source image bands?
Also please specify, if the model has been generated from the same set of images or the pictures have been replaced by modified image set?
-
Hi Alexey,
Here is the log, with two attemps, one for the texture, one for the tiled model, both failed with the same message.
The images are 8bits tif (only RGB) converted from 14bits DNG, and have not been replaced.
Thank you for looking into it !
-
Hello jnb,
Is it also possible to attach any image from this set or send an image example from the project to support@agisoft.com?
-
Image sent on your mail !
-
Hi, i tried to Align a chunk of 1200 fotos from DJI inspire 2 with X5S with metashape. At the end of the process of alignment it gave me an error that it couldn't align all the photos. So when i checked the sparse cloud, only 1/3 of the photos where aligned.
Today i tried the same batch of photos with photoscan 1.4.4 with the same settings and it aligned all the photos?
Is there something different i need to change in metashape to obtain the same results relative to the aligment?
-
Hello FOSI,
Can you please send the complete processing logs from both runs?
-
Hi Alexey.
I just made a run of 600 photos of the same area as before with logging on.
Metashape v7336
im running the same with photoscan 1.4.4 and i'll send you the log later
-
Same batch of 600 fotos on another PC with metashape v7336, same settings
Some fotos not possible to align (same error message as before)
But it aligned a different area of photos from the other PC with metashape v7336
-
Hello FOSI,
Can you please re-save the latest project with only a small subset of images aligned in PSZ format (Metashape Archive file format in Save As dialog) and send it to support@agisoft.com, we'll check what might be the issue. In case you have the same project properly aligned with PhotoScan Pro 1.4, please also send the second project.
-
Just sent it. Thanks
-
I also think there are some changes in the 7336 match.
The same photo 7336 and 1.44 compared, it is obvious that 7336 does not match.
-
I also think there are some changes in the 7336 match.
The same photo 7336 and 7224 compared, it is obvious that 7336 does not match.
Yes i noticed that also.
-
Hi ruyi7952. I did not notice that but it's possible, as you show. Can you confirm that the advanced parameters of the alignment are the same? The alignment is indeed very diferent in your case. Very strange.
I've tested with a more than 700 images set and the alignment was the same as the previous version. Of course, conditions can be different.
Can you share the dataset?
-
Dear Agisoft,
So i made a comparison test between V1.1.6 (i always love this version in comparison) and metashape V1.5 last update to an 2000 years old artifice i used 160 image (the result in the snapshot 1) anyhow V1.1.6 Align all the photos in medium setting while in V1.5 i used the same setting and it only align 63 image (both projects are masked), So what's wrong? this the first bug.
The second bug the same project in V1.5 i changed the setting to high and it align all photos but i notice that the jar rim not align ( i saw it in dens point cloud) very well but in V1.1.6 it was aligned very well (see snapshot 2) Note the Build dens is different in both projects but the problems from Align itself the strange thing in V1.5 when i build the Mesh based on Depth map this bad Align not exist????
Hope it was a bug and fix it.
Best
Mohammed
-
Dear Agisoft,
So i made a comparison test between V1.1.6 (i always love this version in comparison) and metashape V1.5 last update to an 2000 years old artifice i used 160 image (the result in the snapshot 1) anyhow V1.1.6 Align all the photos in medium setting while in V1.5 i used the same setting and it only align 63 image (both projects are masked), So what's wrong? this the first bug.
The second bug the same project in V1.5 i changed the setting to high and it align all photos but i notice that the jar rim not align ( i saw it in dens point cloud) very well but in V1.1.6 it was aligned very well (see snapshot 2) Note the Build dens is different in both projects but the problems from Align itself the strange thing in V1.5 when i build the Mesh based on Depth map this bad Align not exist????
Hope it was a bug and fix it.
Best
Mohammed
Hi Agisoft,
Any feedback about this comparison?
Best
Mohammed
-
I also think there are some changes in the 7336 match.
The same photo 7336 and 1.44 compared, it is obvious that 7336 does not match.
Hello ruyi7952,
Can you share the projects with the alignment results and source images to support@agisoft.com?
-
So i made a comparison test between V1.1.6 (i always love this version in comparison) and metashape V1.5 last update to an 2000 years old artifice i used 160 image (the result in the snapshot 1) anyhow V1.1.6 Align all the photos in medium setting while in V1.5 i used the same setting and it only align 63 image (both projects are masked), So what's wrong? this the first bug.
The second bug the same project in V1.5 i changed the setting to high and it align all photos but i notice that the jar rim not align ( i saw it in dens point cloud) very well but in V1.1.6 it was aligned very well (see snapshot 2) Note the Build dens is different in both projects but the problems from Align itself the strange thing in V1.5 when i build the Mesh based on Depth map this bad Align not exist????
