T

YiandClaude Opus 4.6 15d5d02afa Split into 3 nodes for proper VRAM management

LTXShotUpscaler now ONLY does latent upscale (tiny model).
New LTXShotRefiner handles second pass sampling (main model).

This gives ComfyUI 3 separate execution steps:
1. LTXShotRenderer  → main model loaded, then offloaded
2. LTXShotUpscaler   → upscale model loaded (~200MB), then offloaded
3. LTXShotRefiner    → main model reloaded for refinement

Matches how the original workflow manages VRAM across nodes.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-05-22 19:04:51 -07:00

__init__.py

Initial commit: LTXDirectorSampler all-in-one node

2026-05-22 14:45:06 -07:00

example_workflow.json

Split into 3 nodes for proper VRAM management

2026-05-22 19:04:51 -07:00

nodes.py

Split into 3 nodes for proper VRAM management

2026-05-22 19:04:51 -07:00

README.md

Initial commit: LTXDirectorSampler all-in-one node

2026-05-22 14:45:06 -07:00

README.md

ComfyUI-LTXDirectorSimple

Simplified all-in-one nodes for LTX Video generation with Director workflow.

Replaces the chain of: LTXDirector → ConditioningZeroOut → LTXVConditioning → LTXDirectorGuide → ConcatAV → Guider → Scheduler → Sampler → SeparateAV → CropGuides → Upsampler → DirGuide2 → ConcatAV → Sampler → SeparateAV → CropGuides → Decode

With just: LTXDirector → LTXDirectorSampler (one node for the entire sampling pipeline)