Software Engineering Courses (SECourses)•11mo ago

That high resolution images tiling into 1024x chunks is working really well I have to say. Running t

That high resolution images tiling into 1024x chunks is working really well I have to say. Running the fine tune on a larger more diverse dataset.

..tarkan That high resolution images tiling into 1024x chunks is working really well I ha...

AiInfluence•2/4/25, 9:50 PM

that's cool how did you do it ? put me on

..tarkan That high resolution images tiling into 1024x chunks is working really well I ha...

AiInfluence•2/4/25, 9:51 PM

is this 1024x1024 quality or x2?

.tarkanOP•2/4/25, 9:51 PM

Thanks. This is upscaled to 4k.

.tarkanOP•2/4/25, 9:52 PM

So 4x upscale in total.

.tarkanOP•2/4/25, 9:52 PM

Also it's the Flux de-distilled model. 2x slower but also 2x better quality.

.tarkanOP•2/4/25, 9:53 PM

Will share my method and the needed pyhton scripts on reddit soon and will post it here too.

.tarkanOP•2/4/25, 9:53 PM

No detail is escaping it.

AAiInfluence @Furkan Gözükara SECourses SOS 🚨 https://www.instagram.com/p/DFqmMs_IKyW/?img_...

Furkan Gözükara SECourses•2/4/25, 10:31 PM

i published a video of it

..tarkan Thanks. This is upscaled to 4k.

AiInfluence•2/4/25, 10:33 PM

what upscale method ? supir ?

FFurkan Gözükara SECourses i published a video of it

AiInfluence•2/4/25, 10:33 PM

where

AAiInfluence where

Furkan Gözükara SECourses•2/4/25, 10:33 PM

https://youtu.be/K7d3c8zLqpM?si=qFay_AhD9WBjeFtK

YouTubeSECourses

OmniHuman-1 Is The Ultimate Level of Generating AI Videos from Imag...

OmniHuman will shake the AI world forever if get published and works as they claimed. This is literally the ultimate video model announced so far. It is not published yet but I will be following it very closely. I hope that they will publish models publicly so we can use locally hopefully. As soon as published I plan to make APP for this model h...

Furkan Gözükara SECourses•2/4/25, 10:33 PM

but no code or model yet

..tarkan That high resolution images tiling into 1024x chunks is working really well I ha...

Furkan Gözükara SECourses•2/4/25, 10:33 PM

nice

AiInfluence•2/4/25, 10:40 PM

I think that china will be the next super power

AiInfluence•2/4/25, 10:40 PM

they are dropping "bombs" in the next revolutionary field

AAiInfluence they are dropping "bombs" in the next revolutionary field

Furkan Gözükara SECourses•2/4/25, 10:40 PM

100%

Furkan Gözükara SECourses•2/4/25, 10:40 PM

i hope they bring GPUs

AiInfluence•2/4/25, 10:41 PM

Also @Furkan Gözükara SECourses your research has changed my life in a way I want to thank you for everything

AiInfluence•2/4/25, 10:41 PM

Keep going this community is awesome

AiInfluence•2/4/25, 10:41 PM

Idk how I didn't find it until September

Furkan Gözükara SECourses•2/4/25, 10:59 PM

if i missed any question message me again

.tarkanOP•2/4/25, 11:03 PM

..tarkan Click to see attachment

Furkan Gözükara SECourses•2/4/25, 11:06 PM

nice quality

.tarkanOP•2/4/25, 11:07 PM

@Furkan Gözükara SECourses thank you!

.tarkanOP•2/4/25, 11:07 PM

Feel like I want to train a large fine tune with this method.

.tarkanOP•2/4/25, 11:27 PM

@Furkan Gözükara SECourses thank you for the advice and the amazing configs. Wouldn't have gotten this result without them!

FFurkan Gözükara SECourses 100%

sergio_gonzo•2/4/25, 11:27 PM

Dr furkan the last model that make china work pretty well woth AMD do you think at least can be models that work good in any GPU brand

.tarkanOP•2/4/25, 11:49 PM

Few more outputs. Still first epoch only.

Ssergio_gonzo Dr furkan the last model that make china work pretty well woth AMD do you think ...

Furkan Gözükara SECourses•2/5/25, 12:36 AM

i wish AMD was paying some more attention

..tarkan Few more outputs. Still first epoch only.

Furkan Gözükara SECourses•2/5/25, 12:36 AM

nice

Furkan Gözükara SECourses•2/5/25, 12:49 AM

working on VisoMaster auto installer

Darth Teethius•2/5/25, 12:55 AM

im trying to use the deepface similiarity app, but i'm getting this error message. Anyone know what this is?

Darth Teethius•2/5/25, 12:56 AM

DDarth Teethius Click to see attachment

Furkan Gözükara SECourses•2/5/25, 1:04 AM

yes

Furkan Gözükara SECourses•2/5/25, 1:04 AM

cuda and cudnn installation error

Furkan Gözükara SECourses•2/5/25, 1:05 AM

please install cuda 12.4 and cudnn 9.4 or 9.7

Furkan Gözükara SECourses•2/5/25, 1:05 AM

nice the project working

Furkan Gözükara SECourses•2/5/25, 1:05 AM

hopefully will publish tomorrow after testing on massed compute too

Darth Teethius•2/5/25, 1:11 AM

I have 12.8 installed. Downgrade?

DDarth Teethius I have 12.8 installed. Downgrade?

Furkan Gözükara SECourses•2/5/25, 1:17 AM

yep dont use cuda 12.8 yet

Furkan Gözükara SECourses•2/5/25, 1:17 AM

use 12.4 atm

vp•2/5/25, 1:54 AM

im trying to get caught up on stuff

vp•2/5/25, 1:54 AM

are illustrious models the same as sdxl models?

fab•2/5/25, 6:54 AM

Hi everyone, and hello @Dr. Furkan Gözükara . Thank you so much for all the work and research you share, it's amazing!
Can I ask your advice on what's the best way to train a LORA for SDXL on cloud as of today (best quality possible)?