UniFL shows HUGE Potential - Euler Smea Dyn for A1111
Vložit
- čas přidán 16. 05. 2024
- UniFL has huge portential for Stable Diffusion Models. It is faster, understands the image content better and most importantly: It looks better. Finally we are getting closer to images that look like Midjourney in Stable Diffusion. Love it! Also let's check out the new Euler Smea Dyn for A1111
#### Links from the Video ####
My Workflows on Patreon: / animated-video-102281138
uni-fl.github.io/
github.com/licyk/advanced_eul...
#### Join and Support me ####
Buy me a Coffee: www.buymeacoffee.com/oliviotu...
Join my Facebook Group: / theairevolution
Joint my Discord Group: / discord
AI Newsletter: oliviotutorials.podia.com/new...
Support me on Patreon: / sarikas - Jak na to + styl
the samplers extension works in forge too, just installed it. thanks! 🙂
Always great to see a new informative video from you. Keep going!
Interested in this new euler sampling, will try it tomorrow
Content for Automatic 1111?, you get Automatic Like!
true
Works with comfy UI as well
Another video from the legend. Much appreciated brother!
How to create wallpaper type patterns in stable diffusion? Is there a particular prompt structure or Lora that allows for this?
try "abstract background pattern, black and white and grey, wavey" in different models. I created mine in Midjourney
This does look interesting :)
oooo very nice
will this is also work on the comfy ui?
Is this better than CosXL and ELLA?
Is this available for comfyui?
What about full LORA training for A1111
Haha, I was expecting you to wave two hands with six fingers at the end.
Wonder if it will run on my Pi5?
You back with goodies on auto 1111
In my Stable Diffusion, when I want to install the extension, it tells me that it already exists... but when I see the samplers, the options shown here do not appear, but a Schedule type option appeared that has several options although they are not the same sampler names...
check if you have installed it but not activated it. or try to delete it and reinstall it (folder in extensions folder)
O哥出品 必属精品！哈哈哈
Waiting for ELLA in A1111
images generated with this model have a shallower depth of field, carefully observe the background lines compared to the original image
anybody care to explain to me what pony diffusion is? seen models for it all over civitai
Little pony vibe I think, new model base on SDXL version 😎
I have no clue how to use it. All the reference images I have seen look amazing, but mine area all garbo lol
@@ItsFinigan I have that same problem, all that score stuff in the prompts makes no sense and I've tried to look up what it is. It also doesn't use celeb names without loras, which is a big downer as I like to mix and match names to get a new, mostly consistent character.
@@ItsFinigan It took me a while too. It's really weird because I don't know exactly what I started doing differently. Make sure you're prompting using the old SD1.5 keyword/tag style. I think it was trained on booru tags. Loras seem to be pretty important for getting better results. The positive prompt should start with score_9, score_8_up, score_7_up. You can try adding score_6, score_5, score_4 to the negative.
@@Elwaves2925 I don't think it was made for photorealism, so it knows a lot animated characters instead of celebrities. The positive prompt should start with score_9, score_8_up, score_7_up. You can try adding score_6, score_5, score_4 to the negative.
still using SD1.5 nowdays?
because it's uncensored and community trainable. Other models require too much VRAM.
👋
PS: works in Forge
6:52 arthritis cramped hands.
You should talk about the best NSW models ;)
Dreambooth is irreversibly broken, did anyone notice that
At this point everyone is using kohya or onetrainer. Shame.
"Huuuuge" 🙄
First! 😂
🤮
@@FusionDeveloper 🤢
I can't believe that after all this time you still keep saying "Euler" wrong. Are you just trying to prove your ignorance?
Are you trying to start a fight on the playground? lol