{"slug": "ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering", "title": "LTX 2.3 MSR: Multi-Image Reference Video With Three-Stage Rendering", "summary": "A developer published a ComfyUI workflow for multi-image reference video generation on LTX 2.3 that uses multi-subject reference (MSR) conditioning to keep subjects distinct rather than blending them into an average face. The workflow adds a three-stage rendering pass — structure, refinement and texture — each with a progressively smaller denoise budget so later stages adjust rather than rebuild, and re-anchors on the reference images between stages to counter the drift that causes a single long pass to lose references midway through a clip. The workflow JSONs are hosted on Hugging Face and require LTX 2.3 models, which are preinstalled on a RunningHub cloud instance.", "body_md": "One workflow for multi-image reference video generation on LTX 2.3, with a three-stage render enhancement pass.\n\nYou supply a set of reference images and a prompt, and the graph produces a video where the subjects\n\ncome from the references. MSR (multi-subject reference) keeps them separate instead of blending them\n\ninto an average face.\n\nThree-stage rendering is applied on top: structure pass, refinement pass, texture pass. Each stage\n\ngets a progressively smaller denoise budget so the later stages adjust rather than rebuild.\n\nReference-conditioned video accumulates drift. A single long pass tends to lose the references\n\nhalfway through. Splitting the render lets you re-anchor on the references between stages, which is\n\nwhat keeps the subject looking like the subject at the end of the clip.\n\nComfyUI export, LTX 2.3 models required. The cloud instance on RunningHub already has them.\n\nWorkflow JSONs and covers: [https://huggingface.co/AI-KSK/ksk-video-262-comfyui-workflows](https://huggingface.co/AI-KSK/ksk-video-262-comfyui-workflows)\n\nVideo walkthrough (Chinese): [https://youtu.be/P-D1niNdJ4w](https://youtu.be/P-D1niNdJ4w)\n\nBilibili: [https://www.bilibili.com/video/BV1xsJw6YEg6/](https://www.bilibili.com/video/BV1xsJw6YEg6/)", "url": "https://wpnews.pro/news/ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering", "canonical_source": "https://dev.to/ai-ksk/ltx-23-msr-multi-image-reference-video-with-three-stage-rendering-498", "published_at": "2026-11-14 21:48:15+00:00", "updated_at": "2026-11-14 21:51:08.492403+00:00", "lang": "en", "topics": ["generative-ai", "ai-tools", "computer-vision", "artificial-intelligence"], "entities": ["LTX 2.3", "ComfyUI", "RunningHub", "Hugging Face", "AI-KSK"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering", "markdown": "https://wpnews.pro/news/ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering.md", "text": "https://wpnews.pro/news/ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering.txt", "jsonld": "https://wpnews.pro/news/ltx-2-3-msr-multi-image-reference-video-with-three-stage-rendering.jsonld"}}