Better AI Hands. How to Install Smea Dyn for Stable Diffusion.

Рет қаралды 14,907

Күн бұрын

Пікірлер: 64

@subsudo8641 6 ай бұрын

Since the name Euler comes from the mathematician Leonhard Euler, who was born in Switzerland in the 18th century, I can confirm as a native Swiss that it is pronounced like "Oiler." This is a well-known family name here. That should end the discussion once and for all, at least for those who read this. ;-)

@sandnerdaniel 6 ай бұрын

I guess the discussion is taking place outside of Europe. Leonhard Euler is a famous mathematician and you learn about him in primary school. Of course, you are right about the pronunciation.

@DarkoOG 6 ай бұрын

@@sandnerdaniel It's also pronounced "Oil-er" in English, people who argue against it are illiteralte.

@kneecaps2000 2 ай бұрын

Oiler. No debate required. :)

@madrooky1398 6 ай бұрын

Creating a sampler for dealing with the hand problem is just a mere band aid. The issue is simply that hands are a complex concept and have a much higher variety relative to the rest of the body. The same goes for faces in images where the face is not the focus. But hands are way more difficult. Image augmentation during training/fine tuning is a better approach on the long run. But its difficult. The amount of samples and the right balance means a shxxx ton of work. Another issue are relative low res samples. For a model trained with samples below 1mp it is very difficult to pick up the details of hands. And lets not get started about the "right" tagging.

@ginglyst 6 ай бұрын

indeed, the sampler approach to fix composition issues... I only see more compatibility issues. Like using "Lightning" models that require an "SGM Uniform" variant. Novel idea, but I'm gonna pass for this one.

@Steamrick 6 ай бұрын

Euler was a Swiss mathematician born in Basel - the German-speaking part of Switzerland. In German, the name is definitely pronounced like 'oiler' - same as Europa is pronounced 'oirohpa'.

@nebuchadnezzar916 6 ай бұрын

It is "Oiler", you're correct. I've been dealing with Euler rotations for over 20 years.

@meadow-maker 6 ай бұрын

why is it pronounced Oiler though, it make no sense, I know English is confusing but new words should follow the basic rules of letters having certain sounds and EVERYONE pronounces 'Europe either 'oorup' or 'yourup'. NObody says oirup! So what's the reason?

@Marcio2024-of2tb 6 ай бұрын

@@meadow-maker Euler is not ENGLISH... Euler was a Swiss mathematician

@Warlynn 6 ай бұрын

@@Marcio2024-of2tb that explains a lot.

@Steamrick 6 ай бұрын

@@meadow-maker Euler was a Swiss mathematician born in Basel - the German-speaking part of Switzerland. In German, the name is definitely pronounced like 'oiler' - same as Europa is pronounced 'oirohpa'. Notice the stretch on the second syllable - that's an 'oh', not a 'u'.

@skycladsquirrel 6 ай бұрын

It might be good for inpainting just for the bad limbs. Thanks for the video!

@GrocksterRox 6 ай бұрын

Great vid as always! Your examples focused on Anime, any tests against photoreal and it's impact on both SD15 and SDXL?

@fernando749845 6 ай бұрын

Exactly! Why bother with this anime crap?!!

@luisbriceno5242 6 ай бұрын

@@fernando749845 Maybe because anime models are more precise and responsive than realistic models (thanks Danbooru tags) and they are less boring too.

@southcoastinventors6583 6 ай бұрын

@@fernando749845 If you read the github notes you would have your answer.

@andyrevo8081 6 ай бұрын

Dead jokes can also be stored with your mummy.

@EvlixManiac 6 ай бұрын

@NotThatOlivia 6 ай бұрын

same sampler is in Comfyui - install it via manager as node

@KDawg5000 6 ай бұрын

Would be interesting to try things like "hand shaking" or "holding hands" or "hugging". It seems like those images always confuse the models.

@ADMNtek 6 ай бұрын

from my experience unless I use a Lora any form of body contact is ganna result in jank sometimes i get lucky but often times it is body horror.

@popular75 6 ай бұрын

It helps me,you’re great,I will try it,thanks man

@TitohPereyra 4 ай бұрын

I love automatic

@blademarketing 6 ай бұрын

This is awesome! available for comfuyi as well?

@vi6ddarkking 6 ай бұрын

Does it also work on toes?

@sebastiankamph 6 ай бұрын

Good question! If you test it, let me know

@sandnerdaniel 6 ай бұрын

Is there a scenario for Heun samplers? I always get very subpar results with Heun when testing pretty much anything. Maybe it is better for cartoons?

@drericclinic 6 ай бұрын

I got a runtime error: the size of tensor a (120) must match the size of tensor b (96) at non-singleton dimension3. when trying to use Euler SMEA Dy at the sampling method. any clue how to fix it this problem?

@SyntheticSurrealism 6 ай бұрын

No issues about how you say "Eular", but "centaur" on the other hand... ;). (in english it's sen-tore rather than ken-tore)

@NicoSeymore 6 ай бұрын

Thanks! Does this work in Forge too?

@lastword3386 6 ай бұрын

yes

@streamtabulous 5 ай бұрын

how would one add to comfy

@keenheat3335 6 ай бұрын

didn't control net already solve the hand issue ? or is this method more data efficient or have other benefit above controlnet

@sebastiankamph 6 ай бұрын

I would say that nothing has truly solved hands yet, at least no easy 1-click solutions. There are options out there to improve hands, but we're not fully there yet. With manual intervention, you can sort them pretty good, but we want to automate 😅

@DaniDani-zb4wd 6 ай бұрын

It takes so much longer to generate an image with controlnet and adetailer enable . ESPECIALLY ADETAILER. Why is it using so many resources? It takes almost 3 times longer to generate an image with adetailer on for me. I am running sd with 6bg of vram thiugh so I guess this is expected

@Brandon-x9v8k 6 ай бұрын

What if you run another pass after dpm karras using this new model?

@sebastiankamph 6 ай бұрын

Interesting. One could use adetailer and change it to the new samplers for hands only.

@popular75 6 ай бұрын

does it support SDXL?

@lomaxxbr 6 ай бұрын

I'm a brazilian and I've known since I was a kid that the name is pronounced as "Oiler".

@DMikoGee 6 ай бұрын

Doesn't seem to work with lcm

@danieldimitrovdrums 5 ай бұрын

Getting an error when using it :( - The size of tensor a (80) must match the size of tensor b (64) at non-singleton dimension 3

@frankhall401 6 ай бұрын

I just tried them all with a selection of models (1.5 and XL) using the prompt "normal women looking at their normal hands". The results were hilariously terrible.

@CoconutPete 6 ай бұрын

the reason hands are so difficult for AI is proof the technology was created by aliens

@createfun5706 3 ай бұрын

how to install Sampler: DPM++ 2M Karrass .... i have AMD graphic card ,i want to generate image to image.

@CharlesLijt 6 ай бұрын

at 6:16 ....Nah bro...The image quality is significant significantly worse than any of other samplers in this example. extremely flat, no detail what soever... I would rather go Euler A