31
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: February 14, 2020, 03:13:43 PM »
Any updates on the noise issue? I am noticing it in many more models.
Show Posts
This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.
32
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: February 06, 2020, 07:24:29 PM »
Alexey,
I'll email you a link to download a project.
Best,
Cody
I'll email you a link to download a project.
Best,
Cody
33
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: February 04, 2020, 03:03:54 AM »
I haven't gotten any updates from Agisoft on this, although I have made a few observations and have been able to reduce the noise significantly. I'll try to summarize below.
On models that showed increased noise I noticed that the residuals looked quite systematic (fairly strong patterns) compared to those that did not have increased noise. Parameter selection for models was done using the process described by James et al. To reduce noise I chose the parameter set that gave the lowest RMSE without "fit additional corrections'. Once this was determined I optimized with "fit additional corrections' enabled. Finding a suitable parameter set first seemed quite important. The resulting dense cloud was very close to what I was seeing in 1.5.3. The accuracy with 'fit additional corrections' is also improved as verified by measurements extracted from the model and compared to 3rd party measurements.
It became quite obvious if the dense cloud was going to have noise or not by the residuals. If they were systematic the dense cloud ended up with significant noise and if they weren't the dense cloud was quite good.
I do notice different results with different lenses. I get better results, and better looking residuals, with my 21mm than I do with my 18mm. Perhaps the 18mm is harder to model.
On models that showed increased noise I noticed that the residuals looked quite systematic (fairly strong patterns) compared to those that did not have increased noise. Parameter selection for models was done using the process described by James et al. To reduce noise I chose the parameter set that gave the lowest RMSE without "fit additional corrections'. Once this was determined I optimized with "fit additional corrections' enabled. Finding a suitable parameter set first seemed quite important. The resulting dense cloud was very close to what I was seeing in 1.5.3. The accuracy with 'fit additional corrections' is also improved as verified by measurements extracted from the model and compared to 3rd party measurements.
It became quite obvious if the dense cloud was going to have noise or not by the residuals. If they were systematic the dense cloud ended up with significant noise and if they weren't the dense cloud was quite good.
I do notice different results with different lenses. I get better results, and better looking residuals, with my 21mm than I do with my 18mm. Perhaps the 18mm is harder to model.
34
General / Re: Sony A7rii for photogrammetry and lenses
« on: January 29, 2020, 04:20:53 PM »
Forgot to mention that in addition to the sensor moving freely when the camera is powered of, the lens also moves focus (to infinity I believe). When powered off and then back on the focus moves back to the previous setting, unless the battery is removed then it resets to infinity, but again the precision is unknown. It likely isn't the exact same position and my suspicion is that this creates an even greater error than a shift in sensor position. I would recommend using a fully manual lens and taping the focus ring if possible. If not consider my previous recommendation.
35
General / Re: Sony A7rii for photogrammetry and lenses
« on: January 29, 2020, 04:06:39 PM »
The Sony A7Rii does have sensor stabilization, which likely isn't good for photogrammetry as remotesense has pointed out. The literature recommends turning all such features off. I'm still waiting for a good paper to assess whether or not turning those features off is enough or if cameras with them should be avoided.
I have been successfully using an A7Riii with a Zeiss Batis 18 mm lens with good results. I have compared measurements taken from the models to third party measurements and note differences on the order of a few mm on models ~15 m in size. I do wonder if I would see an improvement if I used a camera with a fixed sensor.
Something to consider is that the sensor does freely move around when the camera is turned off and it is only when the camera is powered on that it is rigidly held in place. If you power the camera off during a shoot and then back on, the sensor may not be in the same position as previously. I don't know what the precision of the sensor positioning system is. I would suggest leaving the camera powered on during an entire shoot, and if that is not possible then possibly creating separate camera calibrations.
Hope that helps
Cody
I have been successfully using an A7Riii with a Zeiss Batis 18 mm lens with good results. I have compared measurements taken from the models to third party measurements and note differences on the order of a few mm on models ~15 m in size. I do wonder if I would see an improvement if I used a camera with a fixed sensor.
