<h1 id="ltx-2">LTX-2</h1>This model card focuses on the LTX-2 model, as presented in the paper <a target="_blank" rel="ugc" href="https://huggingface.co/papers/2601.03233">LTX-2: Efficient Joint Audio-Visual Foundation Model</a>. The codebase is available <a target="_blank" rel="ugc" href="https://github.com/Lightricks/LTX-2">here</a>.LTX-2 is a DiT-based audio-video foundation model designed to generate synchronized video and audio within a single model. It brings together the core building blocks of modern video generation, with open weights and a focus on practical, local execution.<a target="_blank" rel="ugc" href="https://www.youtube.com/watch?v=8fWAJXZJbRA"><img src="https://img.youtube.com/vi/8fWAJXZJbRA/maxresdefault.jpg" alt="LTX-2 Open Source" /></a><h2 id="direct-use-license">Direct use license</h2>You can use the models - full, distilled, upscalers and any derivatives of the models - for purposes under the <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/blob/main/LICENSE">license</a>.<h2 id="comfyui">ComfyUI</h2>We recommend you use the built-in LTXVideo nodes that can be found in the ComfyUI Manager. For manual installation information, please refer to our <a target="_blank" rel="ugc" href="https://docs.ltx.video/open-source-model/integration-tools/comfy-ui">documentation site</a>.<h2 id="general-tips:">General tips:</h2><ul><li>Width &amp; height settings must be divisible by 32. Frame count must be divisible by 8 + 1.</li><li>In case the resolution or number of frames are not divisible by 32 or 8 + 1, the input should be padded with -1 and then cropped to the desired resolution and number of frames.</li><li>For tips on writing effective prompts, please visit our <a target="_blank" rel="ugc" href="https://ltx.video/blog/how-to-prompt-for-ltx-2">Prompting guide</a></li></ul><h3 id="limitations">Limitations</h3><ul><li>This model is not intended or able to provide factual information.</li><li>As a statistical model this checkpoint might amplify existing societal biases.</li><li>The model may fail to generate videos that matches the prompts perfectly.</li><li>Prompt following is heavily influenced by the prompting-style.</li><li>The model may generate content that is inappropriate or offensive.</li><li>When generating audio without speech, the audio may be of lower quality.</li></ul><h2 id="huggingface-repository">HuggingFace Repository</h2><a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/tree/main">https://huggingface.co/Lightricks/LTX-2/tree/main</a>

Lightricks LTXV 2

An x2 spatial upscaler for the ltx-2 latents, used in multi stage (multiscale) pipelines for higher resolution

