This blog post is a collection of artist studies done using Disco Diffusion put together (sourced from Twitter) in one place to help you understand what kind of result you would get using a specific artists name.
This list keeps growing, what started off as a great idea by the creators and contributors received lot of praise on social media so this list is updated regularly. Make sure you bookmark and share with fellow AI Artists.
Currently stands at 662 artists
Contributors
This study was done by several AI artists on Twitter. It was started by @proximasan and @sureailabs and contributed by @EErratica and @KyrickYoung. So if you are into AI Art – which I assume you are which is why you are reading this make sure you follow them on Twitter where they share their wonderful artworks.
Each of the images contains the creator name in the caption as its credit to them that this post is possible.
This post is based on the artists listed in this database created and maintained by the above contributors and can be viewed here. I only include artists that are Represented: Yes or Unsure and are artists not styles. If you are after styles then check out my other two posts: Anything punk and Disco Diffusion Modifiers (styles)
Prompts
For this study the contributors used two main prompts for the AI to generate a series of comparable images:
a beautiful painting of a waterlily pond by <artist name>, Trending on artstation.
a beautiful painting of a building in a serene landscape by <artist name>, Trending on artstation.
The prompt shows some wonderful results based on the artists style known to the model.
Updates
20th March 2022 – re-created the post, using some automation so I can speed up the adding of the artists.
6th April 2022 – added new sets of artists to the post.
20th April 2022 – added 98 new artists to the post
31st May 2022 – added 102 new artists to the post
14th July 2022 – New Gallery Mode available – new capabilities: search and filter multiple artists. Over the next few weeks all artist studies will be transitioned into the gallery mode.
Artist Gallery
You can now search for the artist in the image gallery below or explore in a gallery browser by clicking on the image. All images will move into this gallery to make this page load faster and finding artists easier.
If you'd like to support our site please consider buying us a Ko-fi, grab a product or subscribe. Need a faster GPU, get access to fastest GPUs for less than $1 per hour with RunPod.io
47 Comments
Jason
on 3 March 2022 at 10:19 AM
Very cool. Thanks for sharing this. I use my own hacked version of Disco that works from the command line and have thrown together a quick shell script that can run any prompt on all of these artists. Already blown away by the results!
These are great. Thank you! Can I just ask, Do you use a programe to make the pictures bigger while keeping the detail? I want to do a print (30 by 18 inches) But it looses a bit of detail, and just wanted to know if there is a way round this?
With my art, as soon as its done, i use ESRgain, then in photoshop use blow up 3, but still loses a bit of detail. Is there a programe you use for this? Or is this the best I will get it?
Thanks for your comments. I have used ESRGain in the past but its not as good as Topaz Gigapixel AI, which also uses machine learning to increase the size while preserving the details. I try out all the different models it has for the size I want and stick with the one that produces the correct details when upscaling
Friedensreich Hundertwasser he was an Austrian visual artist and architect 1920-2000
Beatrix Potter⊠she was an English writer and illustrator, she was best known for her childrenâs books
Ădouard Riou⊠he was a French illustrator 1833 â 1900
Really helpful, thanks for sharing this.
I got a different result by setting on Rebecca Guay, more floating strokes and more transparencies. I probably set differently, maybe less Steps or a lower Clip_guidance_scale than yours, who knows.
I’m curious to see if you add any new artists, there are some I really like that are missing here? đ
This is a farce. That youâre all using the work of living artists to inspire use this tech is extremely harmful to those artists, and are encroaching onto their livelihoods. I could see MANY lawsuits on this, especially as this is done without their consent. Disgusting. Take this website down
Yes, by this standard all teaching art should also be banned and forbidden. Because when teaching art, we never got permission from Michaelangelo, Leonado, Van Gough etc.
There is an Eiffel Tower in the generated images, this is also the case on LAION400. This artist has never put an Eiffel Tower in these images. Moreover, the generated images do not correspond at all to the style of the artist. It is probable that other artists in this list have the same problem ?
I found some Manga/Movie/Game “Artists” that I pretty like.
HIDEAKI ANNO (for Evangelion)
Katsuhiro Otomo (for Akira)
Miyazaki Hidetaka (for Dark Souls)
Togashi Yoshihiro (for )
This is absolutely amazing, I’m blown away. The explosion of new creative potential this enables is unbelievable, styles and techniques lost to time and practiced by few alive can be generated, this means new art coming out that draws from almost the entirety of art history, just mind blowing to think about, there are often artists whose styles and aesthetics you love and admire but they are long gone and no one devotes the time to study and create art in their style – but AI can get you an amazingly good approximation its better than nothing and the results really speak for themselves here…
A few things I’ve noticed- artists that specialized in or often featured an element in the majority of their works will have it show up unprompted and morph the result, like Masamune Shirow, Chriss Foss, H.R. Giger, Simon Stalenhag, John Harris, Huge Ferris, Bouguereau, etc so this means that prompts more fitting to their niche in art (mechanical, spaceships, aliens/gothic, nudes, skyscrapers etc) will generate images that feel more like their works which is incredible
And another thing I noticed that’s really blowing my mind and needs further investigation Imo. With regards to Rob Gonsalves the AI seems to not only generate an image in his visual style but also utilizing his clever surrealist technique!! The snow in the distance turning to ice ponds which resemble the shape of water lilys. And in the other prompt there seems to be clouds in the distance which seamlessly shift to packed snow on the roofs of the building in the foreground – this resembles Gonsalves clever surrealist style and technique, so the AI can learn and copy and generate its own attempt at it surprisingly well, its remarkable.
