MiniMax H3 Turbo Lora ComfyUI

Name: MiniMax H3 Turbo Lora ComfyUI
Author: drbaph

Providerdrbaph

Categorytext-to-video

Licenseapache-2.0

Downloads135.1K

Stars0

Overview

The MiniMax H3 Turbo LoRA for ComfyUI brings professional-grade text-to-video generation directly into a modular node-based workflow. Unlike standalone API implementations, this LoRA integration allows developers to fine-tune temporal consistency and motion dynamics within their own local or cloud pipelines. It is specifically designed for those needing high-fidelity video synthesis without the overhead of full model retraining. By leveraging the Turbo architecture, it significantly reduces inference latency, making it viable for iterative prototyping and rapid asset generation. Integration is straightforward via ComfyUI's custom node ecosystem, enabling developers to chain video generation with existing image-to-image or control-net workflows for precise cinematic control.

Highlights

High-fidelity text-to-video generation via ComfyUI nodes
Turbo architecture ensures low-latency video inference
Apache-2.0 license allows flexible commercial integration
Improved temporal stability over standard base models
Seamlessly chains with existing modular AI workflows

Usage

Install

# Install Hugging Face transformers
pip install transformers torch

SDK Usage

# Load model with transformers
from transformers import AutoModel, AutoTokenizer

model = AutoModel.from_pretrained("drbaph/MiniMax-H3-Turbo-Lora-ComfyUI")
tokenizer = AutoTokenizer.from_pretrained("drbaph/MiniMax-H3-Turbo-Lora-ComfyUI")

Hugging Face Download

We recommend downloading the model via the Hugging Face CLI or Hub SDK.

Guidance：Before downloading, install huggingface_hub with:

Guidance

pip install -U huggingface_hub

CLI Download

Download the full repository

huggingface-cli download drbaph/MiniMax-H3-Turbo-Lora-ComfyUI

Download a single file to a local folder (e.g. config.json into ./dir)

huggingface-cli download drbaph/MiniMax-H3-Turbo-Lora-ComfyUI config.json --local-dir ./dir

See the official docs for more CLI options

SDK Download

# 模型下载
from huggingface_hub import snapshot_download
model_dir = snapshot_download('drbaph/MiniMax-H3-Turbo-Lora-ComfyUI')

Git Download

Make sure git-lfs is installed first

Git Download

git lfs install
git clone https://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI

To skip LFS large-file downloads, use:

Skip LFS

GIT_LFS_SKIP_SMUDGE=1 git clone https://huggingface.co/drbaph/MiniMax-H3-Turbo-Lora-ComfyUI

Model files are hosted on the Hugging Face Hub — download directly via HF CLI / SDK / Git, not through this site.

PyTorch / Transformers Usage

Install Transformers

pip install -U transformers torch

Load the model and run inference

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained('drbaph/MiniMax-H3-Turbo-Lora-ComfyUI')
tokenizer = AutoTokenizer.from_pretrained('drbaph/MiniMax-H3-Turbo-Lora-ComfyUI')

Full Documentation

来源: HuggingFace

---
license: apache-2.0
base_model: Comfy-Org/MiniMax-H3
base_model_relation: adapter
pipeline_tag: text-to-video
tags:
- minimax-h3
- lora
- adapter
- comfyui
- text-to-video
- reference-to-video
- audio-video
- synchronized-audio
- few-step
- turbo
- accelerated-inference
- lightx2v
- dynamic-rank
- svd
- safetensors
- bfloat16
- pruned-model
- curve-form
---

MiniMax-H3 Turbo LoRAs for ComfyUI

This repository contains MiniMax-H3 Turbo LoRAs converted and optimized for ComfyUI:

dynamic-rank FL2V and Ref2V LoRAs derived from the official ComfyUI weights in lightx2v/Minimax-h3-Turbo;

pruned/curve-form compatibility conversions of the Turbo LoRAs released by larryvrh.

These LoRAs accelerate MiniMax-H3 video and synchronized-audio generation by reducing the required number of sampling steps.

---

Update — LightX2V FL2V and Ref2V Dynamic-Rank LoRAs

Three dynamically resized BF16 LoRAs are now included. Their source weights were published in the correct MiniMax-H3 ComfyUI layout by:

lightx2v/Minimax-h3-Turbo

New files

| File | Source | Purpose |
| --- | --- | --- |
| minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_resized_avg_rank_21_bf16.safetensors | minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors | Four-step 768p FL2V generation. |
| minimax_h3_fl2v_turbo_8step_v1.0_comfyui_resized_avg_rank_21_bf16.safetensors | minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors | Eight-step FL2V generation with a greater quality focus. |
| minimax_h3_ref2v_turbo_4step_v0.1_comfyui_resized_avg_rank_21_bf16.safetensors | minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors | Four-step reference-to-video generation. |
| minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_resized_avg_rank_64_bf16.safetensors | minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_bf16.safetensors | Four-step 768p FL2V generation, high-fidelity dynamic-rank resize. |
| minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_resized_avg_rank_28_bf16.safetensors | minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_bf16.safetensors | Four-step 768p FL2V generation, compact dynamic-rank resize. |

The LightX2V files were already supplied in the correct ComfyUI namespace and QKV layout. No namespace conversion or AdaLN pruning was required.

