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.

Topics - igor73

Pages: [1] 2 3
General / Change path to photos not available
« on: September 20, 2016, 09:59:21 PM »

I am trying to change paths to my images.  This used to work by right cklickibg the images and chose change path.  The option is not there anymore?!  How can i change the relative path now?

Form the manual. There is no chage path in the context menu.  Screen shot attached

To change the relative paths to the photos
1. Open the context menu of the photo by right-clicking on its title in the Workspace/Photo pane.
2. Select Change Path... command from the context menu.
3. Browse to the folder containing the photo, select it and click OK.
4. Apply to the active chunk or to the entire workspace if all the photos are located in the same folder.
Alternatively, you can move either photos or the project itself with the purpose to make relative paths valid while keeping them the same.

General / Dense cloud selected cameras
« on: April 11, 2016, 03:44:15 PM »
Have noticed that when i build the dense cloud not all cameras are selected.  Today i ran a project with 1200 cameras and around 950 got selected it says in the console..  All of them where aligned,  Is this normal and why ?

I used 2 cameras. A d800 on tripod and a dji inspire drone with x3 camera.  As i mentioned before all cameras alligned. 

General / Agisoft dream machine
« on: September 09, 2015, 08:30:29 PM »
I have been asked to help specify the ultimate Agisoft processing system.  I am thinkng 4 x 10 core frequenzy optimized Xeon CPU´s . Anyone know of a good supplier for pre built workstations?  Preferably in Europe.

  GPU will probably be Titan X and at a minimum 512gb RAM.  What components would you chose?.  Mother board , HD, case , PSU etc. 

General / Re align at higher setting
« on: September 09, 2015, 08:19:16 PM »
Just found this tutorial and the author recommends aligning first at 40 000 point then 60 000 point and then 80 000 points,  He claims doing this incrementally gives better accuracy.  Is this really true?  When you re align all cameras are re-set??

Also he recommends the same method for dense cloud, medium then run again on high.  What do you think?  Seems not like a great idea to me.

General / Aligning huge projects
« on: July 25, 2015, 12:10:18 AM »
I have a huge project on my hands  with more than 14 000 images arbitrary.  I am currently  aligning in chunks of around 2500 images. Processing time for each chunk is around 130 hours.  To process all images in a single chunk would take around  700 hours  or 30 days if i wouldn't  run out of RAM.  Aligning is not that RAM hungry so might be possible on my 64GB system. 

I do not want to try a 30 day processing job and find out that Agisoft got confused and messed the alignment up. 
I have tried in the past aligning several thousand images in a single chunk and got bad results.

 I have the latest version installed.  Have there been improvements?  Should i give it a try or would you recommend processing in chunks anyway? 
One advantage of running all images at once might be that overlapping images from neighboring chunk might help alignment i am thinking.   Also it would be quite convenient to just hit start and check back next month.  Go on a holiday or something  ;D

How would you plan/align a huge arbitrary project like this? 

General / PBR textures
« on: May 05, 2015, 01:28:30 PM »
Planning to make some of my own tiled PBR textures based on scan data.  Will start out with some wood materials. 

Does anyone have a good method for generating accurate specular and gloss / roughness maps?  Normal maps  i plan on using Xnormal or Zbrush  as i have done on scanned objects.  Will use the texture/materials for UE4 and Octane render. 

Need to get rid of or reduce AO from the diffuse and i will try using HDRi tone mapped images for this purpose.   I also have large soft boxed strobes available for even lighting.  Any advice on this?

General / Titan X benchmark
« on: April 09, 2015, 03:41:28 PM »
Got my new cards 2 days ago and ran some tests.  EVGA Titan X superclocked.  Runs at about 1280mhz on boost.  Noticed GPU load was only around 60% during the dense cloud generation .  SLI was enabled during the test. 

Nvidia Titan X Benchmark

The building sample data was used

I used the following settings: (Same as Anand Tech but tie points set to 0 and CPU 0/12)

## Align Photos: High, disabled, 40000.  0 tie points
## Build Dense Cloud: Medium, Aggressive
## Build Model: Arbitrary, Dense, Interpolation=Enabled, 0 face count
## Build Texture: Generic, Mosaic, 4096, 1

System info
Microsoft Windows 7 Professional
Processor   Intel(R) Core(TM) i7-3930K CPU @ 3.20GHz
2x Titan X 12gb cards.  SLI enabled

Align photos

Finished processing in 294.814 sec (exit code 1)

Dense cloud
All CPU cores disabled in agisoft prefrences.  0/12

finished depth reconstruction in 80.37 seconds
Device 1 performance: 1004.91 million samples/sec (GeForce GTX TITAN X)
Device 2 performance: 994.448 million samples/sec (GeForce GTX TITAN X)
Total performance: 1999.36 million samples/sec
Generating dense point cloud...
selected 50 cameras in 0.22 sec
working volume: 1755x2763x1220
tiles: 1x1x1
selected 50 cameras
preloading data... done in 0.325 sec
filtering depth maps... done in 43.934 sec
preloading data... done in 7.956 sec
accumulating data... done in 1.201 sec
building point cloud... done in 0.608 sec
5706624 points extracted
Finished processing in 137.235 sec (exit code 1)

EDIT.  Did the same test on Ultra setting as i noticed the cards where onlu on 60% load.  Ultra pushed the cards to use 80-.83% and boost clock speed went up to 1316 mhz.  Result much better.  If Agisoft xould tweeks someting so 100% of the card was used evene better performance i guess. 

