<p><strong>Create controlled reference-image-to-video clips with MiniMax H3 Fun ControlNet, using depth, pose, and optional SolAttn branches.</strong></p><p><strong>Who it's for:</strong> creators who want this pipeline in ComfyUI without assembling nodes from scratch. <strong>Not for:</strong> one-click results with zero tuning - you still choose inputs, prompts, and settings.</p><h3 id="open-preloaded-workflow-on-runcomfy">Open preloaded workflow on RunComfy</h3><p><a target="_blank" rel="ugc" href="https://www.runcomfy.com/comfyui-workflows/minimax-h3-fun-control-in-comfyui-depth-and-pose-video?utm_source=civitai&amp;utm_medium=referral">Open preloaded workflow on RunComfy (browser)</a></p><p><strong>Why RunComfy first</strong><br />- <strong>Fewer missing-node surprises</strong> - run the graph in a managed environment before you mirror it locally.<br />- <strong>Quick GPU tryout</strong> - useful if your local VRAM or install time is the bottleneck.<br />- <strong>Matches the published JSON</strong> - the zip follows the same runnable workflow you can open on RunComfy.</p><p><strong>When downloading for local ComfyUI makes sense</strong> - you want full control over models on disk, batch scripting, or offline runs.</p><p><strong>How to use (local ComfyUI)</strong><br />1. Load inputs (images/video/audio) in the marked loader nodes.<br />2. Set prompts, resolution, and seeds; start with a short test run.<br />3. Export from the Save / Write nodes shown in the graph.</p><p><strong>Expectations</strong> - First run may pull <strong>large weights</strong>; cloud runs may require a <strong>free RunComfy account</strong>.</p><hr /><h3 id="overview">Overview</h3><p>Turn a reference image into a controlled video. Keep your subject's identity, outfit, and scene consistent. Guide motion with H3 Fun ControlNet. Choose single-control or depth-plus-pose branches. Use optional SolAttn for faster runs. Create clean walks, turns, waves, and full-body movement.</p><p><strong>Important nodes:</strong></p><h2 id="key-nodes-in-the-workflow">Key nodes in the workflow</h2><h3 id="apply-h3-fun-controlnet">Apply H3 Fun ControlNet</h3><p>This node adds a control stream to the current model. It receives control frames extracted from a control video and applies them during the selected sampling window.</p><p>Important settings:</p><ul><li><p><code>strength</code> controls how strongly the model follows the control.</p></li><li><p><code>start_percent</code> controls when the control begins.</p></li><li><p><code>end_percent</code> controls when the control fades out.</p></li><li><p>For single-control runs, start around <code>0.6</code> to <code>1.0</code>.</p></li><li><p>For chained controls, keep the combined strength near <code>1.0</code> or lower.</p></li></ul><p>Ending control around <code>0.6</code> is often useful because it locks motion and structure early while allowing later steps to recover texture and realism.</p><h3 id="minimaxh3imagetovideo">MiniMaxH3ImageToVideo</h3><p>This node builds image-to-video conditioning from a text prompt and optional first frame. Use it when a single first-frame reference is enough and you want the simpler Branch 1 workflow.</p><h3 id="minimaxh3referencetovideo">MiniMaxH3ReferenceToVideo</h3><p>This node provides persistent reference-image conditioning. Use it when the output needs to preserve a character, outfit, face, style, or product identity across the whole video.</p><p><code>ref_image_size: match</code> is faster. <code>ref_image_size: max</code> uses larger reference tokens and can improve identity, but costs more compute.</p><h3 id="cfgguider">CFGGuider</h3><p>This node makes negative conditioning effective. Branch 1 uses a basic guider, so negative text is not strongly applied there. Branches 2 and 3 use <code>CFGGuider</code>, making them better choices when you need to suppress artifacts, unwanted limbs, text, watermarks, or style drift.</p><h3 id="sigma-shift">Sigma shift</h3><p>The sigma shift node adjusts the video and audio diffusion schedules before sampling. Keep it in the chain unless you are intentionally experimenting with H3’s schedule behavior.</p><h3 id="solattnpatch">SolAttnPatch</h3><p>This optional node accelerates the reference branch through sparse attention. Leave <code>morton</code> off, keep conservative exact-block settings when quality matters, and compare output with a fixed seed before deciding whether the speedup is acceptable for your content.</p><h3 id="basicscheduler">BasicScheduler</h3><p>…</p><h3 id="notes">Notes</h3><p><strong>MiniMax H3 Fun Control in ComfyUI | Depth and Pose Video</strong> - see <a target="_blank" rel="ugc" href="https://www.runcomfy.com/comfyui-workflows/minimax-h3-fun-control-in-comfyui-depth-and-pose-video?utm_source=civitai&amp;utm_medium=referral">RunComfy page</a> for the latest node requirements.</p>

MiniMax H3 Fun Control in ComfyUI | Depth and Pose Video

<p>Initial release - MiniMax-H3-Control.</p>

Files with SHA256: 48f398cc0ab86d8d04cdda22f95f5cbf48c3c237631a1f1affd51aafe163406c