Comments (8)
The answer is probably that you can't - unless you have access to a 3D full body capture stage.
In order to build a new drivable identity, you need a 3D capture of that person, track it, run 3D reconstruction, fit the rig and compute joint angles as well as unwrapped textures. You would then use these assets to learn a new body decoder as described in https://arxiv.org/abs/2105.10441.
from audio2photoreal.
Is it possible to use openpose for human key capture and NerF-based human reconstruction to build a new person.
from audio2photoreal.
You'd have to collect a set of training data to learn the motion specifics of that new representation.
If you want to use the pretrained models, you'd have to match the representation of our avatars exactly.
from audio2photoreal.
The answer is probably that you can't - unless you have access to a 3D full body capture stage.
In order to build a new drivable identity, you need a 3D capture of that person, track it, run 3D reconstruction, fit the rig and compute joint angles as well as unwrapped textures. You would then use these assets to learn a new body decoder as described in https://arxiv.org/abs/2105.10441.
Thankes for your reply. I mean is whether the 3D full body capture stage can be replaced by openpose and 3D human body reconstruction, and if so, whether the details of the .npy file in the dataset can be supplemented.
from audio2photoreal.
Ah, I see. No, this will not be possible - or, it will require a new research effort way beyond what this approach is able to do right now. If I understand you correctly, you describe another body decoder that is driven by openpose keypoints. We don't have that pipeline built in this work but of course you can build it yourself.
If you don't have a 3D capture stage, you can just render our avatars from different viewpoints and treat the result as ground truth to build your openpose + NeRF method on top.
from audio2photoreal.
Ok, I'll try. Thank you for this wonderful project.
from audio2photoreal.
Anyone can please give me a detailed explanation of body_pose.npy, face_expression.npy and missing_face_frames.npy in dataset folder.
from audio2photoreal.
The data format is described here: https://github.com/facebookresearch/audio2photoreal?tab=readme-ov-file#dataset
Could you clarify what part exactly is unclear to you? Thanks!
from audio2photoreal.
Related Issues (20)
- Novel view HOT 2
- render_defaults_PXB184.pth
- Local url issue HOT 5
- evaluation code HOT 3
- what are 256 facial codes? HOT 1
- tutorial video on how to make the conversational avatar in audio2photoreal. HOT 1
- video instructions. HOT 1
- About classifier-free guidance train policy HOT 3
- How can I manually rotate an avatar's head? HOT 2
- How to pass avatar renderer conditions HOT 1
- How to change the position of camera/model? HOT 1
- Training the model with different data format HOT 1
- The lips regressor predicts unexpected result HOT 5
- Switching from Recording to Uploading Audio in a Demo: Is it Possible? HOT 1
- Why the data is not as in the README ? HOT 2
- Models and pre-requisites models unavailable HOT 3
- Does it support languages other than English? HOT 1
- Models and pre-requisites models unavailable HOT 3
- What model was used to extract the body pose ? HOT 4
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from audio2photoreal.