Hello Mohammed,
Is it possible for you to share the source image set for our internal tests?
Also have you tried to align the images without adaptive camera model fitting and not applying the masks to the tie points? In the version 1.1.6 these parameters were not available.
-
So i made a comparison test between V1.1.6 (i always love this version in comparison) and metashape V1.5 last update to an 2000 years old artifice i used 160 image (the result in the snapshot 1) anyhow V1.1.6 Align all the photos in medium setting while in V1.5 i used the same setting and it only align 63 image (both projects are masked), So what's wrong? this the first bug.
The second bug the same project in V1.5 i changed the setting to high and it align all photos but i notice that the jar rim not align ( i saw it in dens point cloud) very well but in V1.1.6 it was aligned very well (see snapshot 2) Note the Build dens is different in both projects but the problems from Align itself the strange thing in V1.5 when i build the Mesh based on Depth map this bad Align not exist????
Hello Mohammed,
Is it possible for you to share the source image set for our internal tests?
Also have you tried to align the images without adaptive camera model fitting and not applying the masks to the tie points? In the version 1.1.6 these parameters were not available.
Thanks Alexey for your answer!
Yes i tried to disable the adaptive camera model fitting and changed the tie point mask to key point (in medium setting) and the same result.
How can i share the photos to yo, do you need the Images or project file?
Best
Mohammed
-
Hello Mohammed,
We would need the source images to reproduce the similar difference in results in versions 1.1 and 1.5. If possible, please send the link via some filesharing service to support@agisoft.com.
-
Hello Mohammed,
We would need the source images to reproduce the similar difference in results in versions 1.1 and 1.5. If possible, please send the link via some filesharing service to support@agisoft.com.
Okay i will do it tonight.
Thank you.
-
Hello Mohammed,
We would need the source images to reproduce the similar difference in results in versions 1.1 and 1.5. If possible, please send the link via some filesharing service to support@agisoft.com.
Hi Alexey i send the images link to your mail. the images in Tif format with mask in Alpha channel.
please feed me back after test it.
Best,
Mohammed
-
So i made a comparison test between V1.1.6 (i always love this version in comparison) and metashape V1.5 last update to an 2000 years old artifice i used 160 image (the result in the snapshot 1) anyhow V1.1.6 Align all the photos in medium setting while in V1.5 i used the same setting and it only align 63 image (both projects are masked), So what's wrong? this the first bug.
The second bug the same project in V1.5 i changed the setting to high and it align all photos but i notice that the jar rim not align ( i saw it in dens point cloud) very well but in V1.1.6 it was aligned very well (see snapshot 2) Note the Build dens is different in both projects but the problems from Align itself the strange thing in V1.5 when i build the Mesh based on Depth map this bad Align not exist????
Hello Mohammed,
Is it possible for you to share the source image set for our internal tests?
Also have you tried to align the images without adaptive camera model fitting and not applying the masks to the tie points? In the version 1.1.6 these parameters were not available.
Dear Alexey,
Just another note In V1.5 (metashap) generate mesh based on Depth map couldn't build the mesh for the bottom part with alot of noise in the surface , could you also check this issue? (see snapshot based on depth map).
Best,
Mohammed
-
Mohammed
Are you sure that the bounding box isn't clipping through the model (sounds stupid but I have made the mistake several times and had to reprosse the models/dense clouds because I incorrectly resized the bounding box).
If not, can you check the depth maps to see if they were correctly generated for this part of the model? MetaShape seems to be picky with photos with high noise level compared to 1.4. Lowering the quality (from ultra to high or even medium can improve things)
Mak
-
Mohammed
Are you sure that the bounding isn't clipping through the model (sounds stupid but I have made the mistake several times and had to reprosse the models/dense clouds because I incorrectly resized the bounding box).
If not, can you check the depth maps to see if they were correctly generated for this part of the model? MetaShape seems to be picky with photos with high noise level compared to 1.4. Lowering the quality (from ultra to high or even medium can improve things)
Mak
Dear mak,
Thank you! You are right i forget to correct the boundary box, I process it again in High (after fix the boundary box) and will see the result.
Dear Alexey iam sorry to confuse you, in Depth map issue, i'll process it again and will post it the result again.
Best
Mohammed
-
Mohammed
Are you sure that the bounding box isn't clipping through the model (sounds stupid but I have made the mistake several times and had to reprosse the models/dense clouds because I incorrectly resized the bounding box).
If not, can you check the depth maps to see if they were correctly generated for this part of the model? MetaShape seems to be picky with photos with high noise level compared to 1.4. Lowering the quality (from ultra to high or even medium can improve things)
Mak
I attached the result with Depth map mesh in High setting the bottom part generated but with extra polygons (might because of low images for this part) but there's many noise in the surface, anyhow the result not so bad for me.
