Dall-E, Midjourney, Stable Diffusion (etc) - who's playing, and with which?
Dall-E, Midjourney, Stable Diffusion (etc) - who's playing, and with which?
- Started
- Last post
- 824 Responses
- grafician0
- https://scribblediff…Nairn
- Ah yes forgot the link
Thanks @Nairngrafician
- grafician2
- Fixed it:
https://i.imgur.com/…palimpsest - @pali LOL Season 2?grafician
- Fixed it:
- yuekit5
- Love the melting Apple version.utopian
- these are good :)sted
- Some more here...
https://imgur.com/a/…yuekit - Fascinating how it approximates all the shadows/reflections, reflection of the right column in the first pic is not accuratedrgs
- Yeah it's simultaneously very detailed but also messes things up sometimes.yuekit
- PhanLo4
- Ladyboyutopian
- :-)PhanLo
- Amazing...what happens if you feed it a very detailed drawing I wonder?yuekit
- awesome! good application of AIKrassy
- Cool - thanks for sharing thatstoplying
- how come I never get such results when doing img2img with SD? this is some insane quality. I only get really crappy distorted images.NBQ00
- oh wait, "Scribble"Diffusion. Seems something elseNBQ00
- think you're doing sthg wrong nbq. if you do img2img at strength .5 you need twice as many steps and you may need to drop the guidance scale lowerkingsteven
- sampler to ddim (klms garbles sometimes)kingsteven
- maybe try invoke unified canvas, scaling the in-fill region with a mask is incredibly powerful (for example you can redraw a 128x128 hand at 512x512 and selectkingsteven
- the best from a batch)kingsteven
- lovelyneverscared
- yuekit0
Are there any good articles on how Midjourney works in terms of the mechanics of rendering the images?
- it rearranges pixel noise till it's happy.
diffusion modeling is what they call it.uan - this article is on how gpt (language part) works:
https://writings.ste…uan - So I wonder what then explains it's inability to render only some things? Some people and characters are dead on, others it misses the mark completely.yuekit
- I also noticed after using it a while that the same stock models and scene compositions seem to show up.yuekit
- when you get a few months in and you start to recognise the facial features of prominent (circa 2019) celebrities in your renders... like looking at the matrix:kingsteven
- "theres Benedict Cumberbatch"... "theres Don Rickles"kingsteven
- with SD you notice with text2img how even the cropping of training data to 1:1 influences the layout so you have that to deal with before realising thatkingsteven
- layout in art is overwhelmingly generic and text2img diffusion only really does generic well.kingsteven
- its good to have an eye for how it works but i maintain that typing text prompts in to MJ is a bit of a fad. guys using SD in production are effectivelykingsteven
- creating their own MJs (custom embeddings and interpretation) and then using guided images, inpainting. alongside several AI processes in a workflowkingsteven
- it really requires that bottom up understanding to lessen the randomness when you have an image in mind you want to createkingsteven
- otherwise you become a little goblin man generating 1000 images to pick the one thats randomly 'correct' there's no merit in it long term.kingsteven
- humans are rubbish at identifying randomness and exponentially and images from latents is exponentially random. there's not much differencekingsteven
- in sharing a MJ image and numbers guy sharing every permeation of pixels in a 9 pixel cube. just more pixels.kingsteven
- technically the most impressive MJ image would be a prompt+seed that pulls a high resolution artstation image from the training data and fixes the handskingsteven
- but if i wanted to do that, i could use my AI workflow in SD for fixing hands. and as phanlo below, if you want a layout - sketch it in, if you want a likenesskingsteven
- Thanks that makes sense, are there any others than Stable Diffusion you think are good?yuekit
- train a model. the more you dig the more you find the tools are out there to overcome the limitations reliably.kingsteven
- the big SD platforms are adding workflow editors which is a big deal because staged processing can automate the removal of almost all the identifiable traitskingsteven
- I found this article that explains how Stable Diffusion works with illustrations...
