Download
1 variant available
0
7
License:
On Civitai (generation, training, sharing LoRAs):
covered by Civitai’s agreement with MiniMax
The MiniMax H3 Community License Agreement’s grant excludes the European Union, the United Kingdom, the Republic of Korea and the United States of America.
MiniMax H3
Here is the simple workflow. It works on my side and takes about 10 minutes to generate a 0.3 megapixel 15 second video on 4070TI Super + 32GB RAM. I'm not a pro at all, but as I was looking for beginner-friendly workflow, I didn't find any, so here it is
Main features of this workflow:
Optional Turbo along with other LORAs: Load multiple LORAs in comfortable UI, switch the turbo state with one click
Prompt enhancer: using Florence2 to describe the source image and Hermes-3-Llama (and its system prompt), this workflow improves your existing prompt
Simple properties customizer: tired of calculating resolution manually, looking at all the nodes trying to find the required one? No need! Everything is in one subgraph
Please, leave a comment if you used this workflow, it is really important to me. Have a great generations!
Models links:
unet minimax_h3_fl2va_pruned_int8_convrot.safetensors
clip qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors
vae minimax_h3_video_vae_int8_convrot.safetensors
audio_vae minimax_h3_audio_vae_fp32.safetensors
turbo lora minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors
llm model hermes-3-llama-3.1-8B-lorablated-q4_k_m.gguf