MiniMax H3 + ControlNet Union 2.0 video control and local repaint workflow (VDN 12-step HD edition). One source video drives everything: structural control, audio, tracked mask and inpaint source.
Main features:
- Single 24fps source video (VHS force_rate=24) simultaneously feeds the AIO preprocessor (control_video), audio extraction, SeC point tracking (mask) and the inpaint source
- MODE switch: 1 = pure structural control, 2 = pure local repaint, 3 = control + repaint (auto boolean routing)
- ControlNet Union 2.0 patch (minimax_h3_fun_controlnet_union_pruned_int8_convrot) with AIO preprocessors for depth / pose / edge style control
- SeC full-clip tracking to mask, GrowMask dilation, mask preview
- OpenVDN DMD8 two-pass: 8NFE base, learned latent x2 upscale to 1920x1088, 4NFE refine; native audio AV output at approx 1080P / 24fps
- First frame, last frame and 3 reference image slots for subject/style reference
Suggested workflow:
1. Load one source video (any fps; it is resampled to 24fps)
2. Pick MODE 1/2/3 and the AIO preprocessor you need
3. Add reference images, run the graph and export the safe AV save (approx 1080P, 24fps)
Dependencies:
- minimax_h3_fl2va_pruned_int8_convrot.safetensors (H3 base UNET)
- minimax_h3_fun_controlnet_union_pruned_int8_convrot.safetensors (Union 2.0 patch)
- qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors (CLIP), video/audio VAEs
- minimax_h3_latent_upscaler_3d_fp16.safetensors (learned latent x2)
- OpenVDN/vdn-minimax-h3 (DMD8 stage), KSK T8 custom nodes, SeC model
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2108580245673959425?inviteCode=rh-v1111
Bilibili tutorial video: https://www.bilibili.com/video/BV1pTph6EEUv/
RunningHub registration with invite code rh-v1111 (1000 RH coins): https://www.runninghub.ai/?inviteCode=rh-v1111
Description
MiniMax H3 + ControlNet Union 2.0 video control and local repaint workflow (VDN 12-step HD edition). One source video drives everything: structural control, audio, tracked mask and inpaint source.
Main features:
- Single 24fps source video (VHS force_rate=24) simultaneously feeds the AIO preprocessor (control_video), audio extraction, SeC point tracking (mask) and the inpaint source
- MODE switch: 1 = pure structural control, 2 = pure local repaint, 3 = control + repaint (auto boolean routing)
- ControlNet Union 2.0 patch (minimax_h3_fun_controlnet_union_pruned_int8_convrot) with AIO preprocessors for depth / pose / edge style control
- SeC full-clip tracking to mask, GrowMask dilation, mask preview
- OpenVDN DMD8 two-pass: 8NFE base, learned latent x2 upscale to 1920x1088, 4NFE refine; native audio AV output at approx 1080P / 24fps
- First frame, last frame and 3 reference image slots for subject/style reference
Suggested workflow:
1. Load one source video (any fps; it is resampled to 24fps)
2. Pick MODE 1/2/3 and the AIO preprocessor you need
3. Add reference images, run the graph and export the safe AV save (approx 1080P, 24fps)
Dependencies:
- minimax_h3_fl2va_pruned_int8_convrot.safetensors (H3 base UNET)
- minimax_h3_fun_controlnet_union_pruned_int8_convrot.safetensors (Union 2.0 patch)
- qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors (CLIP), video/audio VAEs
- minimax_h3_latent_upscaler_3d_fp16.safetensors (learned latent x2)
- OpenVDN/vdn-minimax-h3 (DMD8 stage), KSK T8 custom nodes, SeC model
RunningHub workflow page: https://www.runninghub.ai/zh-cn/post/2108580245673959425?inviteCode=rh-v1111
Bilibili tutorial video: https://www.bilibili.com/video/BV1pTph6EEUv/
RunningHub registration with invite code rh-v1111 (1000 RH coins): https://www.runninghub.ai/?inviteCode=rh-v1111