<h2 id="endless-minimax-h3-(with-endless-lipsync)-v1.5">Endless MiniMax H3 (with Endless LipSync) v1.5</h2>When MiniMax H3 came out, I was badly missing the <a target="_blank" rel="ugc" href="https://civarchive.com/models/2701632/endless-wan-22-i2v-svi-2-pro">Endless Wan 2.2 I2V (SVI 2 Pro)</a> features, so at first I look at other options, but most of them included an AIO node that could do everything, and I couldn't use most of my workflow, because they did everything inside that huge node. The only exception to this, was <a target="_blank" rel="ugc" href="https://github.com/NikoDemon80/ComfyUI-H3-Motion-Context">ComfyUI-H3-Motion-Context</a>, which I could easily integrate into my setup. Its big shortcoming though, was that it just created single videos. No way to concatenate them without the lossy step of decoding and re-encoding in a video editor. So, I created a custom node that could do just that, and voila..Endless MiniMax H3 (with Endless LipSync) v1.0 A simple workflow to create MiniMax H3 videos of unlimited duration, using <a target="_blank" rel="ugc" href="https://github.com/NikoDemon80/ComfyUI-H3-Motion-Context">ComfyUI-H3-Motion-Context</a> and <a target="_blank" rel="ugc" href="https://github.com/noembryo/ComfyUI-noEmbryo#h3-motion-context-clip-stitcher">H3 Motion Context Clip Stitcher</a>. I can easily create a 1:30 lip-synced video myself, with a RTX 3060 12GB at around 2 hours (with retries).It can use both FL2AV and Ref2AV, and can also create normal MiniMax H3 videos. The extra parts though, are the saving/loading of the latent from every generation we do, and the stitching of all (or some) of them, whenever we want a full video. Nothing visible at the connections, no indication that there were more than one video. Now, with much less degradation for multiple clips without a camera cut. We can try and re-try every generation, looking for the best one, and then proceed to the next. We can re-do any previous generation if we like too, but we will not be able to use the clips after it (previous clips are not affected), because they are in a way, "fused" with the replaced one.Version 1.6 <ul><li>Recreated the Configuration panel and other subgraphs, to make them more compatible with the newer front-end versions (fixed the reset values after refresh bug)</li><li>Added an optional Clip Refiner optimization group</li></ul>Version 1.5<ul><li>Project name that creates a directory with all project's generations</li><li>Added Purge all clips group</li><li>Connected the 5th Ref Image that was disconnected</li><li>Simplified Audio 1 Forced Multi</li><li><code>Refiner</code> that fixes (almost completely) the overexposed/over-contrasted clips from <code>H3 Motion Context</code>, on a long video without much movement.</li></ul><h3 id="controls">Controls</h3>The RED nodes Enable/Disable parts of the workflow.<ul><li>Generation Mode (select only one)<ul><li>T2V / I2V (FL2VA)</li><li>REF2VA</li></ul></li><li>Optimizations (select as many as you want, but only 1 Attention and/or only 1 Cache) This panel controls<ul><li>All the nodes that are inside the <code>Optimizations/LoRA</code> subgraph.</li><li>The H3 Motion Context Clip Refiner subgraph, that removes much of the degradation that accumulates when stitching together latent clips without camera cuts (it's clearly visible after 4-5 clips, but is apparent from the 2nd-3rd clip). It's not needed if there are camera cuts in the clip, so you can turn it off to speed-up the generation.</li></ul></li><li>Reference Items (select as many as you want) Special usage for the <code>&lt;Audio 1&gt; Forced Multi</code>, more for them later.</li><li>Setup (select as many as you want depending on the goal)<ul><li>Generate starts a generation. You don't need this if you just stitching clips</li><li>Preview enables the main video preview that can show you where the generation is going before it finishes, so you can stop bad generations</li><li>Use previous clip, uses the last part of the previous generation to start the current one, continuing the video. Enable it if you want the current generation to be stitched with the previous video</li><li>Stitch clips, stitches all the clips (depending on the Stitcher's settings) from the <code>h3_context</code> folder (this is the default folder that the H3 Motion Context node uses to save the latents)</li><li>Stitch last only, stitches only the last video with the current</li><li>Save Video just saves the generated video</li><li>Purge all clips, deletes all clips from the current Project directory</li></ul></li></ul>The GREEN nodes are various settings nodes that must be setup. Apart from the Prompt, Seed and LoRA nodes, the most important are these:<ul><li>Configuration It contains all the settings for the video. The most important setting for the clip stitching, is the Save to clip Index number. This specifies the clip file's number, that the latent of the current generation will be saved to (overwriting any previous existing file). This number also tells us who is the previous clip file that we use to start our current video clip.</li><li>Setup multiple Forced Audio clips This is used if we use our own audio for the video, and we want to also stitch many clips together. More about it in the Usage section.</li></ul><h3 id="usage">Usage</h3>Most of the settings are self explanatory (like Optimizations or enabling Reference items). Here, I will just list the main goals of the workflow.The first thing to do, is to set the Project Name and the other video options at the Settings.This Project Name will be used to save your generated clips and videos in a folder of the same name, under the EndlessMMH3 directory.<ul><li>Create a normal video<ul><li>Select Generation Mode</li><li>Enable Generate, Preview and Save Video</li><li>Save to clip Index to 1</li><li></li></ul><blockquote>You get a video and that's it.