triXope Film Auteur LTXV 2.0.5 Node Release for ComfyUI
Deploy the new triXope Film Auteur LTXV 2.0.5 node to your ComfyUI instance, update your workflow to use the integrated Director Mode and autoregressive chunking, and test the new audio reference and NAG integration.
Deploy the node, update your workflows, test the new features, and monitor performance.
Summary
triXope Film Auteur (LTXV) 2.0.5 is a new ComfyUI node that consolidates all LTX features into a single production‑ready suite. The update removes the need for external video extenders or multiple runs, allowing users to input any number of prompts separated by "|" or a single prompt for a long clip. Key additions include Text‑to‑Video, Image‑to‑Video, Audio‑to‑Video, and experimental Image‑Reference‑to‑Video. The node now integrates Ollama for prompt enhancement, Normalized Attention Guidance (NAG), and a Director Mode that supports multi‑shot inference. Infinite length is achieved via autoregressive chunking with built‑in sliding context windows, and users can apply one or two spatial upscale passes or a temporal upscaling option to double the framerate. Face restoration is in progress, and an integrated audio mastering pass (soft limiter & normalization) is available. The node also offers sageattention, fp16 accumulation, chunk feed‑forward, optional model unloading, stage‑1 preview, and an internal real‑time ETA counter.
The release includes a single text‑to‑video workflow and plans for additional workflows. The node supports image batch input for storyboard processing, LTX Add Video IC‑LoRA Guide for reference image control, and an unload models & clear cache switch. Future features will add Prompt Relay, keyframes, RTX Super Resolution upscaler, and more.
Users are encouraged to review the feature list and settings before deployment, and to share feedback or suggestions to further improve the node.
Key changes
- Text‑to‑Video, Image‑to‑Video, Audio‑to‑Video, and experimental Image‑Reference‑to‑Video support
- Ollama integration for prompt enhancement
- Normalized Attention Guidance (NAG) integration
- Director Mode with multi‑shot inference
- Infinite length via autoregressive chunking and sliding context windows
- Temporal upscaling option that doubles framerate and improves motion
- Face restoration work‑in‑progress for artifact removal
- Built‑in sageattention, fp16 accumulation, and chunk feed‑forward for efficiency