Software Engineering Courses (SECourses)•12mo ago

this is huge for flux

Ddxqb you normally can't, this was an experiment that required code changes and is 3x ...

dxqb•1/24/25, 11:45 PM

have you seen that?
I'd much rather train it into a LoRA

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 @Dr. Furkan Gözükara Will you be testing training on Hunyuan video? I am waiting...

Furkan Gözükara SECourses•1/25/25, 12:58 AM

yes

YYobin random thought, can you make flux loras a gguf???

Furkan Gözükara SECourses•1/25/25, 12:59 AM

you can convert them to fp8 but i dont know gguf lora

JJ2_Nedo this new Hunyuan3D 2.0 looks really interesting!

Furkan Gözükara SECourses•1/25/25, 12:59 AM

their demos are excellent

Furkan Gözükara SECourses•1/25/25, 12:59 AM

but the app doesnt produce that quality atm

Furkan Gözükara SECourses•1/25/25, 12:59 AM

authors said they will fix issues

Furkan Gözükara SECourses•1/25/25, 12:59 AM

i am following repo

VVAMP Doc would be buying the Rtx 5090?

Furkan Gözükara SECourses•1/25/25, 12:59 AM

yes waiting first day to buy hopefully

Furkan Gözükara SECourses•1/25/25, 12:59 AM

my machine almost ready

Furkan Gözükara SECourses•1/25/25, 1:00 AM

Furkan Gözükara SECourses•1/25/25, 1:00 AM

currently running 9 hours Prime 95 so far good

VVAMP And doc what do you think about NVIDIA Project DIGITS? ..Nvidia priced it at 300...

Furkan Gözükara SECourses•1/25/25, 1:00 AM

i say worthless

Ddxqb What is the state-of-the-art of creative upscaling? still SUPIR?

Furkan Gözükara SECourses•1/25/25, 1:00 AM

yes

Ddxqb SUPIR was using JuggernautXL when I last used it. Have they upgraded to newer mo...

Furkan Gözükara SECourses•1/25/25, 1:01 AM

not yet

Furkan Gözükara SECourses•1/25/25, 1:01 AM

but authors said they are working supir v2

Furkan Gözükara SECourses•1/25/25, 1:01 AM

although not sure if will be public

NN7-Blackbird how can you do FLUX with negative prompt?

Furkan Gözükara SECourses•1/25/25, 1:02 AM

you can use de-disttiled model

Furkan Gözükara SECourses•1/25/25, 1:02 AM

it has negative prompt

Furkan Gözükara SECourses•1/25/25, 1:02 AM

https://medium.com/@furkangozukara/thoroughly-experimented-with-fine-tuning-dreambooth-training-of-flux-dev-de-distill-pixelwave-13fab8c625bf

Medium

Thoroughly experimented with Fine-Tuning / DreamBooth training of F...

Thoroughly experimented with Fine-Tuning / DreamBooth training of Flux-dev-de-distill, PixelWave v03, Verus Vision and base FLUX Dev model…

MMacsalim @Dr. Furkan Gözükara doc what do you think of this node, is it acuurate?

Furkan Gözükara SECourses•1/25/25, 1:03 AM

i used them in past but with flux doesnt use anymore :d

Furkan Gözükara SECourses•1/25/25, 1:03 AM

no needed

Fabricatedgirls•1/25/25, 3:20 AM

how far do you think we are from having veo 2 quality video models at home

NN7-Blackbird yes, curiously! FLUX has great quality but something about people generated with...

wardensc2•1/25/25, 7:30 AM

It's because you dont know how to prompt with some tweak in ComfyUI, Flux is mile ahead of SD 1.5. Let me show you some images

wardensc2•1/25/25, 7:31 AM

wardensc2•1/25/25, 7:32 AM

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•1/25/25, 7:41 AM

Has anyone used a Lora on a dreambooth?

FFurkan Gözükara SECourses you can use de-disttiled model

AiInfluence•1/25/25, 8:08 AM

is it the same as Q8? in terms of quality

�𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮 Has anyone used a Lora on a dreambooth?

Joshcandle•1/25/25, 10:37 AM

What you mean?

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•1/25/25, 10:38 AM

I mean loading up a lora with a trained dreambooth instead of base flux dev

Joshcandle•1/25/25, 10:39 AM

@Furkan Gözükara SECourses Ive done dreambooth training on this Flux Checkpoint https://civitai.com/models/161068/stoiqo-newreality-flux-sd35-sdxl-sd15?modelVersionId=979329

But somehow i cant extract the lora from it, doesnt matter if i use the Full or the Pruned Model, do you know why that is?

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•1/25/25, 11:56 AM

Is multi gpu dreambooth training possible?

�

𝗡𝗶𝘁𝗶𝗻 𝗣𝗮𝗿𝗲𝘃𝗮•1/25/25, 11:56 AM

Has anyone tried training dreambooth on a 80gb card?

Timson•1/25/25, 12:16 PM

Pro tip: if you are getting low quality results when upscaling image with flux img2img always use at least 0.15-0.2 Init Image Reset To NormInit Image Reset To Norm. Upscale artefacts you see are the result of SwarmUI upscale init image to target resolution BEFORE it is VAE encoded. This works WAY better than increasing image creativity.
Took me a while to figure it out