</blockquote></li><li>Create a lip-synced video<ul><li>Select Generation Mode</li><li>Enable Generate, Preview and Save Video</li><li>Enable &lt;Audio 1&gt; Forced Multi and load an audio file (never mind about the settings, except for the <code>Start offset</code>)</li><li></li></ul><blockquote>You get a video that is lip-synced with the provided audio</blockquote></li><li>Create an Endless video<ul><li>Create a normal video</li><li>Enable Use previous clip</li><li>Save to clip Index to 2 and generate</li><li>Save to clip Index to 3 and generate</li><li>Save to clip Index to 4 and generate</li><li>...</li><li></li></ul><blockquote>You get many small videos, each one of them starts with the ending of the previous one. At a later time, you will use the Stitcher, to stitch them all together to one full video.</blockquote></li><li>Create an Endless video, stitched<ul><li>Enable Stitch clips With every generation, the Stitcher will stitch all the previous clips with the currently generated one. This way you always get the full video to check.</li><li>If you also enable Stitch last only, only the previous and the current videos are stitched together, so you can check the connection without waiting for the full video to be created. You can always stitch them all together at the end.</li><li></li></ul><blockquote>You get a full video every time, or just the last 2 videos connected, for previewing the connection.</blockquote></li><li>Create an Endless lip-synced video, stitched<ul><li>Create an Endless video</li><li>Enable &lt;Audio 1&gt; Forced Multi</li><li>At the Setup multiple Forced Audio clips panel there are some settings.<ul><li>Start offset: At the 1st gen, you put here the initial offset that you want for the song (e.g. where the lyrics start). After every successful generation (when you advance the Save to clip Index number), you must copy here the value that is in the <code>Copy to Next Start offset</code> box.</li><li>Frame offset (ignore if 1st clip): Never mind at 1st generation. After every successful generation (when you advance the Save to clip Index number), you must copy here the value that is in the <code>Copy to Next Frame offset</code> box.</li><li>context_length must be the same value everywhere (here, at the <code>Motion Context</code>, and at the <code>H3 Motion Context Clip Stitcher</code>). It's the number of common frames the 2 video clips use to blend together.</li><li></li></ul><blockquote>You get a full lip-synced video every time, or just the last 2 videos connected, depending on the Stitch clips and Stitch last only settings.</blockquote></li></ul></li><li>Just stitch the clips together You just have to enable the Stitch clips and the Save Video All (or some of them depending on the settings), of the clips in the current <code>h3_context</code> folder, will be concatenated to a single full video.</li></ul>Notes:<ul><li>All generated clips that need to be stitched, must have the same dimensions.</li><li>You can organize past generations inside <code>EndlessMMH3</code> folder, using the Project Name to create the directories.</li><li>You can stitch together clips that are generated from either fl2va or ref2va.</li><li>In this workflow, I don't use the normal ref2va model in the <code>Reference to Video</code> node, but rather the fl2va with the <code>ref_lora_layer20-49adaln</code> LoRA that has better quality. You can check some LoRAs with different weights <a target="_blank" rel="ugc" href="https://huggingface.co/morisoba/ComfyUI_extracted_lora/tree/main/minimax-h3">here</a>, or totally bypass the LoRA and use the normal ref2va model.</li></ul>Models used:<ul><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/MiniMax-H3-experimental/resolve/main/minimax_h3_fl2va_pruned_w4a8_mixed.safetensors">minimax_h3_fl2va_pruned_w4a8_mixed</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/MiniMax-H3-experimental/resolve/a3e7d8da4ae7ba8df0779094cf5ab9d6ee855fe4/minimax_h3_ref2va_pruned_w4a8_mixed.safetensors">minimax_h3_ref2va_pruned_w4a8_mixed</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/sakamakismile/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4/resolve/main/qwen3vl_32b_heretic_minimax_h3_nvfp4.safetensors">qwen3vl_32b_heretic_minimax_h3_nvfp4</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Kijai/MiniMax-H3-experimental/resolve/main/minimax_h3_video_vae_int8_convrot.safetensors">minimax_h3_video_vae_int8_convrot</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors">minimax_h3_audio_vae_fp32</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI/resolve/main/minimax_h3_fl2v_turbo_8step_v1.0_comfyui_resized_avg_rank_21_bf16.safetensors">minimax_h3_fl2v_turbo_8step_v1.0_comfyui_resized_avg_rank_21_bf16</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/morisoba/ComfyUI_extracted_lora/resolve/main/minimax-h3/minimax_h3_ref_lora_layer20-49adaln_proj_rank_256_bf16.safetensors?download=true">minimax_h3_ref_lora_layer20-49adaln_proj_rank_256_bf16</a></li></ul>Custom Nodes used:<ul><li><a target="_blank" rel="ugc" href="https://github.com/rgthree/rgthree-comfy">rgthree-comfy</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/NikoDemon80/ComfyUI-H3-Motion-Context">ComfyUI-H3-Motion-Context</a></li><li><a target="_blank" rel="ugc" href="https://github.com/noembryo/ComfyUI-noEmbryo">ComfyUI-noEmbryo</a></li><li><a target="_blank" rel="ugc" href="https://github.com/xmarre/ComfyUI-Spectrum-MiniMax-H3">ComfyUI-Spectrum-MiniMax-H3</a> (optional)</li></ul>

Endless MiniMax H3 (with Endless LipSync)

Version 1.6 <ul><li>Recreated the Configuration panel and other subgraphs, to make them more compatible with the newer front-end versions (fixed the reset values after refresh bug)</li><li>Added an optional Clip Refiner optimization group</li></ul>

Files with SHA256: 244166e6e9b169c34f1ce41ddc0ea73775ee8481c0c94b7b340ee7daca2dda06