image used to create example video: <a target="_blank" rel="ugc" href="https://civarchive.com/images/123232086">https://civarchive.com/images/123232086</a>WAN 2.2 14B I2V Workflow with SVI 2.0 (Long Video Generation)I couldn't find a workflow based on ComfyUI-WanVideoWrapper nodes that uses the SVI 2.0 LoRA, so I decided to create my own.This workflow includes the following optimizations:<ol><li>Block swap</li><li>TeaCache</li><li>TorchCompile</li><li>Cached prompts, text embeddings, and image embeddings</li><li>RifleXRope frames interpolation in the sampler</li><li>SageAttention</li><li>Smart image resizing to ensure the first frame resolution is divisible by 32 (better resolution for WAN 2.2)</li><li>ApplyNAG (enable negative prompt for cfg=1.0)</li></ol>Feel free to clone 'Extra segment' subgraph to increase total video length.Tested on:<ul><li>Kubuntu 25.10 (kernel 6.18)</li><li>ComfyUI 16.4</li><li>Pytorch 2.10+rocm7.1; Triton 3.6.0; SageAttention-1.0.6 (for ROCm)</li><li>AMD RX 6700XT 12 Gb</li><li>32 Gb RAM</li></ul>P.S. (not included in this workflow) If you want to double your output video FPS, you can add RIFE VFI node before VHS Video Combine node, then change frame_rate in VHS Video Combine node<h1 id="prerequisite">🚀 Prerequisite</h1><ul><li>At least 12 Gb VRAM</li><li>SageAttention installed and enabled</li><li>ROCm 7.1 / CUDA 12+ compatible system</li></ul><h1 id="model-links:">📥 Model links:</h1><h3 id="wan:">🎬 WAN:</h3><ul><li><a rel="ugc" href="https://civarchive.com/models/2456413?modelVersionId=2763096">(CivitAI) DaSiWa-WAN 2.2 I2V 14B BoundBite v10 | Lightspeed | GGUF - Q4 High</a></li><li><a rel="ugc" href="https://civarchive.com/models/2456413?modelVersionId=2763705">(CivitAI) DaSiWa-WAN 2.2 I2V 14B BoundBite v10 | Lightspeed | GGUF - Q4 Low</a></li></ul>You can try Q5 or even Q6 versions<h3 id="wan-vae:">🖼️ WAN Vae:</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/Wan_2.2_ComfyUI_Repackaged/resolve/main/split_files/vae/wan_2.1_vae.safetensors">(Hugging Face) Comfy-Org/Wan_2.2_ComfyUI_Repackaged - wan_2.1_vae.safetensors</a></li></ul><h3 id="t5-text-encoder">📝 T5 Text encoder</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/umt5-xxl-enc-fp8_e4m3fn.safetensors?download=true">(Hugging Face) Kijai/WanVideo_comfy - umt5-xxl-enc-fp8_e4m3fn.safetensors</a></li></ul>You can use any T5 encoder, but don't use 'scaled' versions<h3 id="tae-for-wan-2.1-(for-previews-optional)">👁️ Tae for WAN 2.1 (for previews, optional)</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/blob/main/taew2_1.safetensors">(Hugging Face) Kijai/WanVideo_comfy - taew2_1.safetensors</a></li></ul><h3 id="svi-2.0-pro-lora">♾️ SVI 2.0 Pro LoRA</h3><ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/LoRAs/Stable-Video-Infinity/v2.0/SVI_v2_PRO_Wan2.2-I2V-A14B_HIGH_lora_rank_128_fp16.safetensors?download=true">(Hugging Face) Kijai/WanVideo_comfy - SVI_v2_PRO_Wan2.2-I2V-A14B_HIGH_lora_rank_128_fp16.safetensors</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/WanVideo_comfy/resolve/main/LoRAs/Stable-Video-Infinity/v2.0/SVI_v2_PRO_Wan2.2-I2V-A14B_LOW_lora_rank_128_fp16.safetensors?download=true">(Hugging Face) Kijai/WanVideo_comfy - SVI_v2_PRO_Wan2.2-I2V-A14B_LOW_lora_rank_128_fp16.safetensors</a></li></ul><h1 id="folder-structure">📁 Folder structure</h1><pre><code>/path_to_ComfyUI/
├── models/
│ ├── unet/
│ │ ├── DasiwaWAN22I2V14BSynthseduction_q4High.gguf
│ │ └── DasiwaWAN22I2V14BSynthseduction_q4Low.gguf
│ ├── text_encoders/
│ │ └── umt5-xxl-enc-fp8_e4m3fn.safetensors
│ ├── vae/
│ │ └── wan_2.1_vae.safetensors
│ ├── vae_approx/
│ │ └── taew2_1.safetensors
│ └── loras/
│ ├── SVI_v2_PRO_Wan2.2-I2V-A14B_HIGH_lora_rank_128_fp16.safetensors
│ └── SVI_v2_PRO_Wan2.2-I2V-A14B_LOW_lora_rank_128_fp16.safetensors</code></pre><h1 id="used-nodes">📦 Used nodes</h1><ul><li><a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-WanVideoWrapper">ComfyUI-WanVideoWrapper</a></li><li><a target="_blank" rel="ugc" href="https://github.com/kijai/ComfyUI-KJNodes">ComfyUI-KJNodes</a></li><li><a target="_blank" rel="ugc" href="https://github.com/yolain/ComfyUI-Easy-Use">ComfyUI-Easy-Use</a></li><li><a target="_blank" rel="ugc" href="https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite">ComfyUI-VideoHelperSuite</a></li><li><a target="_blank" rel="ugc" href="https://github.com/Smirnov75/ComfyUI-mxToolkit">ComfyUI-mxToolkit</a></li><li><a target="_blank" rel="ugc" href="https://github.com/crystian/ComfyUI-Crystools">ComfyUI-Crystools</a></li><li><a target="_blank" rel="ugc" href="https://github.com/evanspearman/ComfyMath">ComfyMath</a></li><li><a target="_blank" rel="ugc" href="https://github.com/rgthree/rgthree-comfy">rgthree-comfy</a></li></ul><h1 id="performance">🚀 Performance</h1>For an image 480x608@49 (16x3+1) on AMD RX 6700XT 12 GB (AOTRITON+SageAttn+TunableOps)<ul><li>First run: ~30-35 min (compilation + tuning) first segment</li><li>Subsequent runs: ~5-6 min per segment</li></ul><h1 id="known-issues-and-fixes">⚠️ Known Issues &amp; Fixes</h1><ul><li>OOM: Increase Swap block parameter, decrease Image resolution, FPS, Seconds</li><li>System freeze: Disable <code>fun_or_fl2v_model</code> (if enabled); Change linux kernel to 6.18 or higher</li><li>First run slow: Normal, compilation takes time for every new parameter set (resolution, frames, etc), especially when TUNABLE_OPS enabled.</li></ul>

Files with SHA256: ab271393cdb344c92b658c8f9c183223d4a6d725ce991892059344967455c05f