Software Engineering Courses (SECourses)•10mo ago

@Dr. Furkan Gözükara do we have tutorial / installer for this ? https://github.com/ali-vilab/ACE_pl

@Dr. Furkan Gözükara do we have tutorial / installer for this ?

https://github.com/ali-vilab/ACE_plus

GitHub

GitHub - ali-vilab/ACE_plus

Contribute to ali-vilab/ACE_plus development by creating an account on GitHub.

AiInfluenceOP•3/8/25, 5:16 PM

Has anyone tried it ? if so , let me know please

Jjulius i finished fine tuning on the iphone style now

AiInfluenceOP•3/8/25, 5:18 PM

what token and class did you use?

AAiInfluence @Dr. Furkan Gözükara do we have tutorial / installer for this ? https://github....

Furkan Gözükara SECourses•3/8/25, 5:18 PM

not yet but definitely i need to make

Furkan Gözükara SECourses•3/8/25, 5:20 PM

mmaudio working on kaggle

Furkan Gözükara SECourses•3/8/25, 5:20 PM

really fast

MetaKan•3/8/25, 5:23 PM

hi, Is there any way in Swarm or ComfyUI to remove that overly plastic-looking texture in generations, especially in portraits?

MetaKan•3/8/25, 5:24 PM

everything looks excesively smooth, loosing reality, I need more photographic style

daroth•3/8/25, 5:25 PM

thats just Flux

MMetaKan hi, Is there any way in Swarm or ComfyUI to remove that overly plastic-looking t...

Furkan Gözükara SECourses•3/8/25, 5:26 PM

using custom trained checkpoint or lora

Furkan Gözükara SECourses•3/8/25, 5:26 PM

is solution

Furkan Gözükara SECourses•3/8/25, 5:26 PM

when we train our fine tuned models it also remove that

Ddaroth thats just Flux

Furkan Gözükara SECourses•3/8/25, 5:27 PM

true

FFurkan Gözükara SECourses when we train our fine tuned models it also remove that

MetaKan•3/8/25, 5:36 PM

I'm using one

MetaKan•3/8/25, 5:37 PM

But it's true that maybe It's not the perfect dataset

Ddaroth thats just Flux

MetaKan•3/8/25, 5:39 PM

Yep I see ... Any other model recomendations to get realistic portraits?

MMetaKan Yep I see ... Any other model recomendations to get realistic portraits?

Timson•3/8/25, 5:46 PM

you can't get better results than flux, you just need more effort with it. Consider better prompt engineering

AiInfluenceOP•3/8/25, 5:46 PM

@Timson have you tried this

https://github.com/ali-vilab/ACE_plus

GitHub

GitHub - ali-vilab/ACE_plus

Contribute to ali-vilab/ACE_plus development by creating an account on GitHub.

AAiInfluence @Timson have you tried this https://github.com/ali-vilab/ACE_plus

Timson•3/8/25, 5:48 PM

nope

TTimson nope

AiInfluenceOP•3/8/25, 5:50 PM

any tips on better prompt engineering? is there like a custom gpt out there for flux prompting ?

AAiInfluence any tips on better prompt engineering? is there like a custom gpt out there for ...

Timson•3/8/25, 5:58 PM

nope, just experiment. There is no single solution that fits all cases

Timson•3/8/25, 5:59 PM

Also img2img is a powerful tool, even if you can't get result just with prompt you can use reference images and/or inpainting

daroth•3/8/25, 6:22 PM

theres a myriad of options available. filmic looking loras, checkpoints trained for analog look, you don't need to limit yourself for base flux

Furkan Gözükara SECourses•3/8/25, 6:40 PM

Furkan Gözükara SECourses•3/8/25, 6:45 PM

SuperTurboHero•3/8/25, 6:47 PM

Guys, is it worth upgrading from rtx3060 12gb to rtx4060ti 16gb for local training?

SSuperTurboHero Guys, is it worth upgrading from rtx3060 12gb to rtx4060ti 16gb for local traini...

Furkan Gözükara SECourses•3/8/25, 6:47 PM

it will be a small improvement

Furkan Gözükara SECourses•3/8/25, 6:47 PM

4070 super 16 gb better

SuperTurboHero•3/8/25, 6:49 PM

4070 super 16gb same price as 3090 24gb :/

SuperTurboHero•3/8/25, 6:50 PM

So probably 3090 better of those 2?

SSuperTurboHero 4070 super 16gb same price as 3090 24gb :/

Furkan Gözükara SECourses•3/8/25, 6:50 PM

in that case 3090 better :d

Furkan Gözükara SECourses•3/8/25, 6:50 PM

i recommend used 3090

SuperTurboHero•3/8/25, 6:50 PM

Probably will have to go this route. Thanks

Furkan Gözükara SECourses•3/8/25, 6:51 PM

i moved my home setup to work and purchased 2nd used 3090 for it :d

Furkan Gözükara SECourses•3/8/25, 6:52 PM

dmn isnt this impressive

TTimson Also img2img is a powerful tool, even if you can't get result just with prompt y...

AiInfluenceOP•3/8/25, 7:00 PM

how do I use image2image in swarm? I haven't played with it

AAiInfluence how do I use image2image in swarm? I haven't played with it

Timson•3/8/25, 7:56 PM

it's all in the Dr's video tutorials

FFurkan Gözükara SECourses dmn isnt this impressive

Dazzastrous [4090]•3/8/25, 8:16 PM

nah no seal noise

Dazzastrous [4090]•3/8/25, 8:17 PM

so is this actually an updated improved model for mmaudio doc?

DDazzastrous [4090]so is this actually an updated improved model for mmaudio doc?

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

yes

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

i am adding more features right now

Furkan Gözükara SECourses•3/8/25, 8:26 PM

FFurkan Gözükara SECourses Click to see attachment

cyberbol•3/8/25, 8:27 PM

I love it.

FFurkan Gözükara SECourses Click to see attachment

cyberbol•3/8/25, 8:28 PM

Amazing. Can you share prompt and settings?

FFurkan Gözükara SECourses dmn isnt this impressive

cyberbol•3/8/25, 8:30 PM

The physics of water, etc. Beautiful.

Ccyberbol Amazing. Can you share prompt and settings?

Furkan Gözükara SECourses•3/8/25, 8:32 PM

i will show in tutorial how to prompt

Furkan Gözükara SECourses•3/8/25, 8:32 PM

especially video and images are easy to generate with ai

Dazzastrous [4090]•3/8/25, 8:34 PM

Ah yeh it does images now

vp•3/8/25, 8:34 PM

doc, in kohya lora training do you know if there is a difference between using 240mb clip L vs 1.6gb clip L. Is this something you have tested?

Vvp doc, in kohya lora training do you know if there is a difference between using 2...

Furkan Gözükara SECourses•3/8/25, 8:34 PM

yes i tested

Furkan Gözükara SECourses•3/8/25, 8:34 PM

i feel like 1.6gb clip a bit better

@Dr. Furkan Gözükara do we have tutorial / installer for this ? https://github.com/ali-vilab/ACE_pl

Similar Threads

Similar Threads

Similar Threads