<h3 id="2.2">2.2</h3>For I2V this motion helper node is extremely useful:<a target="_blank" rel="ugc" href="https://github.com/princepainter/ComfyUI-PainterI2V">https://github.com/princepainter/ComfyUI-PainterI2V</a>10/30 High lora was further refined.New I2V 1022 versions are out. They have by far the best prompt following / motion quality yet. (The lora key warning is fine, it just contains extra modulation keys that comfyui does not use. It does not matter.)<a target="_blank" rel="ugc" href="https://github.com/VraethrDalkr/ComfyUI-TripleKSampler">https://github.com/VraethrDalkr/ComfyUI-TripleKSampler</a>T2V versions just got updated 09/28. Probably still best to use a step or two with cfg / without the lora to establish motion with high noise as usual like:2 steps high noise without the low-step lora at 3.5 CFG2 steps high noise with lora and 1 CFG2-4 steps low noise with lora and 1 CFGIts definitely a big improvement either way.T2V:Using their full 'dyno' model for your high model seems best."On Sep 28, 2025, we released two models, Wan2.2-T2V-A14B-4steps-lora-250928 and Wan2.2-T2V-A14B-4steps-250928-dyno. The two models share the same low-noise weight. Wan2.2-T2V-A14B-4steps-250928-dyno delivers superior motion rendering and camera response, with object movement speeds that closely match those of the base model. For projects requiring highly dynamic visuals, we strongly recommend using Wan2.2-T2V-A14B-4steps-250928-dyno. Below, you will find some showcases for reference."<a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Wan2.2-Lightning/blob/main/Wan2.2-T2V-A14B-4steps-250928-dyno/Wan2.2-T2V-A14B-4steps-250928-dyno-high-lightx2v.safetensors">https://huggingface.co/lightx2v/Wan2.2-Lightning/blob/main/Wan2.2-T2V-A14B-4steps-250928-dyno/Wan2.2-T2V-A14B-4steps-250928-dyno-high-lightx2v.safetensors</a><hr /><h3 id="2.1">2.1</h3>7/15 update: I added the new I2V lora, seems to have much better motion than using the old text to video lora on a image to video model. Example is 4 steps, 1 CFG, LCM sampler, 8 shift. I uploaded the new version of the T2V one also.I'm also putting up the rank 128 versions extracted by Kijai, they are double the size but are slightly better quality.I suggest using it with the Pusa V1 lora as well, it seems to improve movement even more: <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/tree/main/Pusa">https://huggingface.co/Kijai/WanVideo_comfy/tree/main/Pusa</a> No need for 2 sampler WF anymore IMO. Just plug it into your normal WF with 1 CFG and 4 steps or so. Could prob sharpen it with another pass if you wanted but it no longer hurts the movement like before for image to video.Full Image to video Lightx2V model: <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v/tree/main/distill_models">https://huggingface.co/lightx2v/Wan2.1-I2V-14B-480P-StepDistill-CfgDistill-Lightx2v/tree/main/distill_models</a><hr />Old: <a target="_blank" rel="ugc" href="https://huggingface.co/lightx2v/Wan2.1-T2V-14B-StepDistill-CfgDistill">lightx2v</a> made a 14B self forcing model that is a massive improvement compared to Causvid / Accvid. Kijai extracted it as a <a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_T2V_14B_lightx2v_cfg_step_distill_lora_rank32.safetensors">lora</a>. Example above was generated in about 35 seconds on a 4090 using 4 steps, lcm, 1 cfg, 8 shift, still playing with settings to see what is best.<hr />Please don't send me buzz or anything, give the <a target="_blank" rel="ugc" href="https://github.com/ModelTC/lightx2v">lightx2v</a> team or <a target="_blank" rel="ugc" href="https://www.myminifactory.com/users/kijai">kijai</a> support if anyone.

