YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

MEGAN VIDEO

MEGAN VIDEO is an experimental open video-generation system.

Planned capabilities

  • Text-to-video
  • Image-to-video
  • Video-to-video
  • Video editing
  • Motion control
  • Character consistency
  • Scene consistency
  • Anime generation
  • Photorealistic generation
  • Toon/cartoon generation
  • Cinematic generation
  • 720p generation
  • 1080p generation
  • Long-form video generation

Training strategy

MEGAN VIDEO is developed progressively.

Initial development targets short clips and efficient training.

Later stages introduce:

  1. Better temporal modeling
  2. Image-to-video conditioning
  3. Video editing
  4. Character memory
  5. Scene memory
  6. Spatial upscaling
  7. Temporal upscaling
  8. Long-form scene generation

Checkpoint strategy

Training checkpoints are synchronized with Hugging Face so Colab sessions can be terminated and resumed.

Hardware

Initial development target:

  • Google Colab
  • NVIDIA T4
  • FP16
  • Gradient checkpointing
  • Small training resolution

Ownership

Model weights, code and training artifacts are maintained under the MEGAN project namespace.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support