Still the comparison i did between V1.1.6 and V1.5, i send the images and wait for the answer.
Best
Mohammed
-
Sometimes when running Build Mesh with Depth maps as the Source data, and Calculate vertex colors, the colors are black.
(http://www.digital-mapping.net/forums/Photoscan/2018/no_vertex_color.jpg)
The log file is attached.
-
Bisenberger,
I reported this bug last week and Alexey was able to reproduce it (see previous messages in this thread). It happens when pictures are shot vertically. Should be fixed in the next update I guess.
Mak
-
Hello bisenberger,
There's a bug in the latest pre-release build related to the vertex color calculation that is reproducible in certain cases of source images proportions.
The fix will be included in the next pre-release update (this weekend, I assume), meanwhile it is suggested to disable vertex colors calculation if you encounter such error for the certain project.
-
Oops, I missed that one Mak11. :-[
Thanks Alexey. Really liking the different meshing options. 8)
-
Alexey, Dmitry,
Is there a Change Log for build 7441?
Cheers
EDIT: Just got updated. Thanks
Mak
-
In the latest pre-release update the issue with the vertex colors for the new meshing option should be fixed.
-
Error_Refine_Seamlines in Build 7441
2018-12-25 09:56:10 used RAM: 0.466331 / 95.9456 GB
2018-12-25 09:56:10 free RAM: 92.3456 / 95.9456 GB
2018-12-25 09:56:10 partition refined in 7.891 sec
2018-12-25 09:56:10 loaded camera partition in 0 sec
2018-12-25 09:56:10 Refining seamlines...
2018-12-25 09:56:10 Refining partition...
2018-12-25 09:56:10 topology loaded: Topology:
2018-12-25 09:56:10 n_vertices: 35837
2018-12-25 09:56:10 n_edges: 56649
2018-12-25 09:56:10 RAM size: 0.0132422 GB
2018-12-25 09:56:10 refining 20 partitions using 12 threads
2018-12-25 09:56:14 partition for image 91 refined in 3.625 sec
2018-12-25 09:56:14 partition for image 89 refined in 3.812 sec
2018-12-25 09:56:15 partition for image 88 refined in 4.281 sec
2018-12-25 09:56:15 partition for image 90 refined in 4.36 sec
2018-12-25 09:56:15 partition for image 83 refined in 4.593 sec
2018-12-25 09:56:15 partition for image 86 refined in 4.563 sec
2018-12-25 09:56:15 partition for image 81 refined in 4.672 sec
2018-12-25 09:56:15 partition for image 85 refined in 4.75 sec
2018-12-25 09:56:15 partition for image 82 refined in 4.766 sec
2018-12-25 09:56:15 partition for image 84 refined in 4.797 sec
2018-12-25 09:56:15 partition for image 87 refined in 4.86 sec
2018-12-25 09:56:17 Finished processing in 1278.2 sec (exit code 0)
2018-12-25 09:56:17 Error: checkInvariant_everyVertexPositionIsJunction: bad junctions map for vertex Vertex:
2018-12-25 09:56:17 vertex.id = 3591
2018-12-25 09:56:17 vertex.position = [384, 327]
2018-12-25 09:56:17 vertex.isGood = 0
2018-12-25 09:56:17 vertex.junctionRadius = 0
2018-12-25 09:56:17 vertex.deleted = 0
2018-12-25 09:56:17 vertex.incidentEdges.size() = 2
2018-12-25 09:56:17
2018-12-25 09:56:17
2018-12-25 09:56:17 got id = -1
>>>
(http://i66.tinypic.com/etedxh.jpg)
-
Hello Diego,
Can you provide the complete log related to Build Orthomosaic operation and number of cameras in the project?
-
Hello,
I'm still having an issue when generating dense cloud or texture with this new version (Error: Unexpected channel count).
It seems to only occur when using pictures with an alpha channel.
Regards.
-
Hello jnb,
Do you encounter this issue even if you do not use Change Path command to swap from DNG to TIFF?
-
Yes, I tried on a new project, on a different set of images, directly in tiff.
The same set of images was working fine on 1.4
-
Hello Diego,
Can you provide the complete log related to Build Orthomosaic operation and number of cameras in the project?
Hello Alexey, attached complete log
-
Yes, I tried on a new project, on a different set of images, directly in tiff.
The same set of images was working fine on 1.4
Hello jnb,
Can you please share the smallest sample set of the images that can be used to reproduce the issue?
-
Hello Diego,
Can you provide the complete log related to Build Orthomosaic operation and number of cameras in the project?
