Hugging Face's logo

lightx2v
/
Minimax-h3-Turbo

Image-to-Video
Diffusers
English
Chinese
t2v
i2v
r2v
Model card Files
xet
 Community
35

does this works with pruned int 8 Ref2va model too?

#1
by shadowmonarchforreal - opened
Discussion
shadowmonarchforreal
•
edited 8 days ago

Title.

Or a seperate version is in training for ref2va?

Thanks for you contribution

shadowmonarchforreal changed discussion title from Mdoes this works with pruned int 8 Ref2va model too? to does this works with pruned int 8 Ref2va model too?
k3soft

I think it should not work, waiting for ref2va
X-niper

@shadowmonarchforreal @k3soft
Hi,
It's distilled on the FL2V(T2V) base, we will release R2V-turbo later. Please stay tuned
CXK996

@shadowmonarchforreal @k3soft
Hi,
It's distilled on the FL2V(T2V) base, we will release R2V-turbo later. Please stay tuned

Can this LoRA be used with a standard LoRA loader (model-only)? What is the recommended strength setting? Should the number of steps also be 6–8?
CXK996

@shadowmonarchforreal @k3soft
Hi,
It's distilled on the FL2V(T2V) base, we will release R2V-turbo later. Please stay tuned

Does it work on the pruned version?Thank you for your answer.
shadowmonarchforreal

@shadowmonarchforreal @k3soft
Hi,
It's distilled on the FL2V(T2V) base, we will release R2V-turbo later. Please stay tuned

Does it work on the pruned version?Thank you for your answer.

Hey kijay has released comfy ui compatible version use that using normal lora loader.

https://huggingface.co/Kijai/MiniMax-H3_comfy
HarunAkgun
•
edited 8 days ago

Thank you guys! It works with R2V only issue is the audio... what should be the shift values i wonder?

Edit - Wanted to add, except from audio, it works way better for r2v than the other turbo alternatives.
mkay76

Yeah, i am also waiting for rev2va, that's the most usable workflow where you can plug in actors and all kinds of references
Really needed! Thanks
shadowmonarchforreal changed discussion status to closed

Sign up or log in to comment