Something to consider is that the sensor does freely move around when the camera is turned off and it is only when the camera is powered on that it is rigidly held in place. If you power the camera off during a shoot and then back on, the sensor may not be in the same position as previously. I don't know what the precision of the sensor positioning system is. I would suggest leaving the camera powered on during an entire shoot, and if that is not possible then possibly creating separate camera calibrations.
Hope that helps
Cody
36
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: January 15, 2020, 08:25:29 PM »
Alexey,
Thank you. The tweaks are the same in both versions. The only one that is listed was there by default (main/refine_max_gpu_multiplier).
The alignment parameters are slightly different between them:
1.5.3
f, cx, cy, k1-3, p1-3
1.6.0
f, cx, cy, k1-3, p1-3, additional corrections
I enabled p3 in 1.6.0 in the console as it was removed from the GUI.
I noticed in the logs that the GPU driver/runtime is different:
1.6.0 is 10020/8000
1.5.3 is 10020/5050
I just finished a small test where I opened a project that was optimized in 1.6.0 with parameters noted above in 1.5.3 and processed the dense point cloud. The speed has returned to normal but the resulting dense cloud is still noisy.
Thank you. The tweaks are the same in both versions. The only one that is listed was there by default (main/refine_max_gpu_multiplier).
The alignment parameters are slightly different between them:
1.5.3
f, cx, cy, k1-3, p1-3
1.6.0
f, cx, cy, k1-3, p1-3, additional corrections
I enabled p3 in 1.6.0 in the console as it was removed from the GUI.
I noticed in the logs that the GPU driver/runtime is different:
1.6.0 is 10020/8000
1.5.3 is 10020/5050
I just finished a small test where I opened a project that was optimized in 1.6.0 with parameters noted above in 1.5.3 and processed the dense point cloud. The speed has returned to normal but the resulting dense cloud is still noisy.
37
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: January 15, 2020, 03:50:56 AM »
Thanks Mak11. The logs show GPU usage to be 100% during depth maps for both versions. 1.6.0 is just much slower. Below are durations for both. Attached is a screenshot showing the difference in noise.
1.6.0
depth map generation = 3:13
filtering = 1:41
Total ~5hrs
1.5.3
depth map generation = 1:04
filtering = 0:51
Total ~2hrs
1.6.0
depth map generation = 3:13
filtering = 1:41
Total ~5hrs
1.5.3
depth map generation = 1:04
filtering = 0:51
Total ~2hrs
38
General / Re: V 1.6.0 - Dense cloud generation time longer, increased noise
« on: January 15, 2020, 02:54:23 AM »
Alexey,
Thanks for the response. I emailed the logs as they to large to post. I will look into getting you the dataset. A few things to note between the two logs:
1.6.0
- The CPU is at 100% with very little if any variation for the duration of processing.
- The GPUs do not seem to be at 100% use very often. I monitor them with ASI Afterburner. They are much more active with 1.5.3.
1.5.3
- The CPU is quite active but not at 100% all of the time. There is some variation.
- The GPUs are much more active and the fans are almost at max speed.
One thing I notice in the logs is that the "timings" are much higher with 1.6.0.
Thanks for the response. I emailed the logs as they to large to post. I will look into getting you the dataset. A few things to note between the two logs:
1.6.0
- The CPU is at 100% with very little if any variation for the duration of processing.
- The GPUs do not seem to be at 100% use very often. I monitor them with ASI Afterburner. They are much more active with 1.5.3.
1.5.3
- The CPU is quite active but not at 100% all of the time. There is some variation.
- The GPUs are much more active and the fans are almost at max speed.
One thing I notice in the logs is that the "timings" are much higher with 1.6.0.
39
General / V 1.6.0 - Dense cloud generation time longer, increased noise
« on: January 14, 2020, 06:58:25 AM »
I have processed the same project in 1.5 and 1.6 and the dense cloud in 1.6 takes longer to generate and the result is much noisier. Have tried the max neighbors tweak = -1 with no improvement. Anyone else experiencing similar issues?
