Software Engineering Courses (SECourses)•15mo ago

@Dr. Furkan Gözükara , I trained my own LoRA Flux with the Flux FP8 model, and I want to generate im

@Dr. Furkan Gözükara , I trained my own LoRA Flux with the Flux FP8 model, and I want to generate images with my LoRA for free on Kaggle. Before I subscribe to your Patreon, I wanted to ask if your Patreon post about "Free Kaggle Account Notebook for SwarmUI with Stable Diffusion 3 and Dual T4 GPU support" , can use FP8 LoRAs ?. I previously tried using Forge , but the images result were blurry and pixelated.

FFalan @Dr. Furkan Gözükara , I trained my own LoRA Flux with the Flux FP8 model, and I...

Furkan Gözükara SECourses•10/4/24, 7:21 PM

yes it can use

Furkan Gözükara SECourses•10/4/24, 7:21 PM

but cant train

Furkan Gözükara SECourses•10/4/24, 7:22 PM

blurry result happens due to cfg not being 1

Furkan Gözükara SECourses•10/4/24, 7:22 PM

we use latest swarmui and comfyui as backend

Furkan Gözükara SECourses•10/4/24, 7:22 PM

works perfect

PPixelMonkey So basically anything below 1024x1024 makes it better i feel

NeonNinjaAstro•10/4/24, 8:34 PM

been getting good results sometimes by using Flux at

512*512

512*512

or even

384*384

384*384

in the first pass

NeonNinjaAstro•10/4/24, 8:35 PM

the VAE is so much better than SDXL so you can do more passes

NNeonNinjaAstro been getting good results sometimes by using Flux at ``512*512`` or even ``384*3...

PixelMonkey•10/4/24, 8:42 PM

How do you do mutliple passes? Are you running it in Comfy Ui? I am still on Web Ui Forge

NeonNinjaAstro•10/4/24, 8:43 PM

yeah in Comfy, but you can do this in Forge as well

NeonNinjaAstro•10/4/24, 8:44 PM

what you need to do is take the result of the first pass

NeonNinjaAstro•10/4/24, 8:44 PM

upscale it using an external upscaler

NeonNinjaAstro•10/4/24, 8:44 PM

and then do the standard img-to-img in Forge using a denoise % less than 1

IWishIWereGood•10/4/24, 9:17 PM

Hi everyone, should I use regularisation images when training a lora on sdxl?

IIWishIWereGood Hi everyone, should I use regularisation images when training a lora on sdxl?

Furkan Gözükara SECourses•10/4/24, 9:17 PM

1024x1024

Furkan Gözükara SECourses•10/4/24, 9:18 PM

testing flux rare tokens

Furkan Gözükara SECourses•10/4/24, 9:19 PM

single words makes such good images

Furkan Gözükara SECourses•10/4/24, 9:19 PM

DDazzastrous [4090]Meshy

xlads•10/4/24, 9:21 PM

Came out really low quality with the details on my face, maybe I’ll try with another image

Ccrystalwizard what you want, really, is to be able to download the mesh and use it in a video ...

xlads•10/4/24, 9:21 PM

Yep!

xlads•10/4/24, 9:24 PM

It would be so cool if we could generate videos with our models using text to video. Is there anything out like this? @Dr. Furkan Gözükara

Xxlads It would be so cool if we could generate videos with our models using text to vi...

crystalwizard•10/4/24, 9:28 PM

there are lots of text to video AIs out there - and meta just annouced this https://ai.meta.com/research/movie-gen/

AI at Meta

Meta Movie Gen - the most advanced media foundation AI models

Meta Movie Gen is our latest research breakthrough that allows you to use simple text inputs to create videos and sounds, edit existing videos or transform your personal image into a unique video.

Xxlads It would be so cool if we could generate videos with our models using text to vi...

Furkan Gözükara SECourses•10/4/24, 11:18 PM

nothing good

Ccrystalwizard there are lots of text to video AIs out there - and meta just annouced this http...

Furkan Gözükara SECourses•10/4/24, 11:18 PM

i am about to publish a video for it

Ccrystalwizard there are lots of text to video AIs out there - and meta just annouced this http...

xlads•10/4/24, 11:21 PM

Yes, I heard about this, but we’re unable to use our own trained models

FFurkan Gözükara SECourses i am about to publish a video for it

crystalwizard•10/4/24, 11:21 PM

you might want to wait till it's more than a research paper

Xxlads Yes, I heard about this, but we’re unable to use our own trained models

crystalwizard•10/4/24, 11:22 PM

yes, well - you could use them with animate diff or live portrait

Ccrystalwizard yes, well - you could use them with animate diff or live portrait

xlads•10/4/24, 11:24 PM

Hmmmm need to look into animate diff

Xxlads Hmmmm need to look into animate diff

crystalwizard•10/4/24, 11:25 PM

https://animatediff.github.io/

Ccrystalwizard https://animatediff.github.io/

xlads•10/4/24, 11:26 PM

Sent you a friend request

Ccrystalwizard https://animatediff.github.io/

xlads•10/4/24, 11:26 PM

Thanks will check it in a few minutes

Xxlads Yes, I heard about this, but we’re unable to use our own trained models

Furkan Gözükara SECourses•10/4/24, 11:54 PM

yes nothing published yet

Furkan Gözükara SECourses•10/4/24, 11:54 PM

https://www.youtube.com/watch?v=hnkuuqJVGK0

YouTubeSECourses

Meta Movie Gen: The Future of Filmmaking is Here

Movie Gen sets a new standard for immersive AI content. Meta announced the Meta Movie Gen just today which is going to change the Cinema, Video and Animation industries forever. This will shock the industry workers.

Paper : https://ai.meta.com/static-resource/movie-gen-research-paper

How Meta Movie Gen could usher in a new AI-enabled era for c...

CNOC•10/5/24, 1:42 AM

@Dr. Furkan Gözükara Have you had any success with the batch 7 training? If the config is ready but no video yet would it be possible to receive the config file to try?

CCNOC @Dr. Furkan Gözükara Have you had any success with the batch 7 training? If the ...

Furkan Gözükara SECourses•10/5/24, 2:12 AM

yes it is working great

Furkan Gözükara SECourses•10/5/24, 2:12 AM

but i still didnt finalize

Furkan Gözükara SECourses•10/5/24, 2:12 AM

let me put into zip file

CCNOC @Dr. Furkan Gözükara Have you had any success with the batch 7 training? If the ...

Furkan Gözükara SECourses•10/5/24, 2:16 AM

https://www.patreon.com/posts/112099700

Patreon

Kohya FLUX Fine Tuning (Full Checkpoints) Training Full Tutorial Fo...

Get more from SECourses: FLUX, Tutorials, Guides, Resources, Training, Scripts on Patreon

OinkSauce•10/5/24, 2:45 AM

Just a quick question on Supir upscaling: are there a collection of presets that match up with some of the tests run? I'm not seeing exactly how to configure this to match up with the grid tests run on the SECourses channel. Is there a go-to configuration that I can start with and experiment from there?

I'm likely most interested in upscaling images generated in Flux and likely 2x to 4x in resolution