Open marian-beszedes opened 7 months ago
Hey @marian-beszedes,
sorry I dont fully understand your question. Why do you need to place your code inside the blenderproc folder? You could also just place your custom renderer next to your other own code (e.g. your main.py script).
Describe the issue
I want to add support for synthetic defects generation on synthetic rendered data. My defects are texture based and I know exactly where are they located on texture of 3d objetcs. I can calculate where are these defects located and if they are visible in terms of bounding boxes on rendered 2d images (using Blender python API). Now I want render these defective objects and store in coco format their bounding boxes and defect type.
Texture with synthetic defects:![Screenshot from 2024-01-09 09-27-31](https://github.com/DLR-RM/BlenderProc/assets/124834518/13608b98-9c36-4a03-846b-9864247ff1bb)
Rendered image with bounding boxes for visible defects:![Screenshot from 2024-01-12 15-31-58](https://github.com/DLR-RM/BlenderProc/assets/124834518/1eba39e7-fb71-4823-a936-eab57d19b3cc)
The question is: Where is the best place to write the code which calculates the position of defects in rendered images?
Minimal code example
No response
Files required to run the code
No response
Expected behavior
I would say that there should be
DefectBoxMapRenderer.py
inblenderproc/python/modules/renderer/
which could generate 2d images with colorboxes, each for one defect. This could be than processed withbproc.writer.write_coco_annotations
. Am I right ?BlenderProc version
2.5.0