Hello Alexey, attached complete log
Hello Diego,
We have managed to reproduce the issue on some synthetic project and the fix will be available in the next update. However, if you can provide any minimal sample dataset that allows to reproduce the issue, please share that with support@agisoft.com.
-
Hi Alexey,
The vertex colors for the new meshing option is working, thanks. :)
I did notice in the console pain that there is an error code showing up, highlighted on this image:
(http://www.digital-mapping.net/forums/Photoscan/2018/error.jpg)
-
Hello bisenberger,
Do you have any other GPUs except for GTX 1070 on your computer?
-
I'm still having an issue when generating dense cloud or texture with this new version (Error: Unexpected channel count).
It seems to only occur when using pictures with an alpha channel.
Hello jnb,
Thank you for providing the sample images. However, it appears that the images that are supposed to have alpha channel do have the fourth channel not marked as alpha properly, so this channel is considered as data channel and not as an alpha. Please check that the application that you are using to add the alpha is saving the data properly.
-
Hi Alexey,
No. I had an AMD card installed, but I removed it from the computer.
-
Hello bisenberger,
Maybe there are still AMD drivers then. You can use Display Driver Uninstaller or similar utilities to remove AMD graphic drivers (and Intel graphic drivers, just for case).
-
Alexey,
Just noticed ni the new Metashape user Manual that the description of the Mesh "• Refinement tool" is probably missing ?
Cheers
Mak
-
Alexey,
Just noticed n the new Metashape user Manual that the description of the Mesh "• Refinement tool" is probably missing ?
Cheers
Mak
Also use Volumetric mask not exist in build mesh?
-
Dense Cloud quality is now a lot better that in previous versions. I'm now confident to use Metashape for reliable work. However a few things failed in batch processing. Yesterday I had two failures in mesh and in orthomosaic generation that were, after, correctly builded manually.
-
I'm really missing the 'filter by class' button in my toolbar :'(
I use classes a lot, and being able to quickly filter, without going trough the menu, sped up my workflow by a lot. Can you give me my button back? ::)
(the new automatic classification is pretty nice though ;) )
-
Hello DrieS,
You can assign a shortcut to filter... and reset filter options via General preferences tab, if you use them frequently.
-
Ok, shortcuts it is then for now!
It would be nice though to be able to customize the toolbar somewhat in the future.
Also; when exporting shape profiles from metashape it doesn't include the values anymore as it did in photoscan. Is there a way to fix this?
-
HI DriesS
Seems they are there... just in a very grey font (if you have eye saver mode activated on your screen, then you will see the data labels clearly.
Something to do with display settings I imagine
Regards,
James
-
on the chrome and firefox both of them shows the message asking me to agree although I have already checked the box. please resolve this issue if yo find out solutions
-
Hello to everyone i also get same error without Refine Seamlines option
-
i also face same issues kindly resolve this and give us solutions
-
Alexey,
Looks like there's something strange happening in 1.5.
Some ot the depthmaps generated don't look right (totally blue) compared to the ones from 1.4.5.
Edit: Also, depth map generation seems to be slower (testing on Ultra High) even compared to previous build of Metashape (pre-releases of 1.5)
Mak
-
Something seems definitely wrong with the depth map generation in the latest build. I'm reprocessing a project this time using the new meshing method (without re-using the old depth-maps so its creating new ones) and as you can see from the screenshots below it's going to take nearly twice as long as it took with a previous version of Metashape (it took 3h56 now it is past 4h with still 3hrs left..).
UPDATE: It took 5h10 to process the new depth maps compared to 3h54 with a previous build.
Mak
-
Alexey,
I did some more testing:
Totatly uninstalled and ereased any trace of Metashape & Photoscan (deleted all registry entries etc) and re-installed the latest build of Metashape. Rolled back to the 18.5.1 Radeon driver after clean unstalled with DDU just in case the GPU was acting funny with the latest release.
Re-ran another set & generated depth maps via the new method to compare processing time with an older build of Metashape and as you can see in the screenshots below the latest build was 37mins slower (2h23). The newest depth maps are also different (maybe a bit more detailed).
Regards
Mak
-
The depth maps are very good in this version. In the latest ones they were not so good. For those that are using UAV surveys in evironments with bushes or trees, this version is probably the best ever. Mak11, I would not sacrifice this quality by processing speed. Try to lower one quality step if you do not mind to get speedier processing times sacrificing quality :-)
-
RHenriques,
I agree that having better depthmaps is a good thing.. But in my case (I don't do UAV projects) it's totally not worth it given the small difference in quality VS big performance penalty especially when the final mesh is not even better (it's actually worst in two projects I tested yesterday). Going down to High is not an option unfortunately as I need the highest detailed mesh possible for baking later.
