<h2 id="support">Support</h2>Everything here is free and stays free — the format spec, the nodes, the workflows, the cartridges, the LoRAs. If it saved you a night of debugging (it contains several hundred of mine), tips keep the 5090 warm:<ul><li>☕ <a target="_blank" rel="ugc" href="https://ko-fi.com/joeygambino">Buy me a coffee on Ko-fi</a></li><li>💛 <a target="_blank" rel="ugc" href="https://github.com/sponsors/jlucasmcrell">Sponsor on GitHub</a></li><li>🔁 <a target="_blank" rel="ugc" href="https://liberapay.com/joeygambino">Liberapay</a> (recurring)</li><li>⚡ Or right here: the Civitai tip button on this page sends Buzz directly.</li></ul><h2 id="type-a-story.-get-one-continuous-video-with-sound.-multi-shot-scenes-render-as-a-single-take-no-last-frame-chaining-no-quality-loss-from-shot-to-shot.-that's-the-whole-pitch.">Type a story. Get one continuous video, with sound. Multi-shot scenes render as a single take - no last-frame chaining, no quality loss from shot to shot. That's the whole pitch.</h2>Both workflows now read left to right: numbered lanes, and you only ever touch lanes 2-4. Everything below the main row is optional.<h2 id="what-you-need">What you need</h2><ul><li>ComfyUI + this node pack (Manager: MiniMax-H3 Multishot, or the zip on this version).</li><li>A MiniMax-H3 checkpoint (links on this page). 24 GB card? Take a GGUF.</li></ul><h2 id="seamless-chain-multi-shot-scenes-as-one-take">SEAMLESS CHAIN - multi-shot scenes as one take</h2>Lane by lane:<ul><li>README - the quick start lives on the canvas itself.</li><li>1 - MODELS - pick your H3 checkpoint and text encoder. VAEs are preset; LoRA slots are empty until you fill one.</li><li>2 - ANCHORS (optional) - a photo to open shot 1 on (enable its gate), and a short voice clip to lock the speaker's voice.</li><li>3 - YOUR PROMPTS - type your idea in the box, or point the switch at a prompt file. The writer expands it into shot prompts. Writing your own? Set the writer to <code>passthrough (raw JSON, skip LLM)</code> and paste shots separated by <code>---</code> lines.</li><li>4 - CONTROLS - size, frames per shot, steps, and <code>take_seconds</code> (total length; 30 is a good first run). The switches stay off unless you installed the pack a switch names.</li><li>5 - ENGINE - nothing to change. The remote encoder lives here if you want the text encoder on a second PC: enter its address, flip the encoder switch, free ~15 GB.</li><li>6 - OUTPUT - your video and its audio save here.</li></ul>Optional panels below the main row: reference images (your character, ref2va checkpoints - folder per character + AUTO REFS on), V2V reference (a clip whose look guides the render), FFLF plates (flf_chain mode only), audio spine (a soundtrack the take follows).<h2 id="extend-take-one-person-talking-as-long-as-you-want">EXTEND TAKE - one person talking, as long as you want</h2>Same lanes, different job: one premise becomes ONE continuous speech cut across windows.<ul><li>1 - MODELS - same as above.</li><li>2 - ANCHORS - a photo of your speaker (shot 1 opens on them) and a voice clip. More useful here than anywhere: one person carries the whole take.</li><li>3 - YOUR PROMPT - ONE premise, one speaker. The writer writes the whole speech. num_shots 0 = it decides. Passthrough works here too.</li><li>4 - CONTROLS - <code>take_seconds</code> is the star: 30 ships, 60 clears TikTok's minute. <code>window</code> stays on auto - it sizes itself to your card.</li><li>5 - ENGINE / 6 - OUTPUT - same as above.</li></ul>Keep takes to about 4 windows for now - very long takes slowly sharpen.<h2 id="rules-of-thumb-(both-workflows)">Rules of thumb (both workflows)</h2><ul><li>Spoken lines: 8-12 words per shot. Short lines sync; long lines garble.</li><li>Say the sounds you want ("rain on the roof, a fridge hum") or it invents its own.</li><li>Keep your character's face in frame - faces carry identity between shots.</li></ul><h2 id="if-something-breaks">If something breaks</h2><ul><li>Red node? Update the pack in Manager, restart, reload the workflow from disk.</li><li>Render crawls at low wattage? Lower resolution or frames per shot, or use the remote encoder.</li><li>Only one of the two workflows shows in your sidebar? Fixed in 2.6.5 - re-download both.</li><li>Still stuck: comment with your console log. I answer.</li></ul>Deep dives: the two articles linked on this page. Every lane also has a short note on the canvas. Detailed guide for people that can read good: <a target="_blank" rel="ugc" href="https://civarchive.com/articles/34046/every-setting-explained-the-seamless-chain-deep-manual">Every setting explained: the Seamless Chain deep manual | Civitai</a>