Lightning Lora, massive speed up for Wan2.1 / Wan2.2 made by Lightx2v / Kijai

Wan21_CausVid_14B_T2V_lora_rank32_v2.safetensors

Wan21_CausVid_14B_T2V_lora_rank32.safetensors

Self-Forcing / CausVid / Accvid Lora, massive speed up for Wan2.1 made by Kijai

Causvid 14B V2

2.2-Lightning-I2V-1030-H

2.2-Lightning-I2V-1022-L

2.2-Lightning-I2V-1022-H

2.2 Lightning I2V H 2?

2.2-Lightning-T2V-0928-L

2.2-Lightning-T2V-0928-H

2.2 Lightning I2V L

2.2 Lightning I2V H

2.2 Lightning T2V L

2.2 Lightning T2V H

Self-Forcing 14B T2V R128

Self-Forcing 14B I2V R128

Self-Forcing 14B T2V

Self-Forcing 14B I2V

Self-Forcing 14B T2V V1

Accvid 14B I2V

Accvid 14B

Causvid 1.3B

<h1 id="wan2.1-vace-14b-(lora-accelerated):-10x-speed-with-causvid-lora-for-3-step-video-creation-y8p4wlr3f">👑 Wan2.1-VACE-14B (LoRA Accelerated): 10x Speed with CausVid LoRA for 3-Step Video Creation</h1><h3 id="skyrocket-your-video-creation:-achieve-~10x-speed-with-wan2.1-and-the-causvid-lora!-sfh0gcbpe">🎬 Skyrocket Your Video Creation: Achieve ~10x Speed with Wan2.1 &amp; the CausVid LoRA! 🎬</h3>the pro Workflow &gt; <a target="_blank" rel="ugc" href="https://civarchive.com/models/1620800">https://civarchive.com/models/1620800</a>II. ℹ️ Important Note / Alert ℹ️ Note: The CausVid LoRA (<code>Wan21_CausVid_14B_T2V_lora_rank32.safetensors</code>) is key to this workflow! It enables a significant speedup, reducing generation time for 81 frames at 720p from approximately 40 minutes to just around 4 minutes on an RTX 4090 (or even faster with quantized models like Q3KL GGUF). This guide focuses on leveraging this LoRA for fast video generation.III. 📌 Overview / Introduction The Wan2.1-VACE-14B video diffusion model, when supercharged by the CausVid LoRA, is designed for high-quality, highly efficient video generation. It particularly excels at 480p and 720p resolutions through a streamlined 3-step ComfyUI workflow. This guide will walk you through the setup process to unlock this accelerated video generation capability, including options for full precision and quantized models like the fast Q3KL GGUF.IV. 🔑 Key Components / Dependencies<ol><li>Diffusion Model (14B):<ul><li>Full Precision: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/diffusion_models/wan2.1_vace_14B_fp16.safetensors">wan2.1_vace_14B_fp16.safetensors</a> (Recommended for compatibility with LoRA examples)</li><li>Quantized (Civitai): <a target="_blank" rel="ugc" href="https://civarchive.com/models/1616692?modelVersionId=1829648">wan2.1_vace_14B_Q4KM.safetensors</a></li><li>Quantized (GGUF - Civitai): <a target="_blank" rel="ugc" href="https://civarchive.com/models/1616692?modelVersionId=1829937">wan2.1_vace_14B_Q3kl.gguf</a> (Used in the 5-min example, requires GGUF loader)</li></ul></li><li>Performance LoRA (Essential for Speed):<ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/blob/main/Wan21_CausVid_14B_T2V_lora_rank32.safetensors">Wan21_CausVid_14B_T2V_lora_rank32.safetensors (original)</a></li></ul></li><li>VAE:<ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/vae/wan_2.1_vae.safetensors?download=true">wan_2.1_vae.safetensors</a></li></ul></li><li>Text Encoder: Choose one:<ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/text_encoders/umt5_xxl_fp16.safetensors?download=true">umt5_xxl_fp16.safetensors</a> (Recommended to match Kijai's wrapper compatibility for the LoRA workflow)</li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/resolve/main/split_files/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors?download=true">umt5_xxl_fp8_e4m3fn_scaled.safetensors</a> (Smaller, fp8 version)</li></ul></li></ol>V. 📁 File Organization / Installation Place the downloaded files in the following structure within your ComfyUI directory:<pre><code>ComfyUI/
├── models/
│ ├── diffusion_models/
│ │ └── wan2.1_vace_14B_fp16.safetensors # Or Q4KM.safetensors, or Q3kl.gguf
│ ├── text_encoders/
│ │ └── umt5_xxl_fp16.safetensors # Or the fp8 version
│ ├── loras/
