search

atonalfreerider / dancer-pose-alignment

Align 2d yolo poses to 3d pose
5 stars 1 forks source link

Fascinating! A few questions! #1

Open oliverban opened 8 months ago

oliverban commented 8 months ago

Hello!

Congrats on a fantastic project, looks interesting to say the least. I have some questions though so that I am correctly interpreting the project.

Does this output JSON skeletons in 3D space and have them animated or is it just a single "frozen" pose? Is it in global-local space or in camera space only? Could it output FBX or BVH skeletons if a converter for the JSON was built? Maybe a CSV intermediary?

Again, if the first question is true than I am excited! :)

atonalfreerider commented 8 months ago

Thank you for your interest! Definitely using this project as an experimental space. If you manage to save the jsons per the readme, they can be viewed in the unity project I linked. They are indeed 3d poses through time.

I tried making gaussian splats but my expertise is limited here.

I have not tested this beyond my initial dataset.

Also, I'm hopeful that the latest code from https://github.com/sanweiliti/RoHM will make a lot of my code obsolete. I'll test and incorporate as soon as it is published.

Thank you for your interest

oliverban commented 8 months ago

Thank you for your interest! Definitely using this project as an experimental space. If you manage to save the jsons per the readme, they can be viewed in the unity project I linked. They are indeed 3d poses through time.

I tried making gaussian splats but my expertise is limited here.

I have not tested this beyond my initial dataset.

Also, I'm hopeful that the latest code from https://github.com/sanweiliti/RoHM will make a lot of my code obsolete. I'll test and incorporate as soon as it is published.

Thank you for your interest

Ah, nice, thanks so much, I'll check the unity project!

I've seen RoHM and am looking forward to seeing it in action for myself! Have you been able to discern when that might be?

No problem, and again, thanks for the hard work!