Any Time, Any Space: A World-State-Centric Framework for Agentic Video Generation
Python agentic-video-generationmutli-agentmutli-agent-systemspatial-consistencyvideo-generationvideo-generator
# DiffSynth-Studio
accelerate
einops
ftfy
modelscope
imageio[ffmpeg]
pandas
peft
protobuf
safetensors
sentencepiece
transformers==4.57.6 # transformers>=5 will break DA3
# Vista4D
matplotlib
opencv-python-headless
openexr
# Depth Anything 3 (minimal dependencies for our use)
addict
e3nn
evo
moviepy==1.0.3
omegaconf
pycolmap
plyfile
trimesh
xformers
# Segment Anything 3 (minimal dependencies for our use)
decord
iopath
kernels
pycocotools
timm
# Camera UI
fastapi
uvicorn[standard]
viser