Mak
-
Quite happy with the new meshing based on depth maps, I get much more quality for a given calculation time and more importantly, for a given set of tie points and therefore still photographs, since re taking field work takes a lot of time, energy and money. I still think that there should be a thresold for generating very small detail based on displacement mapping on subdivision surfaces, since Agisoft generates good meshes anyway. This way I could squeeze more render quality and diffuse consistency in pos rendering operations in 3D packages.
Also I would welcome more advanced operations and workflows with regions (Multiple regions, boolean logics, non destructive workflows, etc). Anyway Agisoft rocks, I just sampled a 17 hectares old city with Agisoft and it is delivering, compared to existing GIS data, xtremely good relative accuracy even without CGP points !!! Of course there is a good workflow involved but heck I am impressed.
-
The new meshing method already greatly result in higher detail models with the "older" depthmaps and from my own testing the newer ones which are on average 30% slower to generate are not really improving things at all (in my particular use cases).
But I do admit that I'm probably in the minority.
Mak
-
I'm getting a lot of crashing when running Build Tiled Model with Quality set to Ultra high. Build 7492.
Here is a link to my log file:
http://www.digital-mapping.net/forums/Metashape/Jan2019_log.zip (http://www.digital-mapping.net/forums/Metashape/Jan2019_log.zip)
-
Hello bisenberger,
Does it seem similar to the following bug report:
https://www.agisoft.com/forum/index.php?topic=10222.0
However, in the log it seems that the tiled model generation procedure has been completed successfully.
Or are you referring to "clGetPlatformIDs failed: CL_UNKNOWN_ERROR_CODE_-1001" lines?
-
Hi Alexey,
It is crashing my computer (blue screen), so maybe it isn't saved to the log. When it crashes my computer the blue screen error code has been different each time. The routine does complete sometimes, and that's probably what you see in the log.
I'll look at the info in the post you refer to.
The "clGetPlatformIDs failed: CL_UNKNOWN_ERROR_CODE_-1001" is something I reported earlier in the thread. My guess is, it has something to do with a graphics card that is no longer installed in the system. I have used the AMD cleanup utility, but the code is still there.
-
Forgot to mention that I don't have the problem when using the other quality settings. I only have the problem when the quality is set to Ultra High.
-
Hi Alexey et al.,
I would like to make a point regarding visibility consistent mesh generation option. I have started running a couple of experiments using Agisoft Cloud.
There does not seem to be an option visible in my build (7492, Mac OS X) to set the depth filtering. When running as a non-batch process, interpolation and depth filtering options appear greyed out. Are they locked? I am not sure, as when I attempted to run mesh generation as a batch process from depth maps, the process failed (355 cameras, planar surface, highest quality). Depth maps were saved (see log: 'downloading 3.xxxGB'), but they came out with aggressive filtering. The reason for failure was not given, but when attempting to run the process again, the error 'null dense cloud' was returned. Obviously something has gone awry somewhere along the road....
Given I work with rock art, and the best images that I capture have little or no noise, it is important that small details are not discarded as outliers, which I have observed can occur when using filtering (smoothing options). There is also the additional benefit of reduced computational cost. Therefore, it is for these reasons I ask about filtering options that many do not completely disable as a matter of course. All this you may of course know already.
I can see the benefits of visibility mesh generation, and Diana Ovod has previously recommended that I try out the method. Unfortunately, due to reasons beyond my control, I have not had the chance to test it on HPC until the launch of Agisoft Cloud. I am particularly interested if the method reduces RAM demands (although we unfortunately can't see performance readouts from EC2), which can be a problem for me, and hence Diana's recommendation. There is also the time saving element through full exploitation of parallelism. I know there are some of our colleagues saying things are now taking longer, but lets see...
Regards,
James
-
Hello James,
Currently depth maps based mesh generation option is creating the depth maps with the Mild Filtering by default and requires it on input, so depth maps calculated with the different filtering option cannot be re-used for this method (at least in the version 1.5.0). There's, however, a workaround how to make it possible to re-use aggressive and moderate depth maps - it's necessary to modify the meta information of the depth maps instance in the chunk, for example, using Python console:
Metashape.app.document.chunk.depth_maps.meta["depth/depth_filter_mode"] = "0"After that the available depth maps could be re-used. However, the Mild option tends to produce better results in most projects.
In the version 1.5.0 the depth maps generation procedure has been revised and optimized, so you shouldn't expect noise for mild filtering depth maps. Moreover, new meshing method should be good for internal noise suppression.
I can try to compare the processing time for some rock-art / stone carving projects that we have for dense cloud and depth maps based meshing methods (both for mild and aggressive filtering) and provide the comparison chart. the new meshing is exploiting GPU for reconstruction, so it can be accelerated with the good discrete GPU cards, whereas the dense cloud reconstruction (especially dense cloud filtering part) time is quite dependent on the image overlap, while the old meshing method could be very RAM demanding for large datasets (hundreds of photos) in High / Ultra quality.
