@Dr. Furkan Gözükara is it better to use the sd vae when training model with dreambooth?
@Dr. Furkan Gözükara is it better to use the sd vae when training model with dreambooth?
Din extra , which model is best for SDXL ? and should i use both upscale ?
KTrial and error 
KI like these the best, but it's amatter of taste and settings
DStable Diffusion ArtAndrew
AI Image upscaler like ESRGAN is an indispensable tool to improve the quality of images generated by Stable Diffusion. In fact, it is so commonly used that
DTaken from Reddit #Upscaler
DOpenModelDB
OpenModelDB is a community driven database of AI Upscaling models. We aim to provide a better way to find and compare models than existing sources.
KAnyone good at design training?! I mean... I can train a model of my face or an object, no problem. I can train a style, like a specific artstyle, but how do I train a design... Like... If I want to make an aeroplane in the Volvo designlanguage?! How would I train that?
Fyes it fits
Fwe run it as fp16- half precision
FKIt fit, but man it took some time running a batch of images. About 8-12 minutes per image
Fyou can copy paste and append them
Fmuch faster for me
FYouTubeSECourses
Download scripts from here : https://www.patreon.com/posts/sota-very-best-90744385
Full length Twitter video here : https://twitter.com/GozukaraFurkan/status/1712665890259447831
SOTA (The Very Best) Image Captioning Models Script For Stable Diffusion And More
1 Click install and use SOTA image captioning models on your computer. Supports 8 bi...
Full length Twitter video here : https://twitter.com/GozukaraFurkan/status/1712665890259447831
SOTA (The Very Best) Image Captioning Models Script For Stable Diffusion And More
1 Click install and use SOTA image captioning models on your computer. Supports 8 bi...
Fit is still slow though it is accurate
Fi think can be used
Fnice
KYou're using VIT-L-14 in the video. Not laion/CLIP-ViT-bigG-14-laion2B
KRight?!
Fi think you need a good dataset and perhaps captions
KThe captions is easily created by CLIP interrogator and some manual labor. Question is what kind of dataset would be needed :/ All training videos (Including your fantastic videos) are of subject/object training or an art style. Not of a design
K
K200 seconds in and is has barely started interrogating :/
KIt takes forever
KI'm at 9% 430 seconds in
KJust slow
KViT-L-14/openai just takes a few seconde
K
Fi also have rtx 3090
Fwill try and let you know speed. mine working good
KThis is very strange
Fi will let you know in 30 min got to get my girl from school
Fstarting web ui
F
KWhat kind of GPU is that? 
K55.9GB
KNevermind!
KMine:
Ftakes lesser than 1 minute
F
KHow?!? 
KJust loading the model takes forever 
Fare you using batch process?
KNot for one image
F1 image loading will take time
Fi didnt count it
K
KLoading to RAM first :/
Fuse latest version