How the LoRAs were resized

Each LoRA was compressed using an exact compact SVD calculated from its LoRA A/B factors without materializing the full dense weight updates.

The process:

1. Treats every attention, MLP and token-refiner projection independently.
2. Separates fused QKV adapters into their original Q, K and V projections.
3. Calculates the exact singular spectrum of each effective B @ A update.
4. Selects a dynamic rank independently for every projection.
5. Preserves one additional safety rank beyond the Frobenius-energy threshold.
6. Rebuilds balanced factors as A = sqrt(S) × Vh and B = U × sqrt(S).
7. Re-fuses Q, K and V using concatenated A factors and a block-diagonal B matrix.
8. Bakes the source alpha / rank scale into lora_B.
9. Removes the no-longer-required alpha tensors.
10. Saves all output tensors as BF16.

All three files target the same projection-level average rank:

Rank sum: 6778

Projection count: 312

Average projection rank: 21.724359

Filename rank: floor(21.724359) = 21

Q, K and V are counted separately. The stored rank of a fused QKV adapter is the sum of its three independently selected ranks.

Resize results

| LoRA | Output size | Average rank | Mean Frobenius norm retained | Baked scale |
| --- | ---: | ---: | ---: | ---: |
| FL2V 4-step v1.0 768p | 284.36 MiB | 21.724359 | 94.95% | 1.0 |
| FL2V 8-step v1.0 | 311.89 MiB | 21.724359 | 97.72% | 0.0625 |
| Ref2V 4-step v0.1 | 311.79 MiB | 21.724359 | 98.33% | 0.0625 |
| FL2V 4-step v1.1 768p (rank 64) | 891.65 MiB | 64.500000 | 98.18% | 1.0 |
| FL2V 4-step v1.1 768p (rank 28) | 375.55 MiB | 28.500000 | 93.59% | 1.0 |

The Ref2V source was reduced from approximately 1,865.57 MiB to 311.79 MiB, an 83.29% reduction. The v1.1 768p source (1,865.57 MiB) was reduced to 891.65 MiB (−52.21%) at rank 64 and to 375.55 MiB (−79.87%) at rank 28.

A separate Frobenius threshold was solved for each LoRA because their singular-value distributions differ. Reusing one fixed threshold would not produce the same average rank.

Ref2V numerical validation

Direct comparison between the original and resized effective LoRA updates produced:

| Measurement | Result |
| --- | ---: |
| Global cosine similarity | 99.9190% |
| Global relative L2 error | 4.0234% |
| Mean projection Frobenius norm retained | 98.3332% |
| Minimum projection Frobenius norm retained | 96.4904% |
| Projection rank range | 2–105 |
| Size reduction | 83.29% |

These are numerical weight-reconstruction measurements, not a perceptual video-quality score.

FL2V 4-step v1.1 768p numerical validation

Direct comparison between the v1.1 source and each resized variant produced:

| Measurement | Rank 64 | Rank 28 |
| --- | ---: | ---: |
| Global cosine similarity | 99.1169% | 97.3279% |
| Global relative L2 error | 13.2601% | 22.9625% |
| Mean projection Frobenius norm retained | 98.1774% | 93.5945% |
| Minimum projection Frobenius norm retained | 98.0817% | 92.6661% |
| Projection rank range | 2–120 | 2–99 |
| Size reduction | 52.21% | 79.87% |

These are numerical weight-reconstruction measurements, not a perceptual video-quality score.

Structural validation

Every resized LightX2V file contains:

| Component | Count |
| --- | ---: |
| BF16 tensors | 416 |
| LoRA A/B pairs | 208 |
| Independently ranked projections | 312 |
| QKV modules | 52 |
| Alpha tensors | 0 |

Validation confirmed:

all tensors contain finite BF16 values;

every LoRA A tensor has a matching LoRA B tensor;

ComfyUI recognizes all 208 adapters and all 416 tensors;

no incompatible or unresolved tensors remain;

QKV rank columns remain block-diagonal and ordered as Q, K and V;

the original effective LoRA scale is preserved;

the exact average projection rank is 21.724359.

---

Workflow Example

The following output was generated with the recommended pruned-model workflow using the v4 step-600 EMA LoRA.

Current workflow

Download the current ComfyUI workflow

Workflow: fl_minimax_h3_turbo_lora_example_workflow.json

LoRA: minimax_h3_turbo_v4_step600_ema_pruned_comfyui.safetensors

Steps: 6–8

Sampler: Euler

Scheduler: Beta

LoRA strength: 1.0

Video sigma shift: 12

Audio sigma shift: 4–6

The previous workflow remains available here:

Download the previous workflow

View the previous workflow output

---

Included Files

LightX2V dynamic-rank LoRAs

| File | Description |
| --- | --- |
| minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_resized_avg_rank_21_bf16.safetensors | Dynamic-SVD resize of the official LightX2V four-step 768p FL2V LoRA. |
| minimax_h3_fl2v_turbo_8step_v1.0_comfyui_resized_avg_rank_21_bf16.safetensors | Dynamic-SVD resize of the official LightX2V eight-step FL2V LoRA. |
| minimax_h3_ref2v_turbo_4step_v0.1_comfyui_resized_avg_rank_21_bf16.safetensors | Dynamic-SVD resize of the official LightX2V four-step Ref2V LoRA. |

Pruned/curve-form compatibility conversions

| File

View official source