Friendly LTX-2.3 T2V+I2V+FLF+Lipsync — Unified Omni v1.0

No subgraph version - just clean Friendly UI

OmniVoice implementation for voice cloning

Unique denoise parameter, which allows to generate up to 20 steps on second pass for better quality

Quality Boost mode with Multimodal Guider implementation - better with Dev model and 30 steps

Use any Dev/Distilled/GGUF model, easily switch between models

Generate with or without audio (t2v and i2v)

Links to best models and loras

Detailed tips