<div data-youtube-video><iframe src="https://www.youtube.com/embed/H5L5sO7dRu8" width="640" height="360" allowfullscreen="true"></iframe></div>
https://youtu.be/H5L5sO7dRu8 
 
Creator: AI-KSK · Collection: Video 327 / six-workflow MiniMax H3 RefMod episode 
 
Workflow file: <code>06_多参考生成_B_TextEncode.json</code> · Cover file: <code>多参考生成 B：TextEncode 控制-6.png</code> 
 
<h3>Main features:</h3> 
 
The second video-generation route puts saved references into <code>MiniMaxH3RefModTextEncode</code> together with the Qwen H3 text encoder and H3 video VAE. The result is conditioning with embedded references and an optional <code>reference_map</code> that can identify inputs as <code>&lt;Video n&gt;</code> or <code>&lt;Picture n&gt;</code>. Unlike the Apply path, the reference-aware text encoder performs the attachment; the active sampler path contains no Apply node, and adding a second Apply with the same assets could duplicate references. 
 
This is a flexible *starting graph* for explicit multi-subject reference assignment, not a packaged finished two-person film. Its original Text Encode prompt field is empty. The fourth selected RefMod is an extra character not built by workflows 01–04. Its active bundle does not include the voice RefMod used in Generation A. 
 
<h3>Exact connected references and settings</h3> 
 
- Main loader slots: <code>character/Rin_Identity</code>, <code>motion/Action_01</code>, <code>scene/Dojo_OneLocation</code>, <code>character/war</code>; each strength <code>1</code>, copies <code>1</code>. 
- Missing asset: <code>character/war</code> is not included and has no creator JSON in this collection. 
- Missing prompt: <code>MiniMaxH3RefModTextEncode.prompt=""</code>; <code>reference_fps=24</code>, <code>max_total_tokens=0</code>. 
- Video latent: <code>832×480</code>, <code>73</code> frames. Seed <code>9527</code> fixed, Euler, beta scheduler, <code>4</code> steps and denoise <code>1</code>; 24 fps H.264 MP4, CRF 19, prefix <code>AIKSK_REF06_TEXT_</code>. 
- An audio VAE is used for output audio decoding, but no <code>voice/Rin_Voice_01</code> reference is selected. 
- A second loader in the original JSON feeds an <code>Inspect</code>/<code>PreviewAny</code> branch; it is not another reference injection into the sampler. 
 
<h3>Suggested workflow:</h3> 
 
First create your character, movement and scene RefMods; supply a separate authorized second-character reference for <code>character/war</code> or replace that selection. Inspect the actual <code>reference_map</code> before writing subject labels: a multi-image asset can appear under one <code>&lt;Video n&gt;</code> reference rather than being treated as several <code>&lt;Picture n&gt;</code> entries. Fill the blank prompt, specifying which mapped subject is on which side and what visible actions occur in the short shot. Check the result against your instruction. At 73/24 the JSON's nominal output length is about 3.04 seconds, not the ten-second A configuration. 
 
<h3>Limits</h3> 
 
Reference labels and encoder context do not provide a two-character skeleton-binding engine, guaranteed voice consistency or proven superiority over Apply. The cover shows a woman with a sword and a glowing TextEncode diagram; it is promotional artwork, not footage produced by this file. 
 
<h3>Dependencies, tutorial and deployment</h3> 
 
This project uses ComfyUI, MiniMax H3 weights and the <a href="https://github.com/Luisacaotica/ComfyUI-MiniMaxH3Mod" rel="ugc">upstream RefMod node pack</a>; workflows involving <code>VHS_</code> nodes require <a href="https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite" rel="ugc">Video Helper Suite</a>. See <a href="https://huggingface.co/Comfy-Org/MiniMax-H3" rel="ugc">Comfy-Org's H3 model inventory</a> for official distribution guidance. The exact int8-convrot video VAE and Turbo 4-step LoRA named in the creator's JSON need separate availability/provenance checks. We distribute this workflow JSON and its original cover, not external checkpoints or third-party media. Respect applicable rights and licenses. 
 
<a href="https://www.compshare.cn/images/323moGjpvfeM?referral_code=A7fTQUYY9lCDFJwA6MMvoq" rel="ugc">Deploy AI-KSK's H3｜RefMod免训练超级参考 image on 优云智算</a>. Referral disclosure: the deployment link includes an AI-KSK referral code and may support the creator. Check the cloud provider's actual billing and environment before deploying. 
 
<a href="https://youtu.be/H5L5sO7dRu8" rel="ugc">Watch the complete six-workflow YouTube tutorial</a>. This project is one module in that episode. Description and settings were prepared by inspecting the source JSON and transcript, not from a new measured benchmark. 
 
