More About SDXL Lightning + Presets For Fooocus
ฝัง
- เผยแพร่เมื่อ 22 ก.ค. 2024
- Easily Make Your Own Presets For SDXL Lightning To Use In Fooocus.
Your favorite and most user friendly stable diffusion tool, Fooocus, just got faster with SDXL Lightning. Now with more checkpoints available we can create presets to really maximize our efficiency. Find the best settings to get going and see that SDXL Lightning stands on its own. With minimal steps and a low CFG we can still create amazing images in a fraction of the time that it takes a standard Stable Diffusion Checkpoint model.
Lightning LoRAs Download: huggingface.co/ByteDance/SDXL...
SDXL Lightning/Turbo/LCM Comparison: anotherjesse.com/posts/sdxl-l...
Envy Zoom Slider XL: civitai.com/models/213307/env...
----
00:00 Intro
00:18 Overview
00:45 Lightning LoRAs
02:32 Turbo + Lightning
03:45 Lightning Checkpoints
05:49 Preset Creations
09:44 Examples
---- - วิทยาศาสตร์และเทคโนโลยี
Lightning LoRAs Download: huggingface.co/ByteDance/SDXL-Lightning/tree/main
The LoRA I Used if interested, Envy Zoom Slider XL: civitai.com/models/213307/envy-zoom-slider-xl-01
Like your Fooocus videos. Awesome work. Thanks for sharing.
@@a.d.8083 No problem, thanks for watching! 😁
Fooocus would be amazing if it had an animation section. Thanks for the easy to understand video sir.
Thanks mate! the zoom lora is a great!
No Problem! I was surprised how well it worked myself, had to share.
Очень полезная информация. Спасибо! ))
Good content 😀
Thanks 😁
Awesome tutorial like always. Thank you!
Only one questions though: How can I enable the "Filter" in CiVitAI page ? I don't have it in my main page. I saw some others have it and refer to it while explaining but I don't have it, I will be very thankful if you can help me with that.
You are welcome!
Hmm, haven't heard of the filter missing. Make sure when you get to the main page, click on the "model" tab right next to the "home" tab in the upper left corner. then the filter option should show up. or go here civitai.com/models
@@JumpIntoAI Thank you very much, yep that's it.
Hi! Little bit off topic but I really need some help with this. I have several real photos taken by me, and I inpaint women on the picture. The woman looks fine, very detailed, but compared to the base picture, she look way too detailed and she really stands out from the pic ( i guess because of higher resolution than the base picture). So like is it possible to get the woman and the base picture more even and more equal to each other??? Is this what refiner would do?
Replied to you in your other comment😉.
So as I see and understand, only Loras from SDXL Lightning website are worth downloading and not SDXL Lightning checkpoints themselves? Better to have Juggernaut version SDXL lightning than them, right? And btw. what are "unet.safetensors"?
Yes, the LoRAs are worth it for models that don't have a lightning version. But the checkpoints on there are just the base SDXL model. Not needed now with Juggernaut/Dreamshaper/etc having their own lightning version.
How I upload modal to fooocus colab ???
I have a question - how do I add these model to my Fooocus?
The lightning Lora is now apart of Fooocus thanks to the recent 2.3.0 update (its the lighting setting in performance)
all other models go into your fooocus\models\checkpoints folder. They must be SDXL to use as a base model. (sd 1.5 can be used as refiners)
Is there any alternative of Google colab for longer run time and space
There are many services now but I can't speak personally about them, I do most things in house. But thinkdiffusion and runpod get suggested often.
@@JumpIntoAIIs Google Colab free to run Fooocus there and enjoy fast speeds?
Where did you learn all this
Like most really, soaking in all the information i can from where ever I can the past year, and hundreds of hours of trial and error.
Anybody figured out how to update the presets if its a pinokio install?
fyi I just named the json one of the presets already there so i could get to it from the pinikio dashboard (in my case anime since i never use that preset)
@@williamdownes2032 Nice, thanks for the info. 👍
Warning... (at least in pinokio) my anime preset got replaced during an auto update (so save a copy)
Is sd 3 is free?
Unless Stability AI change their stance suddenly, yes it should be free to use like the others when it releases.
Help for best faceswap with foocus please, new video for us for faceswap only, best settings we want it please
I will be working on another Consistency video involving face swap. But no promises about best setting since face swap in fooocus is far from perfect.
Me and my friends are eagerly waiting for you to make a video. and we love you@@JumpIntoAI
@@JumpIntoAIIs there a way to achieve it with better results somewhere outside Fooocus?
@@hmm-m Just in general really, faceswap can be finicky in anything. But using Comfy and combination of ip-adapter-faceid and reactor (multiple passes) can get better results. But even then not perfect.
How i can cont@ct you