MiniMax-H3 Multishot — Seamless Chain: multi-shot scenes that render as one continuous take (picture + audio)

Critical fix. If you used the long-form MEMORY workflow in v1.1 or v1.2, it could not have worked. Update the node pack and restart ComfyUI fully — a browser refresh is not enough.The bug<code>H3MultishotMemorySampler</code> called <code>vae_decode_audio()</code> without importing it, raising a <code>NameError</code> at audio decode — after all sampling had finished. So it destroyed a completed render rather than failing fast, which is the worst possible moment to fall over.Checking every commit that has touched that file, the import appears exactly once in all of them, in the other sampler. This node had never worked in any released version. It shipped broken in v1.1 and again in v1.2.Reported with a correct diagnosis and the correct patch by Nebuluss in <a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/ComfyUI-H3-Multishot/issues/1">issue #1</a>. Fixed and verified with a real two-shot render — 247 frames, audio stream present, and the embedded graph confirms the memory sampler itself produced it.Also in this version: H3 Free Text EncoderUnloads the text encoder before the DiT loads. This matters most for reference-to-video, where the Qwen3-VL encoder (~16.5 GB) and the ref2va DiT (~25.9 GB) together oversubscribe a 32 GB card. Measured on an RTX 5090, ref2va with one reference image at 124 frames:<pre><code>encoder resident, DiT fully loaded: 90+ min, killed, no output
encoder evicted + --reserve-vram 7: 12.3 min, success</code></pre>Both are needed. Eviction alone still let the DiT load completely at 25.2 GB with nothing left for activations.The diagnostic that found this was power draw, not utilisation. The GPU reported 98% "utilisation" at 146 W — a 5090 doing real diffusion pulls 400 W+. 98% at 146 W is a card spinning on memory transfers. With headroom it runs at 410 W. Full residency is not the goal; total residency under the card size is.What changed in how this gets releasedAn import that survived two releases means the process failed, not just the code. Two checks now gate every package:<ul><li>Provenance, not existence. My evidence file claimed the MEMORY workflow had a render behind it — pointing at a file actually produced by the other sampler. The gate confirmed the file existed and never asked what made it. It now requires the workflow's own node class to appear in the render's embedded ComfyUI graph.</li><li>A static import check over every shipped <code>.py</code>, so a call that parses fine but <code>NameError</code>s at runtime cannot ship again.</li></ul>Everything from v1.2 is still here<ul><li>Keyframes at any position — up to 6 anchors by fraction or frame index, not just first/last. Measured: an anchor requested at frame 121 landed at frame 122 of 243, reached by continuous motion with no cut.</li><li>H3 Condition Strength and H3 Reference Audio (stereo guard — a mono reference clip crashes the sampler with an unhelpful shape mismatch).</li><li>Image-to-video, the ~4x encoder-eviction speedup for multishot, crossfaded audio seams, and the self-repairing script parser.</li></ul>Links<ul><li><a target="_blank" rel="ugc" href="https://github.com/jlucasmcrell/ComfyUI-H3-Multishot">Node pack on GitHub</a></li><li><a target="_blank" rel="ugc" href="https://civitai.com/models/2833352">MiniMax-H3 GGUF (the DiT)</a></li><li><a target="_blank" rel="ugc" href="https://civitai.com/models/2834385">MiniMax-H3 Text Encoder GGUF</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/MiniMax-H3">VAEs</a></li></ul>SupportEverything I publish is free and stays free. If it saved you a night of debugging, tips keep the 5090 warm: <a target="_blank" rel="ugc" href="https://ko-fi.com/joeygambino">Ko-fi</a> · <a target="_blank" rel="ugc" href="https://github.com/sponsors/jlucasmcrell">GitHub Sponsors</a> · <a target="_blank" rel="ugc" href="https://liberapay.com/joeygambino">Liberapay</a>.

Files with SHA256: 5a8a54643e35d7749d706b3fac5a22287c98ba51dda121e541e01e053c9ff989