As for the issues that you have been observing with the cloud, could you specify, when does it happen?
-
Hello James,
Currently depth maps based mesh generation option is creating the depth maps with the Mild Filtering by default and requires it on input, so depth maps calculated with the different filtering option cannot be re-used for this method (at least in the version 1.5.0). There's, however, a workaround how to make it possible to re-use aggressive and moderate depth maps - it's necessary to modify the meta information of the depth maps instance in the chunk, for example, using Python console:
Metashape.app.document.chunk.depth_maps.meta["depth/depth_filter_mode"] = "0"After that the available depth maps could be re-used. However, the Mild option tends to produce better results in most projects.
In the version 1.5.0 the depth maps generation procedure has been revised and optimized, so you shouldn't expect noise for mild filtering depth maps. Moreover, new meshing method should be good for internal noise suppression.
I can try to compare the processing time for some rock-art / stone carving projects that we have for dense cloud and depth maps based meshing methods (both for mild and aggressive filtering) and provide the comparison chart. the new meshing is exploiting GPU for reconstruction, so it can be accelerated with the good discrete GPU cards, whereas the dense cloud reconstruction (especially dense cloud filtering part) time is quite dependent on the image overlap, while the old meshing method could be very RAM demanding for large datasets (hundreds of photos) in High / Ultra quality.
As for the issues that you have been observing with the cloud, could you specify, when does it happen?
Hi Alexey,
Thanks for the swift response. Perhaps it wasn't so clear in my message, but what I would like to do is to disable filtering. Just to reiterate, with well exposed, focused images, taken in strong light, I am able to process without any filtering because there is no noise evident in the dense surface reconstruction (though I note what you say and have observed myself that version 1.5 has revisions and optimisations in this regard).
Therefore, if I have generated the depth maps via the usual procedure without depth filtering, should the workaround in python work for depth maps generated without filtering too?
In the meantime, to get an idea of the output I have set EC2 to work again with mild filtering, but not as a batch process.
As to when the error with the cloud happens... see the attached log. I think I forgot to attach it last time.
Regards,
James
-
Hello James,
The workaround with Python would work for any depth maps filtering option, it just changes the meta information for the available depth maps to "Mild".
As for the "null dense cloud" problem, what were the processing parameters of the task that you have sent for processing in the cloud?
-
Hello James,
The workaround with Python would work for any depth maps filtering option, it just changes the meta information for the available depth maps to "Mild".
As for the "null dense cloud" problem, what were the processing parameters of the task that you have sent for processing in the cloud?
Hi Alexey,
The processing parameters for the log previously supplied were according the attached screenshot. The process was then started and left to its own devices. When I synced the project, a set of depth maps with aggressive filtering were delivered back. There was not explanation why. I then tried to run the process again, just as you see, but with the reuse depth maps option checked. It was then that the "null dense cloud" error message was returned.
As you have pointed out, the program will have of course become confused, given the default settings in operation as you have previously outlined.
Regards,
James
P.S. Update. I have run the project with the default settings for visibility consistent mesh generation. The process still failed. Depth maps were generated in 1hr33min, with aggressive filtering, despite this not being specified. The project has then synced to the local machine, with no error message. I will now try specifying the set of depth maps using python.
-
Hi
One thing I have noticed with new the meshing method based on depth maps is longer times to build textures, which is expected working with more detailed meshes, but I would be great if this department could get some boost in future versions.
-
Hello James,
The workaround with Python would work for any depth maps filtering option, it just changes the meta information for the available depth maps to "Mild".
As for the "null dense cloud" problem, what were the processing parameters of the task that you have sent for processing in the cloud?
Hi Alexey,
The processing parameters for the log previously supplied were according the attached screenshot. The process was then started and left to its own devices. When I synced the project, a set of depth maps with aggressive filtering were delivered back. There was not explanation why. I then tried to run the process again, just as you see, but with the reuse depth maps option checked. It was then that the "null dense cloud" error message was returned.
As you have pointed out, the program will have of course become confused, given the default settings in operation as you have previously outlined.
Regards,
James
P.S. Update. I have run the project with the default settings for visibility consistent mesh generation. The process still failed. Depth maps were generated in 1hr33min, with aggressive filtering, despite this not being specified. The project has then synced to the local machine, with no error message. I will now try specifying the set of depth maps using python.
Hello James,
There might be a problem with the new meshing method started in the cloud solution, we'll try to fix it as soon as possible. Currently looks like the arguments of the task are not properly interpreted by the cloud and it still tries to build the mesh from the dense cloud, while there's no cloud ("Null dense cloud" error is raised).
