Open hcw70 opened 3 years ago
Hi, Could you contact me at fabien.castan@mikrosimage.com to setup a confcall to discuss about it?
Here are two links on the topic:
https://www.reddit.com/r/photogrammetry/comments/anyr4s/focus_stacking_with_an_automated_turntable/
Perspective-consistent multifocus multiview 3D reconstruction of small objects https://arxiv.org/pdf/1912.03005.pdf
The option to generate a depth map from the image stack is really interesting. Here is a nice demonstration: https://github.com/rAum/FocusStacking (pre-compiled binaries)
@fabiencastan Lets discuss.. I did sent you a mail....
@natowi Thanks for the links. Interresting paper.... 8->
Anything happened in this regard? It would be amazing for macro photogrammetry workflows if Meshroom did the focus stacking automatically
@pepae You could use existing tools like https://github.com/PetteriAimonen/focus-stack or https://pypi.org/project/focus-stack/ and write a new Meshroom node to call a stacking tool. (Similar to this) The input should ideally be one folder with one subfolder per stack of images (at the moment, all the tools I know of, read all the images from on folder only and output one stacked image) otherwise one could define a batch number of images for stacking. The output should be multiple stacked images in one folder. (Example node: https://github.com/natowi/meshroom_external_plugins/blob/master/InstantMeshes.py) Documentation https://meshroom-manual.readthedocs.io/en/latest/feature-documentation/core/nodes.html Additional notes: https://github.com/natowi/meshroom_external_plugins/wiki/2-DevDoc
I was thinking if Meshroom can be used to create a 3D model from a set of focus stacking images. By focus stacking i mean macro photography where i.e. an insect is shoot with shallow depth of field with many images (50-200) while moving the camera towards the insect. So different depth slices become in focus during the sequence.
"Normal" focus stacking software (Helicon Focus / Zerena stacker) create a plain 2d image, (but can produce a 3D depth map based on the image alignment and the sharpness estimation).
Is it possible to setup a pipeline in Meshroom which creates a 3D model from a focus stack?
Roughly processing:
See https://stackoverflow.com/questions/15911783/what-are-some-common-focus-stacking-algorithms