lightricksLTXV2_ltx2SpatialUpscaler.safetensors

ltx-2-spatial-upscaler-x2

ltx-2-19b-distilled-fp8

ltx-2-temporal-upscaler-x

ltx-2-19b-dev-fp4

ltx-2-19b-dev-fp8

ltx-2-19b-dev

ALL QUESTIONS WILL BE ANSWERED IN THE COMMENT :)Cause there is a lot to say about how the model and workflows works... I will try to answer and help you if you get some problems :) (Upload Still in progress...)<hr /><h1 id="upload-in-progress....coming-soon-:">Upload In Progress.... COMING SOON :</h1><ul><li><s>LTX+2 Q6_K.GGUF</s></li><li><s>LTX2 Audio VAE BF16</s></li><li><s>TEXT ENCODER GEMMA 3 12B</s></li><li><s>LTX-2-1B-Embeddings_connector_distill_BF16</s></li><li>Special GGUF EDITED WORFLOW IMG2IMG / TXT2IMG</li><li><s>LORA DISTILLED VERSION</s></li><li><s>SPATIAL UPSCALER</s></li><li><s>TEMPORAL UPSCALER</s></li><li><s>CAMERA CONTROL LORAS.</s></li><li>CONTROLNET AIO LTX2</li><li>Workflows <s>I2V</s> / V2V / <s>T2V </s>/ VDETAILER.</li></ul>(Keep in mind i'm not the owner of this model) I made this upload from the official Public LTX Huggingface repo for Civitai. <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks">https://huggingface.co/Lightricks</a> I just want to centralize the most useful models on this Civitai page, so everyone doesn’t get lost among the many models available on the original repo. I’m not asking for any “buzz” or anything like that, it’s just to help the community have a solid LTX base on Civitai (for people not familiar with Hugging Face). I won’t upload big models, I leave that to people who actually want to use them, even if the benefits of doing so will be really minimal. But I will also centralize here my own releases and base-model variations of unofficial LTX.<hr /><h2 id="ltx-2-fp8-distilled-(fast-and-lightweight)">⚡ LTX-2 FP8 — Distilled (Fast &amp; Lightweight)</h2><h3 id="what-is-ltx-2-fp8-distilled">What is LTX-2 FP8 Distilled?</h3>The FP8 Distilled version is a compressed and accelerated variant of LTX-2, trained to replicate the behavior of the full model while being faster and lighter.Distillation reduces model complexity, making it more efficient — at the cost of some fine-grained detail.<h3 id="key-characteristics">✅ Key Characteristics</h3><ul><li>Faster generation speed</li><li>Lower VRAM requirements</li><li>Quicker prompt response</li><li>Slightly reduced fine detail compared to full FP8</li><li>Excellent quality-to-performance ratio</li></ul><h3 id="best-use-cases">🎯 Best Use Cases</h3><ul><li>Rapid iteration &amp; testing</li><li>Prompt exploration</li><li>Draft videos and previews</li><li>Creators with limited hardware</li></ul>Recommended if: You want speed and accessibility, and are willing to trade a small amount of detail for faster results.<hr /><h2 id="ltx-2-fp8-standard-(full-quality)">🔹 LTX-2 FP8 — Standard (Full Quality)</h2><h3 id="what-is-ltx-2-fp8-(standard)">What is LTX-2 FP8 (Standard)?</h3>The FP8 Standard version is a full-quality LTX-2 model quantized to FP8 precision. It preserves the complete architecture and capabilities of the original model while reducing memory usage.This is NOT a simplified model. Only the numerical precision is reduced — the model’s intelligence, structure, and behavior remain intact.<h3 id="key-characteristics">✅ Key Characteristics</h3><ul><li>High visual fidelity and detail</li><li>Strong temporal consistency</li><li>Full audio-video synchronization</li><li>Lower VRAM usage than FP16</li><li>Stable and reliable for long generations</li></ul><h3 id="best-use-cases">🎯 Best Use Cases</h3><ul><li>Cinematic video generation</li><li>Final renders and high-quality outputs</li><li>Creators who want maximum quality with lower hardware requirements</li></ul>Recommended if: You want the best possible quality in FP8, with no compromise on features or flexibility.<hr /><h2 id="which-one-should-you-choose">🧠 Which One Should You Choose?</h2><ul><li>🎬 Go with FP8 Standard if quality and consistency matter most</li><li>⚡ Go with FP8 Distilled if speed and efficiency are your priority</li></ul>Both versions are fully compatible with ComfyUI workflows and part of the same LTX-2 creative ecosystem.<hr /><h2 id="what-is-ltx-2">📌 What is LTX-2?</h2>LTX-2 is a powerful multimodal AI model that transforms text prompts, images, or other media into fully synchronized audiovisual videos — with motion, dialogue, music, and ambient sound generated in one unified pass. It’s built on a hybrid Diffusion-Transformer (DiT) architecture designed specifically for efficient spatiotemporal generation and audio-video alignment. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2+1</a>This approach lets creators go from idea to cinematic result without stitching separate audio tracks manually — a major step beyond typical text-to-video systems. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a><hr /><h2 id="key-features-and-capabilities">✨ Key Features &amp; Capabilities</h2><h3 id="cinematic-quality-output">🎥 Cinematic Quality Output</h3><ul><li>Native 4K resolution support with playback up to 50 FPS, delivering smooth, high-detail video clips ideal for cinematic, commercial, or creative use. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a></li></ul><h3 id="unified-audio-and-visual-generation">🎵 Unified Audio &amp; Visual Generation</h3><ul><li>Generates synchronized audio — including dialogue, ambience and music — alongside the video in a single generation pass, removing the need for external audio sync tools. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a></li></ul><h3 id="flexible-input-and-output-modes">🔄 Flexible Input &amp; Output Modes</h3><ul><li>Works with text prompts, image references, multi-keyframe conditioning, and more to animate concepts or stills into motion. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a></li></ul><h3 id="performance-modes">⚙️ Performance Modes</h3><ul><li>Multiple performance configurations (Fast, Pro, Ultra) allow creators to balance speed and quality according to project needs — from quick drafts to production-ready renders. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a></li></ul><h3 id="efficient-and-accessible">🧠 Efficient &amp; Accessible</h3><ul><li>Highly optimized for consumer-grade GPUs — efficient enough to run on ~16 GB VRAM hardware with FP8/FP4 quantization options — making AI video production more accessible. <a target="_blank" rel="ugc" href="https://www.reddit.com//r/StableDiffusion/comments/1q5jgnl/ltx2_runs_on_a_16gb_gpu/?utm_source=chatgpt.com">Reddit</a></li></ul><h3 id="open-and-extensible">🛠️ Open &amp; Extensible</h3><ul><li>Fully open weights, codebase, and workflows, enabling fine-tuning, custom LoRAs, and integration into tools like ComfyUI. <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/?utm_source=chatgpt.com">Hugging Face</a></li></ul><hr /><h2 id="improvements-over-earlier-versions">📈 Improvements Over Earlier Versions</h2>Compared to the original LTX family and other open video models, LTX-2 raises the bar in several key areas:✅ Audio Integration Built-In Instead of generating silent videos and requiring post-processing, LTX-2 outputs audio and visual streams together with temporal coherence. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a>✅ Higher Resolution &amp; Frame Rates Supports native 4K at up to 50 frames per second, reaching cinema-grade quality unlike many earlier community models that cap at lower resolutions or fps. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a>✅ Longer Clips Offers extended duration generation (up to ~20 s clips) with continuous quality and audio coherence — exceeding many alternatives. <a target="_blank" rel="ugc" href="https://ltx-2.ai/?utm_source=chatgpt.com">LTX-2+1</a>✅ Expanded Workflows Native support in ComfyUI plus custom workflows empowers users with text-to-video, image-to-video, multi-keyframe conditioning, and creative control nodes. <a target="_blank" rel="ugc" href="http://comfyui.org">comfyui.org</a><a target="_blank" rel="ugc" href="https://comfyui.org/en/ltx-2-now-available-in-comfyui?utm_source=chatgpt.com">+1</a><hr /><h2 id="typical-use-cases">🧠 Typical Use Cases</h2>🔹 Cinematic storyboarding &amp; concept visuals 🔹 Social media &amp; marketing video content 🔹 Animated storytelling &amp; motion design 🔹 Game cutscenes &amp; immersive narratives 🔹 Product visualizations &amp; dynamic adsWhether for rapid prototyping or production output, LTX-2 empowers creators with professional-grade generative video. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a><hr /><h2 id="included-files-and-variants">🧩 Included Files &amp; Variants</h2>Depending on the checkpoint uploaded, this collection may include:<ul><li>Full Model Checkpoints (bf16 / fp8 / fp4) — maximum quality with quantization options</li><li>Distilled Variants — faster iteration with lighter compute cost</li><li>Spatial &amp; Temporal Upscalers — improve resolution or frame rate via multiscale pipelines</li><li>LoRA &amp; Fine-Tuning Packs — custom stylistic or control extension modules <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/?utm_source=chatgpt.com">Hugging Face</a></li></ul><hr /><h2 id="comfyui-integration-and-workflows">🔧 ComfyUI Integration &amp; Workflows</h2>Included workflow templates help you use LTX-2 in ComfyUI with nodes for:📌 Text-to-Video — generate animated clips from prompts 📌 Image-to-Video — animate still images with camera motion and style 📌 Video Conditioning — extend clips forward/backward or refine motions 📌 Keyframe Controls — precise guidance over scene transitionsThese workflows are designed for ease-of-use and creative flexibility while demonstrating best practices for prompt structure and smooth temporal motion. <a target="_blank" rel="ugc" href="https://docs.ltx.video/open-source-model/integration-tools/comfy-ui?utm_source=chatgpt.com">LTX Documentation</a><hr /><h2 id="foundation-model-philosophy">🧠 Foundation Model Philosophy</h2>LTX-2 goes beyond a single task — it’s a foundation model for audiovisual creative AI. Open access to its weights, code, and tools encourages developers, artists, researchers, and hobbyists alike to customize, extend and innovate on a common platform. <a target="_blank" rel="ugc" href="https://huggingface.co/Lightricks/LTX-2/?utm_source=chatgpt.com">Hugging Face</a><hr /><h2 id="summary">📌 Summary</h2>LTX-2 is not just another video model — it is a production-ready, synchronized audio-video foundation model that pushes the boundaries of what open discourse video generation can achieve. With cinematic output quality, flexible workflows, and a fully open ecosystem, LTX-2 stands as one of the most capable generative video tools available today. <a target="_blank" rel="ugc" href="https://ltx-2.io/?utm_source=chatgpt.com">LTX-2</a>

Files with SHA256: 3160fabf8edf0bc4dd8de40353a180813b111ce586b655ad54af9a7b8c6736de