-
Hello James,
Currently depth maps based mesh generation option is creating the depth maps with the Mild Filtering by default and requires it on input, so depth maps calculated with the different filtering option cannot be re-used for this method (at least in the version 1.5.0). There's, however, a workaround how to make it possible to re-use aggressive and moderate depth maps - it's necessary to modify the meta information of the depth maps instance in the chunk, for example, using Python console:
Metashape.app.document.chunk.depth_maps.meta["depth/depth_filter_mode"] = "0"After that the available depth maps could be re-used. However, the Mild option tends to produce better results in most projects.
In the version 1.5.0 the depth maps generation procedure has been revised and optimized, so you shouldn't expect noise for mild filtering depth maps. Moreover, new meshing method should be good for internal noise suppression.
I can try to compare the processing time for some rock-art / stone carving projects that we have for dense cloud and depth maps based meshing methods (both for mild and aggressive filtering) and provide the comparison chart. the new meshing is exploiting GPU for reconstruction, so it can be accelerated with the good discrete GPU cards, whereas the dense cloud reconstruction (especially dense cloud filtering part) time is quite dependent on the image overlap, while the old meshing method could be very RAM demanding for large datasets (hundreds of photos) in High / Ultra quality.
As for the issues that you have been observing with the cloud, could you specify, when does it happen?
Good information on the new meshing. Will eliminate a lot of trail and error.
Thanks Alexey
-
[/quote]
Hello James,
There might be a problem with the new meshing method started in the cloud solution, we'll try to fix it as soon as possible. Currently looks like the arguments of the task are not properly interpreted by the cloud and it still tries to build the mesh from the dense cloud, while there's no cloud ("Null dense cloud" error is raised).
[/quote]
Hi Alexey,
Indeed, there does seem to be a problem with the arguments.
I changed the metadata, using the line of code provided. However, I think it exceeded the storage quota (I received a mail to that effect). What is a little strange is that it didn't save the file. I have exceeded the quota before by 3-7GB before, but in those cases the file was still saved in its entirety. I think it is a good feature that one is able to exceed the storage capacity temporarily, and if necessary pay for that excess. However, in this case, that did not happen. I am running mesh generation from the depth maps again to see if the same thing happens.
Regards,
James
-
Hello James,
There might be a problem with the new meshing method started in the cloud solution, we'll try to fix it as soon as possible. Currently looks like the arguments of the task are not properly interpreted by the cloud and it still tries to build the mesh from the dense cloud, while there's no cloud ("Null dense cloud" error is raised).
Hi Alexey,
Indeed, there does seem to be a problem with the arguments.
I changed the metadata, using the line of code provided. However, I think it exceeded the storage quota (I received a mail to that effect). What is a little strange is that it didn't save the file. I have exceeded the quota before by 3-7GB before, but in those cases the file was still saved in its entirety. I think it is a good feature that one is able to exceed the storage capacity temporarily, and if necessary pay for that excess. However, in this case, that did not happen. I am running mesh generation from the depth maps again to see if the same thing happens.
Regards,
James
Hello James,
As for "exceeded the quota" I suppose it was a problem with our email engine, recently it resent some messages that had already been sent.
-
Hello James,
There might be a problem with the new meshing method started in the cloud solution, we'll try to fix it as soon as possible. Currently looks like the arguments of the task are not properly interpreted by the cloud and it still tries to build the mesh from the dense cloud, while there's no cloud ("Null dense cloud" error is raised).
Hi Alexey,
Indeed, there does seem to be a problem with the arguments.
I changed the metadata, using the line of code provided. However, I think it exceeded the storage quota (I received a mail to that effect). What is a little strange is that it didn't save the file. I have exceeded the quota before by 3-7GB before, but in those cases the file was still saved in its entirety. I think it is a good feature that one is able to exceed the storage capacity temporarily, and if necessary pay for that excess. However, in this case, that did not happen. I am running mesh generation from the depth maps again to see if the same thing happens.
Regards,
James
[/quote]
Hello James,
As for "exceeded the quota" I suppose it was a problem with our email engine, recently it resent some messages that had already been sent.
[/quote]
Hi Zakhar & Alexey,
Interesting about the email engine Zakhar. The process has just failed again with the 'null dense cloud' error when trying to run reusing the depth maps. Therefore, Alexey's conclusion holds.
Regards,
James
-
Is there a way to export a Tiled Model in any of the formats that a 3D Model can be exported in?
The area of model actually isn't that large, I just can't as good of results with the other meshing methods.
I would like to take the Tiled Model into another 3D application for editing and retopo.
-
Hello bisenberger,
You can try to export the model using Agisoft Tiled Archive format in .zip archive, then unpack it and see the common models related to the tiles of the different zoom level.
