Repository files navigation

Ruilong Li ·Julian Tanke ·Minh Vo ·Michael Zollhoefer
Jürgen Gall ·Angjoo Kanazawa ·Christoph Lassner

ECCV 2022

Given multiple sparse video views as well as 3D poses as inputs, TAVA creates a virtual actor consists of implicit shape, apperance, skinning weights in the canonical space, which is ready to be animated and rendered even with out-of-distribution poses. Dense correspondences across views and poses can also be established, which enables content editing during rendering. Without requiring body template, TAVA can be directly used for creatures beyond human as along as the skeleton can be defined.

conda create -n tava python=3.9conda activate tavaconda install pytorch cudatoolkit=11.3 -c pytorchpython setup.py develop

• Seedocs/dataset.md on how to accquire data used in this paper.
• Seedocs/experiment.md on using this code base for training and evaluation.
• Seedocs/benchmark.md for replication of our paper.

• tools/extract_mesh_weights.ipynb: Extracting mesh and skinning weights from the learnt canonical space.

• tools/eval_corr.ipynb: Evaluate the pixel-to-pixel correspondences using ground-truth mesh on Animal Data.

BSD 3-clause (see LICENSE.txt).

• TrainingNeRF on NeRF synthetic dataset: Seedocs/nerf.md for instructions.
• TrainingMip-NeRF on NeRF synthetic dataset: Seedocs/mipnerf.md for instructions.

@inproceedings{li2022tava,title={TAVA: Template-free animatable volumetric actors},author={Li, Ruilong and Tanke, Julian and Vo, Minh and Zollhofer, Michael and Gall, Jurgen and Kanazawa, Angjoo and Lassner, Christoph},journal={European Conference on Computer Vision (ECCV)},year={2022}}