Comfy-Org/MiniMax-H3 · Discussions

#60 opened 20 days ago by

dennismc1974

Measured cost per clip on six rented GPUs (4090, L40S, H100 PCIe, RTX 5090): same weights, graph and seeds

❤️ 1

#59 opened 23 days ago by

susorokin

Running On <4gb Vram

#58 opened 26 days ago by

Jit2024

Whats the deal here, how is Comfy-Org able to distribute the diffusion-models to US and EU under the community license?

2

#57 opened 28 days ago by

PuppetVision

Sharing: MiniMax-H3 Character Card Workflow !

🔥👍 4

#56 opened about 1 month ago by

sunnyboxs

Ref2VA: strong red/blush patches appear on cheeks in free-generated frames

#55 opened about 1 month ago by

hidetaro7

What node is needed to use embeddings files?

#53 opened about 1 month ago by

darkenHUB

为啥生成视频一直有噪音呢

#51 opened about 2 months ago by

lin996worker

minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors

👍 2

#47 opened about 2 months ago by

Mimeniia

Error

#45 opened about 2 months ago by

Heouzen

problem with off-screen voiceover【画外音问题】

7

#44 opened about 2 months ago by

xuhai518

Where did the nvfp4 files go?

2

#43 opened about 2 months ago by

prevod

MiniMax H3 Audio Fix and 4 step Turbo LoRA for Low VRAM GPUs

#41 opened about 2 months ago by

LabMike3D

角色说话时其它音效和背景音量都特别小，几乎听不到

#40 opened about 2 months ago by

sansan3333

minimax_h3_fl2va_pruned_fp8_scaled.safetensors Failed - Forbidden

👍 2

7

#39 opened about 2 months ago by

dankrif

Model Reloads on prompt/image change?

#38 opened about 2 months ago by

Dalvogalbo2

Anyone tried reduce the original 24fps to 16fps?

#37 opened about 2 months ago by

Lidianevasili

MiniMax H3 video vae int8 convrot and 8-step Turbo LoRA

👍 1

7

#35 opened 2 months ago by

LabMike3D

very slow

#34 opened 2 months ago by

Heouzen

Strix Halo 8060s

🤗 3

5

#33 opened 2 months ago by

bnnoirjean

Monitors go blank after a few minutes of running

12

#32 opened 2 months ago by

Zeengee

Why MiniMax H3 Ruins Faces on Wide Shots?

24

#30 opened 2 months ago by

LabMike3D

Fail forbidden trying to download minimax_h3_ref2va_pruned_fp8_scaled.safetensors

#28 opened 2 months ago by

hanger102

Nvfp4 Pruned model for first last frame and reference to video

#27 opened 2 months ago by

benc122

Just a quick benchmark

🔥❤️ 5

6

#26 opened 2 months ago by

UntMods

Minimax reference to video

#25 opened 2 months ago by

Galo64

Not supported for Apple silicon

5

#24 opened 2 months ago by

Rhystic1

Could you release an FP32 vae?

5

#23 opened 2 months ago by

pausecatito

Please release an INT4-quantized version of MiniMax-H3

👍 4

9

#22 opened 2 months ago by

makisekurisu-jp

Update README.md

#21 opened 2 months ago by

ryanlee-dev

Approximating VAE? (Or video preview in general)

9

#20 opened 2 months ago by

inflatebot

Upscaling 10s or 15s video to 2K failed in ref2va workflows

#19 opened 2 months ago by

anitman

[Help] RX 9070 XT (AMD) 16g vram + 5600x 32g + win11 – CUDA invalid argument when loading MiniMaxH3VideoVAE / any success stories?

8

#17 opened 2 months ago by

litacc

int4 convrot text encoder?

👍 3

20

#16 opened 2 months ago by

deepfree

Well, just in case anyone is wondering if can do i2v + ref, the answer is yes.

👀❤️ 21

8

#15 opened 2 months ago by

Nuke1229

Using minimax text encoder forTextGenerate node

#14 opened 2 months ago by

MHoela

Prompt injected into every output file