Bilibili tutorial: https://www.bilibili.com/video/BV1F2nZ6SEA7/

MiniMax H3 Multi-Reference TextEncode

<div data-youtube-video></div>
https://youtu.be/H5L5sO7dRu8 
 
Creator: AI-KSK · Collection: Video 327 / six-workflow MiniMax H3 RefMod episode 
 
Workflow file: <code>06_多参考生成_B_TextEncode.json</code> · Cover file: <code>多参考生成 B：TextEncode 控制-6.png</code> 
 
Main features: 
 
The second video-generation route puts saved references into <code>MiniMaxH3RefModTextEncode</code> together with the Qwen H3 text encoder and H3 video VAE. The result is conditioning with embedded references and an optional <code>reference_map</code> that can identify inputs as <code>&lt;Video n&gt;</code> or <code>&lt;Picture n&gt;</code>. Unlike the Apply path, the reference-aware text encoder performs the attachment; the active sampler path contains no Apply node, and adding a second Apply with the same assets could duplicate references. 
 
This is a flexible *starting graph* for explicit multi-subject reference assignment, not a packaged finished two-person film. Its original Text Encode prompt field is empty. The fourth selected RefMod is an extra character not built by workflows 01–04. Its active bundle does not include the voice RefMod used in Generation A. 
 
Exact connected references and settings 
 
- Main loader slots: <code>character/Rin_Identity</code>, <code>motion/Action_01</code>, <code>scene/Dojo_OneLocation</code>, <code>character/war</code>; each strength <code>1</code>, copies <code>1</code>. 
- Missing asset: <code>character/war</code> is not included and has no creator JSON in this collection. 
- Missing prompt: <code>MiniMaxH3RefModTextEncode.prompt=""</code>; <code>reference_fps=24</code>, <code>max_total_tokens=0</code>. 
- Video latent: <code>832×480</code>, <code>73</code> frames. Seed <code>9527</code> fixed, Euler, beta scheduler, <code>4</code> steps and denoise <code>1</code>; 24 fps H.264 MP4, CRF 19, prefix <code>AIKSK_REF06_TEXT_</code>. 
- An audio VAE is used for output audio decoding, but no <code>voice/Rin_Voice_01</code> reference is selected. 
- A second loader in the original JSON feeds an <code>Inspect</code>/<code>PreviewAny</code> branch; it is not another reference injection into the sampler. 
 
Suggested workflow: 
 
First create your character, movement and scene RefMods; supply a separate authorized second-character reference for <code>character/war</code> or replace that selection. Inspect the actual <code>reference_map</code> before writing subject labels: a multi-image asset can appear under one <code>&lt;Video n&gt;</code> reference rather than being treated as several <code>&lt;Picture n&gt;</code> entries. Fill the blank prompt, specifying which mapped subject is on which side and what visible actions occur in the short shot. Check the result against your instruction. At 73/24 the JSON's nominal output length is about 3.04 seconds, not the ten-second A configuration. 
 
Limits 
 
Reference labels and encoder context do not provide a two-character skeleton-binding engine, guaranteed voice consistency or proven superiority over Apply. The cover shows a woman with a sword and a glowing TextEncode diagram; it is promotional artwork, not footage produced by this file. 
 
Dependencies, tutorial and deployment 
 
This project uses ComfyUI, MiniMax H3 weights and the <a href="https://github.com/Luisacaotica/ComfyUI-MiniMaxH3Mod" rel="ugc">upstream RefMod node pack</a>; workflows involving <code>VHS_</code> nodes require <a href="https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite" rel="ugc">Video Helper Suite</a>. See <a href="https://huggingface.co/Comfy-Org/MiniMax-H3" rel="ugc">Comfy-Org's H3 model inventory</a> for official distribution guidance. The exact int8-convrot video VAE and Turbo 4-step LoRA named in the creator's JSON need separate availability/provenance checks. We distribute this workflow JSON and its original cover, not external checkpoints or third-party media. Respect applicable rights and licenses. 
 
<a href="https://www.compshare.cn/images/323moGjpvfeM?referral_code=A7fTQUYY9lCDFJwA6MMvoq" rel="ugc">Deploy AI-KSK's H3｜RefMod免训练超级参考 image on 优云智算</a>. Referral disclosure: the deployment link includes an AI-KSK referral code and may support the creator. Check the cloud provider's actual billing and environment before deploying. 
 
<a href="https://youtu.be/H5L5sO7dRu8" rel="ugc">Watch the complete six-workflow YouTube tutorial</a>. This project is one module in that episode. Description and settings were prepared by inspecting the source JSON and transcript, not from a new measured benchmark. 
 
Bilibili tutorial: https://www.bilibili.com/video/BV1F2nZ6SEA7/

Files with SHA256: 29f8e9f3061dd7c65d515ccde08ce5173d6caf0759a701d6e606c086e17f3f39