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 - ManyPixels

Pages: 1 2 [3]
31
General / Re: Wide angle lens for full frame Nikon mirrorless camera
« on: December 05, 2022, 01:02:48 PM »
When using lenses with noticeable distortion, it is interesting to launch a project or part of the project to be processed, to make a good alignment and then to export the calibration. You can then import it back into the final project and fix all the parameters. This greatly helps in getting good matches.


You can of course release the parameters once the alignment is finished and run an optimization to improve the result.

32
General / Pricing in 2.0
« on: December 05, 2022, 12:55:04 PM »
With the arrival of LIDAR and Laser data integrations, wouldn't it be time to split the sale into 3 versions?

1.Standard – $180
2. Pro – $900 and functions identical to 1.8 (without the bugs)
3. Ultra – $3500 and functions identical to 2.0 (without the bugs)

This way no one will have paid too much for the features they needed. Pros would become Ultra and a new category would appear. For sure some customers will say "I bought Agisoft 6 months ago and don't need to import laserscans, so I've paid an extra 2600$ for nothing" but there sure will be solutions to face this type of problem. I see no other problem and it's a win-win operation.

33
General / Manual image matching
« on: December 05, 2022, 12:52:35 PM »
One more feature that has already been requested: The ability to manually tell the software which images match. It is not uncommon for two images to not match when they clearly do. You would have to be able to tell the software to “match these two images together” and find the matches by any means possible. The only way to do this currently is via markers, but this is very time consuming and manual operation is expensive compared to automatic operations.

34
General / Can’t set mesh CRS in empty project
« on: December 05, 2022, 12:51:54 PM »
When opening an empty project, setting the CRS wherever possible and trying to import a mesh with a well-defined coordinate system, it is only possible to import it in local coordinates. On the other hand, if you import a dense cloud, regardless of its CRS, it is then possible to import the mesh and choose its CRS (even if it is different from the dense cloud)

35
General / Better overall RAM management
« on: December 05, 2022, 12:50:45 PM »
General memory management is simply abysmal. How can you run out of memory with 256 GB available? It may be understandable to need 64GB for large projects, but it should never be more, since no step requires to simultaneously process more than 200 images in parallel (even at very high definition and counting the depth, points, and all related data, this is not 64GB).

There is nothing worse than launching a process and watching the RAM fill up endlessly until the software crashes. This should never happen. There are many programs that use all this RAM if it is available, but if it is not available, they also do very well! The ideal is always to be able to set a usable maximum. It's the same with the processor cores, it's interesting to be able to free up one or two to be able to continue working alongside. You can look at astrophotography processing software to see software that are good at managing resources. Siril and Astro Pixel Processor are very good examples.

36
General / Very poor and bad documentation
« on: December 05, 2022, 12:49:38 PM »
Alexey Pasumansky, after writing 14,000 forum posts, have you ever considered rewriting the documentation properly rather than repeating the same things over and over again on the forum? I never found anything useful in the documentation. You may find this extreme, but it is the truth. The functions and their options are explained so briefly that the only way to know which option to choose in the processes is to stupidly try all the possible solutions and compare the result. And as far as the Python API goes, that's guesswork and prayer.

The log spits out dozens of lines, but we don't really understand what's going on. A complete documentation with an exhaustive description of the functions as well as the lines which appear in the log would be welcome and would free you up a lot of time to deal with the real problems!

Nobody will blame you if you focus your resources for 6 months on fixing everything that is problematic rather than adding new features with their new bugs. On the other hand, paying $3,500 to never get to the end of a project without having problems where there shouldn't be gives you a very bad image.

37
General / List of tweaks
« on: December 05, 2022, 12:48:27 PM »
I know this topic has come up many times in the past, but it's still relevant.

After spending a lot of time on these forums, I noticed that a solution that came up often to fix a problem was to add tweaks. However, they are incomprehensible.

It would be useful to make a list of available tweaks with a clear explanation of their effect.

38
General / Stability of depth maps calculation
« on: December 05, 2022, 12:47:24 PM »
I encountered big problems during the second step of generating depth maps (depth map filtering). The bigger the project, the more likely the crash. It seems related to graphics cards (2x RX 6900 XT). Before you ask, yes my drivers are up to date (and I feel like I have more crashes with latest update 22.11.1).

I ended up succeeding in the calculation using only the CPU and I noticed that on this problematic step, the time was identical or even less than the calculation time on the GPU! About 10s to filter a block of 39 depth maps on CPU (AMD Threadripper Pro 3955WX) and about 12s on the 2 GPUs.

It would be good to be able to adjust the parameters of material use more finely by step. A better division of steps would also allow for better monitoring. Here I restarted the calculation about 10 times on GPU with systematically a crash at a random point in the 2nd step, i.e., after 3 hours of calculation. So, I will have lost 30h while the whole thing only took 10h on CPU and working the first time.

39
General / Merge chunks: resources consumption and method
« on: December 05, 2022, 12:44:57 PM »
Calculation by chunk is useful on very large projects, but unfortunately if you want to merge tie points on large projects, memory management is catastrophic (out of memory with 256GB available). The steps look identical to the alignment, so why is there no fine task subdivision like with all the other processes? This would make it possible to manage large projects. Computing time is rarely a problem because the machines can be left to run on their own. What is problematic is stability. I no longer count the number of crashes encountered after 5-6 hours or more of calculation. It's wasted time.

The method also seems weird, the images are duplicated, and this causes problems later when calculating the depth (the identical images are no longer exactly in the same place, which creates an infinite depth). What we want after having aligned the chunks by images, it is simply that the identical images are merged and their correspondences also (either in an additive way for the couples being in the two chunks, or by recreating the correspondences only for the couples present in the two chunks, not for everything!).

40
After performing several tests, I notice that the outliers are better managed by using “Build texture” than using “Build Orthomosaic”. There are also fewer disparities in brightness and sharpness.
Also, performance is better with mesh texturing.
It would therefore be nice to have an option to extract an orthomosaic directly from the texture of a mesh (top view, side view, etc.). It is very convenient to make interior orthomosaics, per floor. But currently, the only way to do this is to successively cut the mesh to generate an orthomosaic, which recalculates all the orthoimages at each step, even with same projection. The performances are therefore very poor when what we want is simply a cut of the textured mesh (like a screenshot of the workspace at high resolution, which means it could simply be an option when capturing view).

Pages: 1 2 [3]