Bob Thiry

>ValueError: image has wrong mode see https://github.com/thygate/stable-diffusion-webui-depthmap-script/issues/241#issuecomment-1574171475

>OSError: cannot write mode I;16 as JPEG It's trying to save a 16 bit single channel image as jpeg, which does not support this format. This can be due to...

> @thygate I hoped this was bandaid-patched in [0b7d7ef](https://github.com/thygate/stable-diffusion-webui-depthmap-script/commit/0b7d7ef6e19b5b71a298872ec65ed965499628eb) and also [1065ec1](https://github.com/thygate/stable-diffusion-webui-depthmap-script/commit/1065ec10919fb6798c306f7d6783064a446d5cfb). Offtopic: thank you for your email, did my response got into your inbox? thanks for the email. yeah...

> My tool of choice for Python is PyCharm, it debugs nicely I guess :) That works too ;) Hopefully this last fix works, I noticed the other day that...

Could you try adding this to install.py at the end, without the `if not launch.is_installed("timm"):` check : `launch.run_pip("install --force-reinstall timm==0.6.12", "timm requirement for depthmap script")`

if you don't care about the inpainted parts, why not use the simple mesh with occlusion, that takes only seconds to generate and you can delete the faces and use...

seems like a package version incompatibility with numba .. You can remove the numba imports, remove the lines that start with @njit and replace the prange with range, only the...

The inpainting is slow yes, the simple mesh is very fast, have you tried that ? Colors are stored per vertex in the exported formats (wavefront obj and ply), are...

Make sure a vispy backend is installed, and that you're running on a colab with GPU .. See https://github.com/vt-vl-lab/3d-photo-inpainting/issues/23

I have not tried to run it on colab, but perhaps this guy is onto something .. remove frame buffer for jupyter https://github.com/hyqhyq3/stable-diffusion-webui-depthmap-script/commit/6573d7d4adaeb143ef99f86e03c251b06948c73e https://github.com/vispy/jupyter_rfb/