In principle, it should be possible to open in some model editing application the models related to the bottom (most detailed) level and merge them, however, note that the tiles are generated with some overlap and there wouldn't be a possibility to import the tiled model back.
-
Thanks Alexey,
I had a look at the zip. I can load the models but the placement of the many mesh pieces are all centered at the origin (on top of each other). It would be a huge job to snap all the pieces to place. Put it on the wish list to be able to export a higher resolution mesh from a Tiled Model in a more common format.
The thing that the Tiled Model did that I can't duplicate with the other methods is to remove a lot of trees and give me a good mesh of the ground surface. Do you have any suggestions on how I can achieve this with the other meshing methods?
Bill
-
Hello Guys,
Just a "heads up" from me. I encountered the error message "unexpected channel count" during building a dense cloud.
Apparently this error occurs when one or more images have more than one alpha channel
If you have multiple alpha channels in an image, delete the ones you do not need and you should be good to go.
Best Regads,
Chris
-
Hi Alexey,
I was mistaken about the quality of the other meshing options. Build Mesh with Depth maps Ultra high is very close to the Tiled Model. What made me think it was different was that I was using Split in chunks script with Build Mesh, Ultra, and Arbitrary settings. I assumed that it used the same meshing routine as Build Mesh with Depth maps Ultra high from the Workflow pulldown. Apparently that is not the case. Would it be possible to modify the script to use the same meshing routine as the Workflow/ Build Mesh command with Depth maps?
Bill
-
Hi,
I tried the last Version metashape yesterday but i have a Problem with showing my photos. After a doubleclick on a Photo i See the tiepoints and the Markers, But there is no Photo and the horizontal Photo-Menü is empty.
I have a lapop with i7 and gpu NVIDIA 965, the buildin gpu is dismarked in the Settings.
Photoscan is working error-free.
Thanks
Dieter
-
Hello Dieter,
You need to install latest NVIDIA drivers.
-
It works,
thank you.
-
I was mistaken about the quality of the other meshing options. Build Mesh with Depth maps Ultra high is very close to the Tiled Model. What made me think it was different was that I was using Split in chunks script with Build Mesh, Ultra, and Arbitrary settings. I assumed that it used the same meshing routine as Build Mesh with Depth maps Ultra high from the Workflow pulldown. Apparently that is not the case. Would it be possible to modify the script to use the same meshing routine as the Workflow/ Build Mesh command with Depth maps?
Hello Bill,
Yes, it is possible to modify the split-in-chunks script and use Metashape.DataSource.DepthMapsData for the source of the mesh generation. In this case the dense cloud generation and duplicating may be omitted.
-
I received a "index Overflow" message, see attached Protokoll. The memory was ok, it was at 50%.
This is a Projekt with 3030 20mpix Pictures.
I split it in three chunks, but it doesn't work.
I want to create it in best settings. With how many Pictures i can work in a single chunk?
Thanks
Dieter
-
I received a "index Overflow" message, see attached Protokoll. The memory was ok, it was at 50%.
This is a Projekt with 3030 20mpix Pictures.
I split it in three chunks, but it doesn't work.
I want to create it in best settings. With how many Pictures i can work in a single chunk?
Thanks
Dieter
It seems to vary. I've had Tiled Model complete on Ultra High setting with up to 400 12MP photos and I have had it fail on a projects with as few as 264 12MP photos, with out of memory warning. My computer has 64GB of ram. I have used split into chunks also, but the results aren't as good as with a single chunk, and the preset options for Face count, High, Medium, and Low are also missing in the batch process dialog for Build Mesh and Build Tile Model.
-
Hello Dieter,
We are working on the fix for this issue, however, it would be likely implemented in the version 1.5.2.
At the moment I can suggest to use High quality instead of Ultra. Or even Medium to begin with.
-
Ok, thank you, i will try it.
Next Problem. I have a Pointcloud from a scanner, i Import it in Metashape (left Side in the Picture).
Then i mesh the pointcloud and the result is the Right side in the Picture. there are big holes. I#ve tried everything, it doesn't work.
Thanks
Dieter
-
Hello,
I try to test cloud processing option, but i've received the following error:
"Open SSL library not found....".
Then i've installed Open SSL library from this link https://slproweb.com/products/Win32OpenSSL.html (https://slproweb.com/products/Win32OpenSSL.html), but error still persists.
Have you got any idea "why?".
-
Hello adam_s,
Cloud processing option requires OpenSSL 1.0.1 or 1.0.2 installed.
I can suggest to install both 32-bit and 64-bit libraries of the same version (for example, 1.0.2) and check if it helps.
-
@khedar
I encountered the same error.
@agisoft support team
I want your help.
I found the answer. Sorry.
Best regards,
kaz