Software Engineering Courses (SECourses)•10mo ago

I have a 4090 with 16 GB of dedicated VRAM, 64 GByte of RAM on a i9 13900HX and with 9:16 default re

I have a 4090 with 16 GB of dedicated VRAM, 64 GByte of RAM on a i9 13900HX and with 9:16 default resolution, 20 steps, 160 frames last night processed 8n over 10 hours. (laptop)
Maybe I am doing something wrong?

AA. T.I have a 4090 with 16 GB of dedicated VRAM, 64 GByte of RAM on a i9 13900HX and ...

Furkan Gözükara SECourses•3/8/25, 1:39 PM

160 frames

Furkan Gözükara SECourses•3/8/25, 1:39 PM

it is long

Furkan Gözükara SECourses•3/8/25, 1:39 PM

10 seconds

Aikage•3/8/25, 3:14 PM

Also laptop 4090 != Desktop 4090 afaik

MmihutJedi I left thre resolution at 480p and reduced the inference steps to half (25), and...

Aikage•3/8/25, 3:15 PM

Yeah frame dropping is a good option bc you can interpolate them back later using something like flow frames as long as you don't go too nuts

cyberbol•3/9/25, 2:19 PM

https://primecai.github.io/dsd/

Ccyberbol https://primecai.github.io/dsd/

Furkan Gözükara SECourses•3/9/25, 2:26 PM

i tested this on myself

Furkan Gözükara SECourses•3/9/25, 2:26 PM

results were horrible :d

FFurkan Gözükara SECourses results were horrible :d

cyberbol•3/9/25, 2:26 PM

ouhh, good to know

Dazzastrous [4090]•3/11/25, 7:41 AM

its not the same one doc, Ive tried it and got good results

A. T.OP•3/11/25, 5:30 PM

Hi.
I would like to ask what is the best Open Source project, that can be installed locally, to create avatars and use them for generating videos, eventually with text to speech capability (voice cloning?) and nice GUI.
Alternatively, what are free or paid version online?
Thank you in advance.

AA. T.Hi. I would like to ask what is the best Open Source project, that can be instal...

Furkan Gözükara SECourses•3/11/25, 5:46 PM

what you can do is

Furkan Gözükara SECourses•3/11/25, 5:46 PM

generate avatar with flux

Furkan Gözükara SECourses•3/11/25, 5:46 PM

train flux

Furkan Gözükara SECourses•3/11/25, 5:46 PM

generate more images of that avatar

Furkan Gözükara SECourses•3/11/25, 5:47 PM

train again

Furkan Gözükara SECourses•3/11/25, 5:47 PM

then you can consistently generate it

Furkan Gözükara SECourses•3/11/25, 5:47 PM

then you can use wan 2.1 to generate videos

Furkan Gözükara SECourses•3/11/25, 5:47 PM

for text to speech i dont have up to date tutorial but you can use elevenlabs

Zono50•3/11/25, 6:26 PM

should the wan2.1 take 40 minutes to generate a 5 second clip on a 3090? using the default settings with tiled unchecked

ZZono50 should the wan2.1 take 40 minutes to generate a 5 second clip on a 3090? using t...

Furkan Gözükara SECourses•3/11/25, 6:32 PM

yes

Furkan Gözükara SECourses•3/11/25, 6:32 PM

still we are lacking teacache

Furkan Gözükara SECourses•3/11/25, 6:32 PM

hopefully it will half the duration

art-ai_usa-girl•3/12/25, 10:09 PM

https://huggingface.co/spaces/Skywork/skyreels-a1-talking-head

Skyreels A1 Talking Head - a Hugging Face Space by Skywork

Aart-ai_usa-girl https://huggingface.co/spaces/Skywork/skyreels-a1-talking-head

Furkan Gözükara SECourses•3/12/25, 10:15 PM

lets try

Furkan Gözükara SECourses•3/12/25, 10:20 PM

looks even weaker than v express

Furkan Gözükara SECourses•3/12/25, 10:20 PM

Furkan Gözükara SECourses•3/12/25, 10:21 PM

and gave only 4 second output lol

Furkan Gözükara SECourses•3/12/25, 10:22 PM

generating v express

FFurkan Gözükara SECourses and gave only 4 second output lol

art-ai_usa-girl•3/12/25, 10:43 PM

yeah only 4 second

Aart-ai_usa-girl yeah only 4 second

Furkan Gözükara SECourses•3/12/25, 10:49 PM

github has long

Furkan Gözükara SECourses•3/12/25, 10:49 PM

i may make installer

FFurkan Gözükara SECourses i may make installer

art-ai_usa-girl•3/13/25, 1:10 AM

thank you ,,,, you are the best

woni•3/13/25, 8:49 AM

guys does stable diffusion in google collab still working?

art-ai_usa-girl•3/13/25, 10:09 PM

has anybody tested this ? https://github.com/NJU-PCALab/STAR

GitHub

GitHub - NJU-PCALab/STAR: STAR: Spatial-Temporal Augmentation with ...

STAR: Spatial-Temporal Augmentation with Text-to-Video Models for Real-World Video Super-Resolution - NJU-PCALab/STAR

Aart-ai_usa-girl has anybody tested this ? https://github.com/NJU-PCALab/STAR

Furkan Gözükara SECourses•3/13/25, 10:13 PM

i plan to make app for this

Inoshi•3/14/25, 5:22 PM

hi ! there is a text to voice model Sesame ai released https://huggingface.co/sesame are you going to create something around this ?

sesame (Sesame)

IInoshi hi ! there is a text to voice model Sesame ai released https://huggingface.co/se...

Furkan Gözükara SECourses•3/14/25, 10:39 PM

yes i plan to but how good it is? i saw today

Inoshi•3/14/25, 11:10 PM

i am trying to figure out how to control the voice better, but in terms of cloning it sounds pretty decent. i havent really tried other text to speech models though

Furkan Gözükara SECourses•3/15/25, 10:56 AM

@Capybara you can open your thread here. more fitting section