-
Notifications
You must be signed in to change notification settings - Fork 7
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
No Node able to be Processed default pipeline #46
Comments
Traceback (most recent call last): |
So it seems if I right click a node after this I can still click "Compute"... will report back |
[15:27:18.448603][info] CameraInit report:
==============================================================================
|
Hmm.... Second time doing that same right click->compute on the same node using default pipeline seemed to work: Using 7 CPU thread(s) for feature extraction. ==============================================================================
|
Hi @chboishabba, it looks like you eventually got it to work using the default pipeline. I see that it registered 61 out of your original 76 images...can you see any reason or pattern why 15 of the images were not registered with the others? I also notice that some of your images are 4080x3072 (landscape) and that some are 3072x4080 (portrait). This is causing MeshroomCL to compute two separate intrinsic camera models, and so it will not optimize the principal point parameters (i.e. it will assume that the optical axis is perfectly centered on the camera sensor). You might get slightly better StructureFromMotion results if you rotate all your images to be the same dimension, or if you try the "OpenCL MVS" pipeline (which uses a different StructureFromMotion algorithm to locate the camera positions). |
The portrait is the likely culprit.. Fantastic work and thank you very much for providing this tool. I was obsessed with the photogrammetry tool that got bought by Discord that used the motion telemetry and gave a sparse point-cloud while you took video input. Do you know of any tools that do similar these days? Is it outputting which set of images are in which group? I'm not sure what I did to make it work..it might have been giving me an error telling me there were those two groups but i just kept pressing the process button and it made an output eventually... I will try rotating all correctly..is it useful to report back the result? |
[15:17:36.859124][info] CameraInit report:
- # views listed: 76
- # views with an initialized intrinsic listed: 76
- # views without metadata (with a default intrinsic): 0
- # intrinsics listed: 2
Does not appear to create logs but happy to provide if you could share the expected path.
It seemed the app wanted to work fine with OpenCL MVS but I haven't let it run past very early stages. Will continue running this mode and confirm success.
The text was updated successfully, but these errors were encountered: