Software Engineering Courses (SECourses)•2y ago

using Supir to 8x a 800x1700 image on my rtx 3080 with 10vram taking about 2.5 hrs to upscale if i

using Supir to 8x a 800x1700 image on my rtx 3080 with 10vram taking about 2.5 hrs to upscale
if i upgrade to like a 4090 with 24gb will it take like 20-30 mins to upscale to 8x?

FFurkan Gözükara SECourses i would compare supir low text cfg upscaled

kareem9000•4/3/24, 10:19 AM

https://www.patreon.com/posts/supir-full-1-new-99349514

Patreon

SUPIR Full Tutorial + 1 Click 12GB VRAM Windows & RunPod / Linux In...

Get more from SECourses: Tutorials, Guides, Resources, Training, MidJourney, Voice Clone, TTS, ChatGPT, GPT, LLM, Scripts by Furkan Gözükara on Patreon

kareem9000•4/3/24, 10:19 AM

so technically i would try to run the dataset by supir before i start training on dreambooth

ArtChicken•4/3/24, 10:20 AM

Is it possible to extract a lora from finetuned model with Onetrainer?

joshtech•4/3/24, 10:33 AM

@Dr. Furkan Gözükara Thanks to your great guides I now have a really great dreambooth model that reproduces the face correctly in most situations

. However I still have big problems with hands

. I need to generate a lot of images of the person and I dont have time to manually fix certain images. That's why I have to sort out about 80% of ALL images because the hands are misformed. Adetailer does not do a good job at fixing hands too. Do you have any recommendation for a workflow that either avoids hands in the image or at least fixes them sometimes? I cannot use "hands in pocket" because the images are NSFW mostly... Thanks for your ideas. You do an amazing job, really!

Jjoshtech @Dr. Furkan Gözükara Thanks to your great guides I now have a really great dream...

ArtChicken•4/3/24, 11:53 AM

try to use embeddings: they don't do miracles, but sometime prevent big mistakes (that adetailer can fix more easily); try to search specific embeddings on civitai and play around...

TTuanny using Supir to 8x a 800x1700 image on my rtx 3080 with 10vram taking about 2.5 ...

Furkan Gözükara SECourses•4/3/24, 3:06 PM

it would take few minutes

Furkan Gözükara SECourses•4/3/24, 3:07 PM

unless it uses shared vram

Furkan Gözükara SECourses•4/3/24, 3:07 PM

i can test it

Furkan Gözükara SECourses•4/3/24, 3:07 PM

tell me your final resolution that you see

Kkareem9000 so technically i would try to run the dataset by supir before i start training o...

Furkan Gözükara SECourses•4/3/24, 3:07 PM

yes

Furkan Gözükara SECourses•4/3/24, 3:07 PM

but i would compare

Furkan Gözükara SECourses•4/3/24, 3:07 PM

both original and upscaled

AArtChicken Is it possible to extract a lora from finetuned model with Onetrainer?

Furkan Gözükara SECourses•4/3/24, 3:07 PM

yes but i prefer kohya better

Jjoshtech @Dr. Furkan Gözükara Thanks to your great guides I now have a really great dream...

Furkan Gözükara SECourses•4/3/24, 3:08 PM

hands issue coming from the base model

Furkan Gözükara SECourses•4/3/24, 3:08 PM

and thanks

rafraf•4/3/24, 3:19 PM

@Dr. Furkan Gözükara in kohya, can "Epoch" be set to "0" or empty if I set Max Steps to 10000? I am having issue when i set sample every 100 steps but leave Sample every "0" epochs. Should I leave sample every x epochs? I'm struggling finding what is the accepted parameter

rafraf•4/3/24, 3:26 PM

standard_1.json4.4KB

virginia•4/3/24, 6:26 PM

@Dr. Furkan Gözükara can webui still be used?

FFurkan Gözükara SECourses both original and upscaled

kareem9000•4/3/24, 9:20 PM

Can you explain what this means? do you mean do an original supir and then an upscale supir?

AIGambino•4/4/24, 12:12 AM

one trainer won't train what am I missing

AAIGambino one trainer won't train what am I missing

DarkView•4/4/24, 12:22 AM

tried a fresh install?

