Two scanned sets merged from completely different cameras/resolutions?

[MeHoo]

Long story short.. I shot video and took pictures.  They both solve separately, but what is the best method for merging them together?  Dense clouds?  Mesh?

Just curious if others have gotten better results either way.  Thanks.

[jtuhtan]

I would try to merge the clouds, then clean them up in MeshLab or some other software before creating the mesh.

For complex surfaces, merging meshes and then trying to clean them in Photoscan is probably going to be quite difficult.

[MeHoo]

Ok thanks.  I'll throw them over to Geomagic and see what I can get.  Never exported then re-imported a mesh into PScan before.  Should be interesting. 

[Alexey Pasumansky]

Hello MeHoo,

And have you tried to align both sets in the single chunk, or there's insufficient overlap between them?

[MeHoo]

I tried and it failed.  I think PScan thinks they are too different.  If I could manually align them and scale them to be closer, I wish it would then know they were supposed to be the same and merge better. 

[bigben]

Just to be sure, you tried loading both sets of images into a single chunk and the align images step failed? Can you post a sample image from each camera of an overlapping area?  I'd also check Tools > Camera Calibration and check that there are only 2 cameras with the right set of images in each.

[MeHoo]

Two chunks.  One with the video frames, one with the photo frames.  Aligning the two chunks with the photo frames as the main chunk just won't.  They look close enough to me.  I'll check the camera calibration.  Running a 30 hour process right now.  I also ran out of memory (I have 256 gigs) trying to mesh the full photo-based scan on high.  Photoscan seems really really inefficient when meshing.  When the areas of point cloud are meshed, memory should be freed up in some form of garbage collection.  I can make 70 million polygons in a number of other packages and get nowhere near that much memory usage.

[bigben]

If you try the images in one chunk, the points between the different cameras will be used to calibrate the lens parameters for both. If there are too many images for this to be practical, split them up into separate groups of corresponding views.