Project structure of: Doubiiu/DynamiCrafter
DynamiCrafter
Computer Vision, Deep Learning, ML, Video
configs
gradio_app.py
Create video examples with Gradio and DynamiCrafter.
lvdm
Computer vision, ML, Python, linguistic variation, memory.
basics.py
Convolutional layers, modules, normalization, instance of hybrid conditioner.
common.py
Utility functions for data processing, mixed-precision, tensor ops, and image/map detection.
distributions.py
Python distributions code for image gen, ML tasks.
ema.py
EMA PyTorch class for training and parameter management.
models
Conditional Image Generation Models: AutoencoderKL, DDPM & Adaptive Embeddings.
autoencoder.py
AutoencoderKL model, colorization, IDFirstStage, encode/decode, optimizer, logging
ddpm3d.py
3D Conditional DDPM Model for Image Generation
samplers
DDIM samplers for PyTorch image generation and adaptive outputs.
utils_diffusion.py
Timestep embeddings, diffusion schedules for efficient DDIM/DDPG models. Prevents singularities.
modules
Language processing modules for diverse tasks.
attention.py
Efficient attention components for transformer models.
encoders
OpenCLIP T5 & Resampler Encoder Modules
networks
Autoencoder and 3D object reconstruction networks
x_transformer.py
X-Transformer with positional embeddings and multi-head attention.
prompts
README.md
Generate storytelling videos from tables using open-source DynamiCrafter tool.
requirements.txt
Python library dependencies and versions.
scripts
Scripts for model inference and evaluation.
evaluation
Evaluating distributed inference and DDPM tasks scripts.
gradio
run.sh
Run Dynamic Rafter's model inference with custom configurations.
run_mp.sh
Run multi-GPU DynamiCrafter inference script.
utils
utils.py
Utility functions and config string processing for parameter counting, matching, and object instantiation.