https://jalammar.git…yuekit - The part I was most curious about is how SD and other AIs come up with the idea for the layout of the image in the first place. How do they decide which thingyuekit
- to put in the foreground, what to include and what to leave out etc.yuekit
- https://jalammar.git…yuekit
- The article doesn't actually go into a great deal of detail about this but it must be happening during the "Conditioning" phase in that diagram above.yuekit
- https://www.youtube.…
https://www.youtube.…
some Computerphile explanations I remembered watchinguan - and about the composition part...I remember watching a clip from about a year ago, where they said it has composition rules (from data) built in.uan
- I think they use something like the google image likes or something similar to create those models about pleasing compositions and they are built in.uan
- but you can override them...the description you use to generate the image you want can override those 'default' compositions.uan
- i think what you're looking for in that conditioning phase is the aesthetic scoring in the laion datasets. an AI trained on human responses to images has ratedkingsteven
- every image. for example SD favours watercolours because (along with other conditioning) the aesthetic AI that tagged the training data loves watercolours.kingsteven
- maybe read about markov chains and GANS for an understanding of diffusion. i think a lot of these articles assume some understanding of AIkingsteven
- it rearranges pixel noise till it's happy.
- sted4
- We've been expecting you.
Welcome home.palimpsest - niceutopian
- We've been expecting you.
- jonny_quest_lives-2
- What's going on here?monNom
- Most lawyers would say if you are currently involved in a lawsuit with Getty about copyrights and how u incorporate images in your training datajonny_quest_lives
- It might be a good idea to lay low and avoid posting a video of your new blender plug in with a "trending on artstation" value parameterjonny_quest_lives
- As it raises questions as to how that parameter is referenced by tge software... i.e a real time? or snapshot of artstation trending within the week?jonny_quest_lives
- PhanLo4
- Apologies I was only able to draw with the trackpadPhanLo
- Lol https://i.imgur.com/…PhanLo
- I did actually draw a graffiti outline that came out ok https://i.imgur.com/…PhanLo
- Haha, this is mad. Are these created on Midjourney? I still haven't played much with itIanbolton
- hehe, these are on https://scribblediff… Ian, you doodle out a thing and then it creates the image on a prompt. Pretty funPhanLo
- It's apparently a slightly different way of making images compared to Stable Diffusion, so will be interesting to see where it goes.PhanLo
- it is stable diffusion, it just has its conditions set to use edge detection (and a more aesthetically tweaked model)kingsteven
- lol these are greatutopian
- shapesalad0
During an artists livestream, a viewer screenshot the almost finished artwork, ran it through an AI to finish it, posted it on Twitter, then when the artist finished their artwork, the person with the AI art claimed the artist had stolen their artwork and copied it.
- NBQ000
Can somebody point me to the correct github installation link of local Stable Diffusion GUI for Mac?
I googled tons but for some reason I can no longer find it.
Only the InvokeAI version or DiffusionBee.
Any idea?
- Looking for that GUI version that is shown in the video further down posted by ShenanigansTVNBQ00
- automatic1111kingsteven
- think so anyway, he's using darkmode theme https://github.com/A…kingsteven
- there's no installer for automatic1111 on mac any more, but it's just a couple of lines in consolekingsteven
- ThxNBQ00
- _niko5
It's cool to play around with this new tech but it feels a little Brett Bash-y to profit from it or to present yourself as an illustrator/artist/designer
- no using dictionaries either, you hack writers.imbecile
- If you work on advertising it's ok to rip off artists. Fuck them.NBQ00
- So what's the difference when "illustrators, artists and designers" borrow and steal ideas from: Behance, Dribbble, Pinterest, ArtStation, etc?utopian
- Everything is a Remix
https://www.everythi…utopian - if we use Hydro74 as an example, he's got a cool established style that he's honed over the years, if he starts using AI to further his art then all the power_niko
- ...to him, it's just an extension and progression of him as an artist, but if I start asking AI to make art and illustrations in his style and claim it as my ow_niko
- ...own and profit from it and start flexing like I'm special and a genius then I think that's where the problem lays._niko
- Haven't people already done that before AI?palimpsest
- watch out hydro, i'm coming for your jerb! https://i.imgur.com/…imbecile
- Hydro Who?
Get a Brain...Morans Get Jerbs!
https://i.imgur.com/…utopian - Yeah it creates some ethical dilemmas...you don't feel total ownership of the stuff it cranks out. But I think artists will just need to take it further thanyuekit
- everyone else. Also presenting a coherent vision or narrative is actually kind of difficult with these tools. It's good at generating random things that areyuekit
- impressive but good luck trying to get it to implement your exact idea.yuekit