Wan 2.2 5B Fun Control ComfyUI template (video_wan2_2_5B_fun_control): control video wiring, bypassed Canny, one 20-step sampler at 24 fps and 121 frames

Wan 2.2 5B Fun Control in ComfyUI: the official template Scope. This guide reads the official video_wan2_2_5B_fun_control.json in Comfy-Org/workflow_templates at commit 0cc0fba562efe9f68e48050c2b82ed6495050bc2 (SHA256 9e671aa9...e5cc, 21,007 bytes). All 18 nodes and 19 links were traced statically. The file has no subgraphs and no switch nodes. Nothing was run, and no model, image, video or linked page was downloaded. 1. Prerequisites - A ComfyUI build with the core nodes Wan22FunControlToVideo (saved at version 0.3.56), LoadVideo, GetVideoComponents and Canny (0.3.49) and the older core nodes (0.3.45). Those are saved version strings, not a tested minimum. The graph uses no custom node pack. Note 73 says the model supports Canny, Depth, Pose and trajectory control; only Canny is in the graph, and other preprocessors would need nodes the file does not include. - Files, as declared by the loaders and the Model Links note (node 59, sizes as stated there): - models/diffusion_models/wan2.2_fun_control_5B_bf16.safetensors (9.32 GB). One checkpoint, not a high-noise and low-noise pair. - models/vae/wan2.2_vae.safetensors (1.31 GB). This is the Wan 2.2 VAE; the 14B templates use wan_2.1_vae. - models/text_encoders/umt5_xxl_fp8_e4m3fn_scaled.safetensors (6.27 GB). Together about 16.9 GB. No LoRA is used. 2. Control video and preprocessing - LoadVideo 66 (example video_wan2_2_5B_fun_control_control_video.mp4) feeds GetVideoComponents 67 (link 121). Its images output (link 122) goes to Canny 68 (saved 0.1 and 0.6, the low and high thresholds in the core node's order). Canny's output (link 127) reaches the control_video input of Wan22FunControlToVideo 60. PreviewImage 69 shows Canny's output (link 123). - As saved, Canny 68 and PreviewImage 69 are bypassed. A bypassed node passes its image input through, so the clip's frames reach control_video unprocessed. The example file name suggests it is already a control video. It was not viewed. - With ordinary footage, enable Canny 68 (and PreviewImage 69 to check the edges), or feed another preprocessor's images to control_video. - GetVideoComponents also outputs audio and fps, but neither is connected. The output fps is set separately in CreateVideo. - These nodes sit in the group "Step 3 - Control video and video preprocessing". 3. Reference image LoadImage 70 (example video_wan2_2_5B_fun_control_start_image.png) feeds ref_image on Wan22FunControlToVideo 60 (link 129). It sits in the group "Step 2 - Start_image". 4. Size, length and fps Wan22FunControlToVideo 60 holds width 704, height 704, length 121 and batch size 1 (widget names from the node's order; the file does not label them). Nothing is linked into them. CreateVideo 57 sets 24 fps, so 121 frames are about 5.04 s. Note 73 says the model was trained with 121 frames at 24 FPS, which matches the saved length and fps. 121 has the 4n + 1 form that the Wan 2.2 reference code is documented to use (see the frame-count guide). How the node handles a control clip that has more or fewer than 121 frames, or a different size, is not in this file. Prepare a 121-frame control clip at the output size and aspect ratio. 5. Prompts and sampling - Prompts: positive CLIPTextEncode 6 (a character in tattered hooded clothing running through a post-apocalyptic world, with a trailing newline) and negative 7 (the usual Wan Chinese quality list). Both use CLIPLoader 38 (umt5_xxl_fp8_e4m3fn_scaled, type wan). VAELoader 39 loads wan2.2_vae. - Model: UNETLoader 37 (wan2.2_fun_control_5B_bf16) feeds ModelSamplingSD3 48 (shift 8), then KSampler 3. - KSampler 3: seed 436961904444694 set to randomize, 20 steps, CFG 5, uni_pc, simple, denoise 1. Positive, negative and the starting latent come from Wan22FunControlToVideo 60 (links 113 to 115). Nothing is linked into steps or CFG. - Output: VAEDecode 8, then CreateVideo 57 at 24 fps, then SaveVideo 58 (prefix video/ComfyUI, format auto, codec auto). Expect one 704x704 video of 121 frames, about 5 s, following the control clip's motion with the reference image's look. Note 73 warns that, with only 5B parameters, the result most of the time will not be very good. That is an author statement, not tested here. 6. Run it 1) Put the three files in the folders above. 2) Open the template. 3) Load your control clip in LoadVideo 66. If it is not already a control video, enable Canny 68 or add another preprocessor. 4) Load your reference image in LoadImage 70. 5) Edit prompts 6 and 7, and size and length on node 60. 6) Run. Expect one video under video/ComfyUI. There is no second group to enable or bypass. 7. Failure cases and checks - Ordinary footage used as control: with Canny bypassed, raw frames go in as the control signal. Enable a preprocessor first. - Other control types: Depth, Pose and trajectory control are named in note 73, but the file contains no nodes for them. - Clip length, fps or size mismatch: see section 4. The output fps is fixed at 24, whatever the clip's own fps. - Wrong VAE or checkpoint: this template names wan2.2_vae and the 5B Fun Control checkpoint, not wan_2.1_vae or the 14B Fun Control pair. - File-list note: the folder tree in note 59 puts a stray closing parenthesis after the checkpoint file name. The loader names the file without it. - The Tutorial link in note 59 points at docs.comfy.org/tutorials/video/wan/wan2_2 and has a line break before its closing parenthesis, so it may not resolve as written. It was not fetched. - Quality: note 73 expects weak results most of the time from the 5B model. 8. How it differs from the 14B Fun Control template One checkpoint, one KSampler and no LoRA group, where the 14B template has an expert pair, two KSamplerAdvanced stages and a bypassed 4-step LoRA group. It runs 20 steps at CFG 5, where the 14B default is CFG 3.5, and uses 704x704, 121 frames, 24 fps and wan2.2_vae, where the 14B template uses 640x640, 81 frames, 16 fps and wan_2.1_vae. Both leave Canny bypassed and use the same control-video wiring pattern. Evidence and rights. Evidence status: source_inspected, not_run, unreviewed. This is shared-founder desk curation by the tutorial lane. The workflow repository declares the MIT licence; that covers this analysis of the template, not the Wan models, example media or outputs, which keep their own terms.

{
  "citations": [
    {
      "url": "https://raw.githubusercontent.com/Comfy-Org/workflow_templates/0cc0fba562efe9f68e48050c2b82ed6495050bc2/templates/video_wan2_2_5B_fun_control.json",
      "note": "Exact pinned template; node, link and group byte ranges and excerpt hashes are in the source pins."
    },
    {
      "url": "https://raw.githubusercontent.com/Comfy-Org/workflow_templates/0cc0fba562efe9f68e48050c2b82ed6495050bc2/LICENSE",
      "note": "Repository MIT licence at the same commit (existing native source)."
    }
  ]
}