Comments (2)
Hello,
Various things including the skeleton structure are somewhat hard-coded so be careful if you are using your own data and own network weights and want to update them.
Hence it is highly recommended that you retrain the model with your specific requirements.
However you have to do proper pre-processing before doing the same.
You have to build the database of terrain patches which are later
fitted to the motion. For this simply run the following.
python generate_patches.py
This will sample thousands of patches from the heightmaps found in
data/heightmaps
and store them in a database called patches.npz
. This
should take a good few minutes so be patient.
Now you can begin the process of preprocessing the animation data - converting
it into the correct format and fitting the terrain to each walk cycle. For this
you run the following:
python generate_database.py
This process can take some time - at least a couple of hours. It uses all the
data found in data/animations
and once complete will output a database called
database.npz
. If you want to change the parameterisation used by the network
this is probably the place to look - but note that the preprocessing for this
work is quite fiddily and complicated so you must be careful when you edit this
script not to introduce any bugs. You will also have to remember to update the
runtime to match.
Once you've generated database.npz
you can begin training.
from pfnn.
Thank you very much.
You said 'Various things including the skeleton structure are somewhat hard-coded', and I wonder what is 'skeleton structure'? Does it include the 3D locations of the joints? If I have a character which has the same skeleton basic structure (I mean the same joint index, the same father&son joints relationship) and the different joint locations, can I input the character to the trained model in this code without retraining?
from pfnn.
Related Issues (14)
- [RESOLVED] memory error when running generate_patches.py
- invalide syntax in skeletondef.py HOT 2
- Help with computation of NN inputs relating to future trajectories from joystick
- Could you release the character 3D model? HOT 1
- How can I get the mirror bvh file?
- got different result about .phase , _footstep.txt
- Where is the JOINT_WEIGHTS from?
- values of p for equation 7 in paper HOT 1
- How can I run demo on MacOS?
- How to generate the gait file HOT 1
- I'm trying to run the demo on Windows. How do I make the character move? HOT 2
- how to create mirrored versions of all captures?
- rewrite the python code for training HOT 1
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 pfnn.