Instructions to use lightx2v/Minimax-h3-Turbo with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Diffusers
How to use lightx2v/Minimax-h3-Turbo with Diffusers:
pip install -U diffusers transformers accelerate
import torch from diffusers import DiffusionPipeline from diffusers.utils import load_image, export_to_video # switch to "mps" for apple devices pipe = DiffusionPipeline.from_pretrained("lightx2v/Minimax-h3-Turbo", dtype=torch.bfloat16, device_map="cuda") pipe.to("cuda") prompt = "A man with short gray hair plays a red electric guitar." image = load_image( "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/guitar-man.png" ) output = pipe(image=image, prompt=prompt).frames[0] export_to_video(output, "output.mp4") - Notebooks
- Google Colab
- Kaggle
what's the training difference butween minimax_h3_fl2v_turbo_4step_v1.0_768p and minimax_h3_fl2v_turbo_4step_v1.1_768p
#47 opened about 15 hours ago
by
gaga-lee97
Strange Surprise! with minimax_h3_fl2v_turbo_4step_v1.1_768p_comfyui_bf16.safetensors
3
#46 opened 4 days ago
by
agapes
Ref2Va
๐ 17
7
#45 opened 5 days ago
by
muzahid2222
It seems that the current ref2v turbo model does not perform well in a 9:16 video aspect ratio, and is prone to ghosting or overlapping images in the upper half of the screen.
2
#44 opened 6 days ago
by
lvfengchun
Why is 1 file scanned unsafe?
๐๐ 7
3
#43 opened 7 days ago
by
Trackah123
Recommended inference settings for FL2V Turbo 4-step v1.1 (768p)
๐โค๏ธ 15
17
#42 opened 7 days ago
by
lightx2v
I found great settings
โค๏ธ๐ 10
25
#41 opened 10 days ago
by
hanswurstwerner
I2V! I have used accidently used the ref2V turbo lora in combo with fl2V model, 8 steps and the character preserving and face consistency is 10/10!!
6
#40 opened 11 days ago
by
Sikaworld1990
Decent quality but slow-speed video results
1
#39 opened 11 days ago
by
Jam747
thoughts and impressions from using it so far
๐ 3
11
#38 opened 12 days ago
by
m8rr
8+ step loras seem to make sense in terms of quality of video and audio. question
๐ 2
#37 opened 12 days ago
by
MavenDE
4steps motion issue
1
#36 opened 12 days ago
by
Rkss
wait 8 step
2
#34 opened 13 days ago
by
Lollo001lollo
ref2v lora blurs the hands and the overall quality is bad.
๐ 1
6
#33 opened 13 days ago
by
Kinduun
I tested EVERY sampler ร scheduler combo for MiniMax-H3 + LightX2V FL2V Turbo 4-step 1.0 so you DON'T need to
๐ 11
4
#32 opened 13 days ago
by
Peiyi2025
sound effects issues
๐ 1
2
#31 opened 13 days ago
by
FragmentY
best sampling settings for the 4 step ref2v?
10
#30 opened 13 days ago
by
jartoph
Help Us Test Reference-to-Video + Audio โ Share Your Images and Prompts
๐ 18
5
#29 opened 14 days ago
by
lightx2v
Could you share the expected timeline for releasing the 8โstep turbo LoRA for Ref2VA?
3
#28 opened 14 days ago
by
aronzhan
motion issues
๐ 10
6
#26 opened 15 days ago
by
kerbitaliooo
what schedulers + samplers?
๐ 2
6
#25 opened 15 days ago
by
thevegit0
Issues with nvfp4
#24 opened 15 days ago
by
Araskijo
8-step 768p variant?
๐คฏ 1
4
#23 opened 15 days ago
by
wywywywy
8 step issue with other loras
2
#22 opened 15 days ago
by
Plaguekind
Questions regarding the download and use of v1.0 LoRA (for beginners)
๐ค 3
3
#21 opened 15 days ago
by
CXK996
4-step 768p V1.0 test results
๐๐ฅ 9
15
#20 opened 15 days ago
by
lightx2v
What are the weights and samples?[minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors]
๐ 1
2
#17 opened 16 days ago
by
RedHn
8-step V1.0 test results
๐โค๏ธ 9
7
#16 opened 16 days ago
by
X-niper
Congrats and huge Thank you for the 8-step loras !
๐ฅ 6
4
#15 opened 16 days ago
by
necroryona
The Rev2Video Turbo model is currently in training. Please stay tuned
๐๐ 42
5
#14 opened 17 days ago
by
X-niper
Ref2โVA + FL2V 4โstep turbo LoRA causes artifacts on fight scenes, request for native Ref2โ4step LoRA
โ 6
#13 opened 17 days ago
by
aronzhan
minimax_h3_Ref2v_lightx2v_turbo_4step_v0.1_comfy version please??
๐ 14
#12 opened 17 days ago
by
dennismc1974
Will try the r2v version?
๐ 2
#11 opened 18 days ago
by
amazingk
Diltillation of pre-distilled model
#10 opened 18 days ago
by
drakexp
The color became dull.
5
#9 opened 19 days ago
by
sdqq1234
PEFT Training Params
2
#8 opened 19 days ago
by
QrusherZA
Preview for Ref2VA?
2
#7 opened 19 days ago
by
krigeta
Will this Lora work if I use 8 steps? And if so, should I set the Lora strength to 0.5 instead of 1?
4
#6 opened 19 days ago
by
Debich
Thanks, youโre my hero!
โค๏ธ๐ 11
2
#4 opened 20 days ago
by
sunnyboxs
Thanks and ComfyUI conversion
๐๐ฅ 34
4
#3 opened 20 days ago
by
Kijai
V0.1 is a preview version
โค๏ธ๐ฅ 26
11
#2 opened 20 days ago
by
X-niper