Ultra Results

finished depth reconstruction in 2004.43 seconds
Device 1 performance: 1326.94 million samples/sec (GeForce GTX TITAN X)
Device 2 performance: 1279.33 million samples/sec (GeForce GTX TITAN X)
Total performance: 2606.27 million samples/sec

Finished processing in 4388.76 sec (exit code 1)


Finished processing in 144.503 sec (exit code 1)

Build Texture
9056814 faces extracted in 32.709s
Calculating vertex colors...
processing nodes... done in 0.016 sec
calculating colors... done in 20.289 sec
Finished processing in 144.503 sec (exit code 1)
Parameterizing texture atlas...
Packing 2117 charts...
Blending textures...
blending textures... ************************************************** done in 49.94 sec
postprocessing atlas... done in 0.117 sec
Finished processing in 154.71 sec (exit code 1)

General / Comparing different photoscan sofwares
« on: March 17, 2015, 04:26:43 PM »
Interesting discussion on the sketchfab forum.  Agisoft is doing well.

General / Interactive 3D model viewer
« on: March 16, 2015, 01:42:41 AM »
Have an idea making a large area scan with a drone and making it interactive so the viewer can click areas of this larger area and bring up more detailed 3D models of a smaller area.  Anyone know any software that can do this?  Verold might be able to do it    Any onther options.  Must be possible to view on internet.

General / Titan X
« on: March 15, 2015, 02:18:21 AM »
This should be worth waiting for.  Holding the trigger on my new build and this news made me wonder if i should wait.  Any new CPU generations to expect for the autumn as well?

General / Duplicating chunks RAM problem
« on: February 16, 2015, 02:28:06 AM »
I have a large arbritatrary project with 3000 images. Everything is aligned and dense cloud is generated already.  I have 64GB RAM and for meshing i have to split to chunks.  In previous models i have duplicated the chunk 4 times if i need to split in 4.  I then resize the region and remove unnecessary cameras from the duplicated chunks.  Problem with this project is that its so huge that duplicating a chunk takes up about 7GB of RAM.  I tried removing all my cameras and also tried cropping the dense cloud but this only saved about 0.2 GB.  Now with 4 duplicates i have used 30GB before even starting to mesh.  Not a great start. 

So what are my options?  One option would be to not duplicate the chunk and just rezize region process then export the mesh, rezize again process and export etc.  Problem  is thats anoying because i cant set up a batch process that way.  I just want to leave the computer running and batch process all the chunks.  Also to merge the models again might be problematic this way? 
Any options i have missed?

And a feature request.  Make it possble to make several regions in a single chunk so we don´t have to use a work around for generating mesh in chunks.   BTW im using the stadard version.  Maybe Pro has this? 

General / 8000 EUR computer build recommended hardware
« on: February 06, 2015, 10:00:16 PM »

Could i get some advice on a 8000 EUR computer build please?    Will also be used for rendering with Octane render so Nvidia CUDA cards a must.  Minimum 6GB memory on the card so i guess at the moment best option is Titan or Titan Z? 

Would like to fit 256GB of RAM if possible  and plenty of CPU power.  So what parts would you buy for this budget?

General / Camera auto lens correction
« on: February 05, 2015, 11:31:08 AM »
Nikon cameras has built in lens distortion (lens correction)  processing that you can turn on and off.  Will turning it on benefit Agisoft?  Is applying lens correction to RAW images before converting to JPG a good idea or not? 

General / Aligning chunks workflow
« on: February 05, 2015, 11:07:50 AM »
I hace a very large arbitrary project of some 5000 images.  I have split this up in to 4 chunks before aligning.  It is a monumental task processing this and a chunk of 1000 images takes around 30-35 hours to align  on High setting and generic off. 

The disappointment is that aligning chunks does not seem to save any time at all really.  Aligning 2 chunks takes about 50 hours, nearly the same time.  Pre select imaged pairs is turned off , high setting, 40 000 points..  I had to divide up in chunks because Agisoft failed completely to align when i tested aligning 5000 images in one single chunk, only 3 images out of 5000 aligned.    Now splitting the project up in chunks in a  logical manner Agisoft aligners everything very well. Last chunk every single image of more than 1000 images aligned. 

How can i make the align chunk faster?  Obvious options to consider would be changing accuracy to medium and enable image pairs.  Will this effect accuracy a lot? 

Camera align is super fast but the problem here is that i did not include any cameras from neighboring chunks in the chunk so this will not work.  Maybe that would have been the best way?  Any way i can add cameras from another chunk to an already processed chunk? 

General / New Drone from DJI- Inspire
« on: November 13, 2014, 07:16:36 PM »
So what do you all  think of this drone for photogrammetry?  The camera specs are not that detailed.  Aspherical lens with no distortion and 94 degree they claim.  Sounds good..  4k video and 12mp stills.  Burst mode and RAW images.  The drone specs seems to be way better  than the Phantom.

Camera specs


94°FOV (Field Of View)

Sony EXMOR 1/2.3”

f/2.8 (20mm equivalent)
9 Elements in 9 groups
Aspherical lens element
Anti-distortion filter
UV filter
Still Photography Modes

Single shoot
Burst shooting (BURST: 3/5/7 frames,AEB: 3 or 5 bracketed frames at 0.7EV Bias)
Time lapse
HD Video Recording Modes

UHD (4K):4096x2160p24/25,3840x2160p24/25/30
HD: 1280x720p24/25/30/48/50/60
Max Bitrate Of Video Storage

Supported File Formats

Video: MP4/MOV (MPEG-4 AVC/H.264)

Pages: [1] 2 3