│ │ └── Wan21_CausVid_14B_T2V_lora_rank32.safetensors
│ └── vae/
│ └── wan_2.1_vae.safetensors
</code></pre>VI. 🎨 Model Showcase: Rapid 720p Cinematic Shots This setup, featuring Wan2.1-VACE-14B and the CausVid LoRA, excels at producing 720p (and 480p) video clips with remarkable speed—even faster with quantized GGUF models. It's ideal for quick iterations, creative experimentation, and efficient content creation, all streamlined by a 3-step workflow.VII. 🛠️ Step-by-Step Tutorial / Workflow The core of this accelerated workflow relies on integrating the CausVid LoRA.<ul><li>Context for the LoRA and its benefits can be found in <a target="_blank" rel="ugc" href="https://www.reddit.com/r/StableDiffusion/comments/1knuafk/causvid_lora_massive_speedup_for_wan21_made_by/">Kijai's Reddit Post announcing the LoRA</a>. The workflow is designed to generate videos in approximately 3 main steps within ComfyUI after initial model and LoRA loading. 💡 A ComfyUI workflow JSON demonstrating this 3-step process (especially one for the Q3KL GGUF version if available) would be invaluable. If you have one, consider linking it here: <code>[LINK_TO_YOUR_COMFYUI_WORKFLOW.json]</code></li></ul>VIII. 💡 Usage Tips / Best Practices / Performance<ul><li>Model &amp; LoRA Configuration: For maximum speed and quality, ensure you are using the appropriate 14B model (e.g., <code>wan2.1_vace_14B_fp16.safetensors</code> or <code>wan2.1_vace_14B_Q3kl.gguf</code>) paired with the <code>Wan21_CausVid_14B_T2V_lora_rank32.safetensors</code> LoRA.</li><li>Text Encoder: The <code>umt5_xxl_fp16.safetensors</code> text encoder is recommended for best compatibility with existing examples and Kijai's original demonstrations. The fp8 version can save VRAM.</li><li>Resolution: This setup is optimized for 480p and 720p video generation.</li><li>Performance Gains:<ul><li>Without LoRA (fp16): An 81-frame 720p video might take ~40 minutes on an RTX 4090.</li><li>With CausVid LoRA (fp16): The same video can be generated in ~4 minutes on an RTX 4090.</li><li>With CausVid LoRA &amp; Q3KL GGUF: Potentially even faster, around 5 minutes or less for similar output on capable hardware with a GGUF loader.</li></ul></li><li></li></ul><h2 id="credits-and-acknowledgements-w8b8flj08">🌟 Credits &amp; Acknowledgements</h2>Original Wan 2.1 models repackaged for ComfyUI by Comfy-Org: <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged">Wan 2.1 ComfyUI Repackaged on Hugging Face</a>. The performance-boosting CausVid LoRA (<code>Wan21_CausVid_14B_T2V_lora_rank32.safetensors</code>) was extracted and shared by Kijai. Original announcement and details: <a target="_blank" rel="ugc" href="https://www.reddit.com/r/StableDiffusion/comments/1knuafk/causvid_lora_massive_speedup_for_wan21_made_by/">Kijai's Reddit Post</a>. Quantized GGUF and Safetensors versions available on Civitai, enabling broader accessibility and speed. Gratitude to the developers of the underlying CausVid techniqueMy Other ComfyUI Guides(in workflow):<ul><li>📽️ ComfyUI VFI: Video Duration Extension Guide</li><li>Guide for 🧠 ComfyUI-OllamaGemini Custom Node Setup</li><li>🎞️ ComfyUI LayerStyle for Video: Full Guide</li></ul><h2 id="developer-information-an4xir22z">👨‍💻 Developer Information</h2>This guide was created by Abdallah Al-Swaiti:<ol><li><a target="_blank" rel="ugc" href="https://huggingface.co/ABDALLALSWAITI">Hugging Face</a></li><li><a target="_blank" rel="ugc" href="https://github.com/al-swaiti">GitHub</a></li><li><a target="_blank" rel="ugc" href="https://www.linkedin.com/in/abdallah-issac/">LinkedIn</a></li><li><a target="_blank" rel="ugc" href="https://github.com/al-swaiti/ComfyUI-OllamaGemini">ComfyUI-OllamaGemini</a></li></ol>For additional tools and updates, check out my other repositories.✨ Create Dreamy Videos with WAN 2.1 VACE and Pastel Dream! ✨

Wan21_CausVid_14B_T2V_lora_rank32

Wan21_CausVid_14B_T2V_lora_rank32 (1).safetensors

v1.0

causvid

Files with SHA256: e5664f709f39b8352e8487e412efdac5c2fc283f21d06e289a19344daaaa198d