Video Generative Models as Geometry Learner
Paper • 2608.28549 • Published • 12
import torch
from diffusers import DiffusionPipeline
# switch to "mps" for apple devices
pipe = DiffusionPipeline.from_pretrained("happy0612/GeoNeXt", dtype=torch.bfloat16, device_map="cuda")
prompt = "Astronaut in a jungle, cold color palette, muted colors, detailed, 8k"
image = pipe(prompt).images[0]GeoNeXt
Video Generative Models as Geometry Learner
GeoNeXt is a unified framework for monocular depth and surface normal estimation, repurposing pretrained video generative models for geometry prediction through a next-frames formulation.
- Paper: Video Generative Models as Geometry Learner
- Project page: https://happy-hsy.github.io/projects/GeoNeXt/
- GitHub: https://github.com/Creative-Intelligence-Studio/GeoNeXt
- Model checkpoints: happy0612/GeoNeXt
The repository contains checkpoints for GeoNeXt-Wan and GeoNeXt-SVD. Follow the instructions in the GitHub repository to run inference for depth and normal estimation.
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support