Forum

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.


Messages - Vlad

Pages: 1 [2]
16
What is the best tradeoff? Better go with a lower density cloud and higher quality model? Should we downscale the photos to save up RAM? We'd want both the final texture and model to be as detailed as possible of course so any advice on an optimal workflow would be greatly appreciated!

If you'll ask this question on Sketchfab forum me and some other members can answer with some ideas based on many experiments. Just because not so good discuss about cons and pros of app and competitors on forum one of this app.

17
General / Re: Underlying Algorithms for PhotoScan
« on: July 16, 2016, 04:49:30 PM »
As I understand from old days nothing changed.
After some comparison with other modern photogrammetry software commercial and opensource I found that only photoscan required masks. And create mesh with known amount of polygons that reproduce dense cloud shape with all it noise.
After read some of research papers with dense, mesh etc. reconstruction I understand that
all photogrammetry work with 3D volume in Photoscan stop after dense cloud generation.
After this step is only "brute force" generation of mesh with other post processing of this data.( poison, decimate, laplassian smooth, etc)
While already exist different algorithms that use depth map data in mesh reconstruction and mesh refinement steps.
Autodesk Remake, Bentley ContextCapture, CapturingReality and OpenMVS use such algorithms. And results in this programs or toolsets is looks clean and have more details (details != noise) than from Photoscan.

I hope Agisoft team made something with this and shift quality of it great app to quality comparable with other modern software by changing current used but autdated algorithms to modern.


18
Interesting post in sketchfab blog about upcoming 3Digify software.
https://blog.sketchfab.com/try-3digify-photogrammetry-software-free/
With some examples that looks amazing.

Software as is, looks like steps back in usability, may be because of early beta.
But technics them use can be expanded to any photogrammetry software, especially in Photoscan, because it allow use different photo for cloud calculation and for texture calculation.

19
But, ContextCapture required too much manual Tie point placements, or it can not align not perfectly taken photos.
Photoscan make this part mostly automatically.

20
Men, I test Bentley ContextCapture and with High (0.5px) precision it easily beat Agisoft Photoscan with the same Ultra-High (0.5px) quality.
Also if we compare dense clouds from the same dataset Photoscan is looks like blobby texture ContextCapture looks like clean solid mesh.
Also for ContextCapture no problem calculate 600,000,000 (600mln) dense cloud on 32Gb RAM (actually 300mln for one chank).
And speed is amasing.

21
Well after testing ContextCaprute i see that in required HQ source for produce right results.
Agisoft Photoscan can produce good results from much lower quality, for example from 4K video taken by iPhone 6S.

Context Capture fast but faster failed.
Photoscan can align photos with different exposure/lightness/etc settings but slooooooow.

But damn, Photoscan Must work faster, and must be less memory hungry :(

23
General / Re: Telecentric lens settings (Focus Stacking)
« on: May 13, 2016, 10:53:51 AM »

24
After some fails, because of lack of memory in >3 days of calculations of dense clouds on not so old computer with 32Gb DDR4, i7 6700, gtx960 and IntelHD, i try to search for competitors.
RealityCapture for this moment in deep beta, and former Acute3D ContextCapture for this moment looks like most fastest and amazing in quality photogrammetry tools.

Interesting comparison between Photoscan and Acute3D
https://community.acute3d.com/forum/agisoft-photoscan-acute3d-smart3dcapture

How this is possible, when Photoscan use both GPU at 100%, and eat all memory, competitors can work >3x time faster using only about 10-20% of only one GPU (Acute3D) and required about half of 32Gb memory for work with huge (in terms of photoscan) amount of photos?
And quality of photoscan results worster than from competitors :(

This is worst situation when calculation take 2-3 days but you can lost all your work, because you can't save and resume?
I love UI/UX and power of Photoscan but memory and speed is real problem.

Thank you.

Pages: 1 [2]