40
General / Re: Agisoft Metashape 1.6.0 pre-release
« on: November 13, 2019, 04:45:54 PM »
Could you please provide more information on the additional calibration coefficients? What additional coefficients are used compared to the standard Brown model and how are the equations different?
Thank you
Thank you
I am curious to know what Fit additional corrections option do in Optimize Cameras dialog?....
What are these corrections and when should we use this option?
Hello Paulo,
Fit additional corrections option enables additional calibration coefficients that allows a better fit of the lens distortion model. This is especially important for wide angle lens with large radial distortions, e.g. DJI Phantom 4 RTK camera.
This option is mainly intended to be used in cases when highly accurate reference data is available (RTK/PPK) and it can help to achieve higher accuracy when no GCPs are available.
41
General / Re: Adaptive Camera Model Fitting which step
« on: July 31, 2019, 01:28:45 AM »
I'm quite interested in the as well. I have noticed that when I check adaptive fitting during the alignment stage I get lower error than if I don't check it. The more surprising part is that if I optimize with adaptive fitting checked after aligning without adaptive fitting checked, the error is higher than if I simply aligned with adaptive fitting checked. This suggests that they are indeed doing something different.
42
General / Re: Strange Radial Distortion from SONY FE 24mm F1.4 GM lens
« on: July 29, 2019, 03:06:51 PM »
I see similar results, although to a lesser degree, with my A7R III and 18mm Batis. This is with stabilization and all in camera lens corrections off.
43
General / Re: Camera Calibration and Distortion Plot
« on: March 08, 2019, 01:57:06 PM »
JMR,
Thanks for the reply. I think you are right, the vertical axis being unsigned is what is making it look so odd. Including P3 and P4 do appear to improve the model based on a few observations: 1) lower control to check point RMSE ratio and 2) more randomly orientated and lower image residuals, although there is greater correlation between parameters when included (see image attached).
Thanks for the reply. I think you are right, the vertical axis being unsigned is what is making it look so odd. Including P3 and P4 do appear to improve the model based on a few observations: 1) lower control to check point RMSE ratio and 2) more randomly orientated and lower image residuals, although there is greater correlation between parameters when included (see image attached).
44
General / Camera Calibration and Distortion Plot
« on: March 07, 2019, 11:27:17 PM »
Hello,
I've been experimenting with camera calibration parameters and have processed a model with two different configurations. The configurations are as follows:
1 - F, Cx, Cy, B1, B2, K1, K2, K3, K4, P1, P2
2 - F, Cx, Cy, B1, B2, K1, K2, K3, K4, P1, P2, P3, P4
These two configurations give quite different distortion plot results and I am trying to understand the meaning of them (see image attached). The first distortion plot seems quite realistic, increasing distortion with increasing radius. The second distortion plots looks odd, but gives better RMSE on check points, and there is lower distortion in the decentering plot. Is the second plot realistic? Or is it a case of over parameterization? Any advice you have is appreciated.
I've been experimenting with camera calibration parameters and have processed a model with two different configurations. The configurations are as follows:
1 - F, Cx, Cy, B1, B2, K1, K2, K3, K4, P1, P2
2 - F, Cx, Cy, B1, B2, K1, K2, K3, K4, P1, P2, P3, P4
These two configurations give quite different distortion plot results and I am trying to understand the meaning of them (see image attached). The first distortion plot seems quite realistic, increasing distortion with increasing radius. The second distortion plots looks odd, but gives better RMSE on check points, and there is lower distortion in the decentering plot. Is the second plot realistic? Or is it a case of over parameterization? Any advice you have is appreciated.
45
General / Re: Lines in Dense Cloud
« on: January 19, 2019, 01:49:31 AM »
Mak,
Thanks for the suggestion. There doesn't appear to be any white lines in the images or any errors with the regenerated texture. See dense cloud and texture attached.
Thanks for the suggestion. There doesn't appear to be any white lines in the images or any errors with the regenerated texture. See dense cloud and texture attached.