Rrafraf @Dr. Furkan Gözükara in kohya, can "Epoch" be set to "0" or empty if I set Max S...

Furkan Gözükara SECourses•4/4/24, 12:30 AM

never tried

Vvirginia @Dr. Furkan Gözükara can webui still be used?

Furkan Gözükara SECourses•4/4/24, 12:30 AM

yes

Kkareem9000 Can you explain what this means? do you mean do an original supir and then an up...

Furkan Gözükara SECourses•4/4/24, 12:30 AM

i mean do 1 training with original

Furkan Gözükara SECourses•4/4/24, 12:30 AM

1 with upscaled

Furkan Gözükara SECourses•4/4/24, 12:30 AM

and compare

AAIGambino one trainer won't train what am I missing

Furkan Gözükara SECourses•4/4/24, 12:31 AM

i dont know you can ask developer on their discord

Furkan Gözükara SECourses•4/4/24, 12:31 AM

he is super active

Furkan Gözükara SECourses•4/4/24, 1:09 AM

purchased Poco X6 pro and transferring all installed apps taking huge time

Pelolisu•4/4/24, 1:32 AM

poco x6 is good? i have a poco gt x3

Pelolisu•4/4/24, 1:35 AM

i think they only change screen type in x6 haha

AIGambino•4/4/24, 1:45 AM

wheres the tier2 for onetrainer @Dr. Furkan Gözükara ?

AIGambino•4/4/24, 1:45 AM

for 24gb 3090 cards

FFurkan Gözükara SECourses purchased Poco X6 pro and transferring all installed apps taking huge time

VAMP•4/4/24, 1:54 AM

You also like performance Doc...i have redmi k50 with dimensity 8100...works good..

VAMP•4/4/24, 1:54 AM

do you also root your devices doc?

VAMP•4/4/24, 1:54 AM

mine is rooted

PPelolisu poco x6 is good? i have a poco gt x3

Furkan Gözükara SECourses•4/4/24, 2:00 AM

i have poco x3 pro

Furkan Gözükara SECourses•4/4/24, 2:00 AM

upgrading to poco x6 pro - going to give mine to my wife

Furkan Gözükara SECourses•4/4/24, 2:00 AM

it is super powerful phone

AAIGambino wheres the tier2 for onetrainer @Dr. Furkan Gözükara ?

Furkan Gözükara SECourses•4/4/24, 2:00 AM

no more tier 2 after recent updates

Furkan Gözükara SECourses•4/4/24, 2:00 AM

all are tier 1

VVAMP do you also root your devices doc?

Furkan Gözükara SECourses•4/4/24, 2:01 AM

no i dont root

Furkan Gözükara SECourses•4/4/24, 2:01 AM

in past i did in few phones but not needing it

FFurkan Gözükara SECourses upgrading to poco x6 pro - going to give mine to my wife

Pelolisu•4/4/24, 2:03 AM

i want do this too git mine to my wife haha

Pelolisu•4/4/24, 2:03 AM

they always have the leftovers haha

PPelolisu they always have the leftovers haha

Furkan Gözükara SECourses•4/4/24, 3:28 AM

well my wife prefers getting clothes so she gets them

FFurkan Gözükara SECourses well my wife prefers getting clothes so she gets them 😄

VAMP•4/4/24, 4:08 AM

women usually love two things - clothes and jewellery

Morpheus SX343•4/4/24, 5:22 AM

Hi all,
has anyone yet tried to be more accurate while training specific body parts, e.g. abs and pecs of a trained man, etc.
Or legs, waist and bottom? Obviously you cannot always have the face on the pictures if you dont always wanna take full body shots, especially for lower bodyparts.

I'm wondering what is the possible way to use closeup pictures to be more detailled.
Right now I'm not labelling / captioning my pictures.
But I dont know how the trainer is supposed to know that this bottom belongs to "ohwx man" or similarly.

Has anyone experience if this specific task can be improved by only labelling those specific images exactly with using the instance prompt? "bottom of ohwx man from behind". Or stuff like this?

FFurkan Gözükara SECourses yes but i prefer kohya better

ArtChicken•4/4/24, 9:36 AM

using Supir to 8x a 800x1700 image on my rtx 3080 with 10vram taking about 2.5 hrs to upscale if i

Similar Threads

Similar Threads

Similar Threads