MiniMax H3 FL2VA first/last frame + keyframe timeline control, with Motion Lab action repair - a single ComfyUI workflow that pins your story to exact timestamps and generates one continuous cinematic shot.
This workflow is built around the MiniMax H3 FL2VA model (minimax_h3_fl2va_int8_convrot.safetensors) and drives it through a multi-pass SamplerCustomAdvanced pipeline with res_multistep sampling, a Qwen3-VL 32B text encoder (qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors), separate video VAE (minimax_h3_video_vae_fp16) and audio VAE (minimax_h3_audio_vae_fp32), plus a compatible 3D latent upscaler pass at 1.5x before the final save.
Main features:
- First/last frame control: lock the opening and closing frames of the shot.
- Keyframe timeline control: align reference pictures to exact timestamps (e.g. 0.00s / 8.00s) so character, camera and scene evolve exactly as planned.
- Motion Lab action repair: cleans motion artifacts on fast-action sequences.
- Two-stage output: an 8-second PASS1 preview plus the final AllInOne video, with a 3D HiRes 1.5x upscale pass before saving.
Suggested workflow:
1. Prepare first/last frames and any intermediate keyframes for your shot.
2. In the prompt, state which picture aligns to which timestamp, then write one integrated_multimodal_description covering the whole sequence as a single continuous shot.
3. Run PASS1 to verify the timeline beats, then run the final pass with the 1.5x 3D upscale enabled.
Notes: requires MiniMax H3 FL2VA int8 weights, the Qwen3-VL 32B encoder and both video/audio VAEs. Keep the description as one uninterrupted motion - hard cuts are not part of this pipeline's design.
RunningHub (one-click run): https://www.runninghub.ai/zh-cn/post/2095023418159714305?inviteCode=rh-v1111
Bilibili tutorial: https://www.bilibili.com/video/BV1L4tn6WEVb/
Models & resources:
MiniMaxAI/MiniMax-H3-Turbo-Lora: https://huggingface.co/MiniMaxAI/MiniMax-H3-Turbo-Lora
lightx2v/Minimax-h3-Turbo: https://huggingface.co/lightx2v/Minimax-h3-Turbo
fal/MiniMax-H3-Realism-People-LoRA: https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA
Model pack (Quark): https://pan.quark.cn/s/9ec095b95838
RunningHub signup with invite code rh-v1111 gets 1000 RH coins: https://www.runninghub.ai/?inviteCode=rh-v1111
Description
MiniMax H3 FL2VA first/last frame + keyframe timeline control, with Motion Lab action repair - a single ComfyUI workflow that pins your story to exact timestamps and generates one continuous cinematic shot.
This workflow is built around the MiniMax H3 FL2VA model (minimax_h3_fl2va_int8_convrot.safetensors) and drives it through a multi-pass SamplerCustomAdvanced pipeline with res_multistep sampling, a Qwen3-VL 32B text encoder (qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors), separate video VAE (minimax_h3_video_vae_fp16) and audio VAE (minimax_h3_audio_vae_fp32), plus a compatible 3D latent upscaler pass at 1.5x before the final save.
Main features:
- First/last frame control: lock the opening and closing frames of the shot.
- Keyframe timeline control: align reference pictures to exact timestamps (e.g. 0.00s / 8.00s) so character, camera and scene evolve exactly as planned.
- Motion Lab action repair: cleans motion artifacts on fast-action sequences.
- Two-stage output: an 8-second PASS1 preview plus the final AllInOne video, with a 3D HiRes 1.5x upscale pass before saving.
Suggested workflow:
1. Prepare first/last frames and any intermediate keyframes for your shot.
2. In the prompt, state which picture aligns to which timestamp, then write one integrated_multimodal_description covering the whole sequence as a single continuous shot.
3. Run PASS1 to verify the timeline beats, then run the final pass with the 1.5x 3D upscale enabled.
Notes: requires MiniMax H3 FL2VA int8 weights, the Qwen3-VL 32B encoder and both video/audio VAEs. Keep the description as one uninterrupted motion - hard cuts are not part of this pipeline's design.
RunningHub (one-click run): https://www.runninghub.ai/zh-cn/post/2095023418159714305?inviteCode=rh-v1111
Bilibili tutorial: https://www.bilibili.com/video/BV1L4tn6WEVb/
Models & resources:
MiniMaxAI/MiniMax-H3-Turbo-Lora: https://huggingface.co/MiniMaxAI/MiniMax-H3-Turbo-Lora
lightx2v/Minimax-h3-Turbo: https://huggingface.co/lightx2v/Minimax-h3-Turbo
fal/MiniMax-H3-Realism-People-LoRA: https://huggingface.co/fal/MiniMax-H3-Realism-People-LoRA
Model pack (Quark): https://pan.quark.cn/s/9ec095b95838
RunningHub signup with invite code rh-v1111 gets 1000 RH coins: https://www.runninghub.ai/?inviteCode=rh-v1111
comfyui
workflow
workflows
video generation
comfyui workflow
first last frame
minimax h3
fl2va
aiksk
motion lab
keyframe timeline control
Details
Downloads
18
Platform
CivitAI
Platform Status
Available
Created
9/2/2026
Updated
9/2/2026
Deleted
-
