Alissonerdx/Minimax-H3-ComfyUI

MiniMax H3 LoRAs for ComfyUI

Experimental LoRAs for MiniMax H3, packaged for ComfyUI and tested mainly with the ref2va weights.

These are research models, not finished tools. Results vary substantially with the source clip, resolution, prompt and reference images.

Models

Model	Purpose	Trigger	Documentation	Workflow
minimax_h3_lms_v1.0_r64	Sharpness/detail enhancement using an aligned source-video guide	Fixed enhancement caption	LMS / Sharpness guide	minimax_h3_lms_workflow.json
minimax_h3_style_transfer_v1.0_r64	Transfer a visual style from an image or text description onto a video	style_transfer:	Style Transfer guide	minimax_h3_style_transfer_workflow.json
minimax_h3_vfx_edit_v1.0_r128	General prompt-directed VFX editing with an aligned source video	vfx_edit:	VFX Edit guide	minimax_h3_vfxedit_wokflow.json
minimax_h3_vfx_edit_v1.0_r128_ffp	Higher-resolution VFX Edit version with experimental first-frame propagation	vfx_edit:	VFX Edit and FFP guide	minimax_h3_vfxedit_wokflow.json
minimax_h3_head_swap_v1.0_r32	Experimental reference-image head replacement over an aligned source video	head_swap:	Head Swap guide and prompt template	minimax_h3_head_swap_workflow.json
minimax_h3_character_swap_v1.1_ref2va_fro (recommended)	Replace the main guide subject with the character in a reference image while following the guide performance; 80 MB SVD extraction	body_swap:	Character Swap guide, prompt template and examples	Shot Planner / native nodes
minimax_h3_character_swap_v1.0_ref2va_r64_gl	Previous Character Swap release (rank 64)	body_swap:	Character Swap guide	Shot Planner / native nodes
minimax_h3_motion_transfer_v1.0_ref2va (new)	Motion Transfer: the character in the reference performs the guide video's motion; carries ostris' training adapter V2 merged in (185 MB)	motion_transfer:	Native Guide Latents	Planner Guide Latents
minimax_h3_motion_transfer_tara_v1.0_ref2va (new)	Motion Transfer with the guide as a native video reference, aligned frame by frame by TARA (BFSMiniMaxH3NativeTemporal); training adapter V2 merged in (126 MB)	motion_transfer:	Native TARA	Planner TARA
minimax_h3_motion_transfer_ffp_v1.0_ref2va (new)	Motion Transfer FFP: edit the first frame (for example replace the character) and the edit follows the whole video. VFX Edit FFP (rank 128) × 1.0 + SolarRenderer control × 0.5, SVD-compressed to 99% of the Frobenius energy (230 MB)	vfx_edit:	Henge (Motion Transfer · First Frame)	Licenses
solarrenderer_control_comfy_fro_dyn99 / full rank	SolarRenderer's control LoRA converted to ComfyUI H3 keys; the _fro_dyn99 copy keeps 99% of each module's Frobenius energy (757 MB instead of 5.9 GB). Used alone by Henge's Motion Transfer SolarRenderer FFP, on SolarRenderer's own conditioning	SolarRenderer six-part prompt	Henge (Motion Transfer SolarRenderer FFP)	Licenses
solarrenderer_4step_comfy_fro_dyn99 / full rank	SolarRenderer's four-step LoRA converted to ComfyUI H3 keys; _fro_dyn99 at 99% of the Frobenius energy (1.4 GB instead of 5.9 GB). Load it before the control LoRA	—	Henge	Licenses
minimax_h3_live_wallpaper_v1.0_ref2va_r32	Animate one to three reference images; often more faithful to facial features, but with less predictable camera behavior	live_wallpaper:	Live Wallpaper guide	minimax_h3_live_wallpaper_workflow.json
minimax_h3_live_wallpaper_v1.0_ref2va_r64	Camera-aware continuation of R32 for prompt-directed camera behavior and temporal reference frames	live_wallpaper:	Live Wallpaper guide	minimax_h3_live_wallpaper_workflow.json

Motion Transfer and TARA

Four downloadable workflows: Guide Latents / native, Guide Latents / Planner, TARA / native, TARA / Planner.

References are 1..N pictures with roles assigned by the prompt, not a required person/background pair. Several images can describe one identity, multiple characters, outfits, objects, scenery or style. A separate background image is optional. See the Motion Transfer guide for model setup, multiple-reference examples and the modified temporal Comfy Kitchen kernel required for TARA's fused one-pass Kitchen path. The TARA workflows default to Flash partitions, which does not require that extension.

Training adapter V2. The Motion Transfer LoRAs were trained on top of ostris' training adapter V2 (minimax_h3_ref2va_training_adapter_v2.safetensors) kept frozen. Since 2026-11-02 the two files above carry it: LoRA x1 + adapter x1 merged by exact rank concat, then per-module SVD keeping 99% of the Frobenius energy (185 MB and 126 MB instead of 620 MB + 310 MB). Load only turbo + motion LoRA; do not load the adapter again. In same-seed tests they matched the full LoRA + separate adapter, including TARA's frame sync across camera cuts. The previous files (without the adapter) are kept in loras/base_loras_ignore/.

TARA-H3 — Temporally Aligned Reference Attention. Native reference conditioning alone did not reliably reproduce the timing of a source performance in H3, while aligned guide latents (Add Guide) copy it frame by frame. TARA keeps the motion video as native reference tokens, translates its temporal RoPE coordinates onto the generated video's clock, and adds a bounded temporal penalty only to generated-video queries attending to motion-reference keys; image references keep their native coordinates and the attention change adds no learned layers. Motion Transfer TARA v1.0 was trained for this method. Node: BFSMiniMaxH3NativeTemporal in ComfyUI-BFSNodes. Technical report (v0.1.2, experimental, not peer reviewed): TARA-H3: Temporally Aligned Reference Attention for Native Video Conditioning.

All of these run in Henge, the open-source studio built on them.

Latent Upscaler

h3_upscaler_lms_v0.1 is the recommended FP16 release of an experimental latent-upscaler checkpoint. It was trained for 2,000 steps with the same sharpness dataset used for the LMS LoRA and initialized from the LBH-123-AI/Minimax_h3_latent_Upscaler project and its minimax_h3_latent_upscaler_3d_fp16.safetensors checkpoint.

See the Latent Upscaler guide for setup, scope and limitations.

Common setup

Place LoRAs in ComfyUI/models/loras/. The included workflows list their required models and custom nodes inside the graph. Keep the output aspect ratio matched to the source clip and use an H3-supported frame count (17n + 5: 5, 22, 39, 56, 73, 90, 107, 124, ...).

The LMS, Style Transfer, VFX Edit and Head Swap models use an aligned source-video guide through MiniMaxH3AddGuide at frame_idx = 0. This is different from a native image/video reference: the guide shares the generated video's spatial and temporal grid.

License and credits

Licenses are specified by asset. The separate Character Swap license and Motion Transfer license apply only to the Character Swap and Motion Transfer LoRA weight files listed in LICENSES.md; they do not apply to the other LoRAs, workflows, docs, or examples. Use of these LoRAs, including commercial use in videos and films, requires no prior permission. Organizations with over US $1,000,000 in annual gross revenue must credit its origin for commercial use. Offering the LoRA as a hosted API, SaaS, or other remote model service is not allowed. See the full Character Swap and Motion Transfer licenses and check each other asset's applicable terms before use.

Base model credit to MiniMaxAI, ComfyUI-ready weights to Comfy-Org, and guide-latent training support to ostris/ai-toolkit and ComfyUI-AIToolkit-MiniMaxH3.