I have a few artists suggestions I think would yield interesting varied results-
Is there a way to write the prompt to ensure it looks for the right artist? I tried with paleoartist John Conway, and can’t get Disco Diffusion to use the right references…
So the “Trending on artstation” is throwing things off a bit, and handing it subject matter directly is also throwing things off in evaluating if DD can recognize the artist at all. While the results are cool.. several of these artists are not even trained into DD.
The way you can tell is to first type “Artwork by nobody” which will generate a control image. Then you can do “Artwork by (artist name here)” ie “Artwork by Tyler Eldin” and see what you get. If the result looks like the work of the artist with their subject matter. The training model is strong.. if however it looks similar to the control.. then the training model is weak or non existent. (basically the further from the control image, the stronger that training model is)
This will also indicate to you what kind of training it has by the colours it uses and the subject matter it picks which will tell you how you should expect it to behave when given new things to create. Does it have any colours of its own? Does it just hold textural detail? or is there unique shape and subject matter? can it even adjust composition? Stronger training models can do more.
Weaker models will only work well when the subject matter is similar to the kind of work the artist normally does. I did this test 100 artists from this list (plus a few off my own) I found 17 that failed miserably and a few which were only partial fails. Hopefully this helps someone else.
This is an amazing resource that has done more to boost my appreciation for visual arts (and a great deal of the artists whose styles have been rendered here) than anything in the 36 years I’ve been alive.
It’s also an amazing resource for my future creations of art in collaboration with these amazing AIs.
Thank you very much to all of the contributors, and every one else involved.
thanks for the great collection and references
some suggestions:
Akira Kurosawa
Alejandro Jodorowsky
Christian Riese Lassen
David Lynch
Fra Angelico
Franz von Stuck
Fujiko F. Fujio
George A. Romero
Georges Ferdinand Bigot
Giotto di Bondone
Hajime Sorayama
Hayao Miyazaki
Hideaki Anno
Hishikawa Moronobu
Isao Takahata
Jan van Eyck
Jim Davis
Jim Jarmusch
JirĆ Takidaira
John Carpenter
John Lasseter
Katuhiro Otomo
Kuroda Seiki
Machiko Hasegawa
Ogata KĆrin
Osamu Tezuka
Ridley Scott
Shotaro Ishinomori
Stanley Kubrick
Takeshi Kitano
Taro Okamoto
Tite Kubo
TĆshĆ«sai Sharaku
Utagawa Hiroshige
Cool but not so much. No way on this webpage to select all names to paste them raw on a word or excel file. One’s gotta type them down one by one.
Sounds nice, but pretty useless at the end.
Very cool. Thanks for sharing this. I use my own hacked version of Disco that works from the command line and have thrown together a quick shell script that can run any prompt on all of these artists. Already blown away by the results!
Cheers!
last two are from the same person? or is the second to last supposed to be h.r. giger?
last two are ZdzisĆaw BeksinÌski
These are great. Thank you! Can I just ask, Do you use a programe to make the pictures bigger while keeping the detail? I want to do a print (30 by 18 inches) But it looses a bit of detail, and just wanted to know if there is a way round this?
With my art, as soon as its done, i use ESRgain, then in photoshop use blow up 3, but still loses a bit of detail. Is there a programe you use for this? Or is this the best I will get it?
Thanks for your comments. I have used ESRGain in the past but its not as good as Topaz Gigapixel AI, which also uses machine learning to increase the size while preserving the details. I try out all the different models it has for the size I want and stick with the one that produces the correct details when upscaling
Are these all done with standard out of the box settings?
Hi Ethan,
Yes, except in some cases ViTL14 is activated in addition to default settings.
incredible, thank you so much for your work. I’ve gotten a lot of mileage out of it.
Thanks to the contributors who continue to create more artist studies.
This site is awasome!
I suggest:
Beatrix Potter… was an English writer and illustrator, she was best known for her children’s books
Ădouard Riou… was a French illustrator 1833 – 1900
This site is awasome!
I suggest:
Friedensreich Hundertwasser he was an Austrian visual artist and architect 1920-2000
Beatrix Potter⊠she was an English writer and illustrator, she was best known for her childrenâs books
Ădouard Riou⊠he was a French illustrator 1833 â 1900
Thanks for the suggestions Lorenzo, the guys have taken note of these artists you suggest.
Really helpful, thanks for sharing this.
I got a different result by setting on Rebecca Guay, more floating strokes and more transparencies. I probably set differently, maybe less Steps or a lower Clip_guidance_scale than yours, who knows.
I’m curious to see if you add any new artists, there are some I really like that are missing here? đ
I will continue to add them to this post as the contributors continue creating more artist studies
It would be great to have this without “Trending on Artstation”
You can always run your own prompt without it.
This is a farce. That youâre all using the work of living artists to inspire use this tech is extremely harmful to those artists, and are encroaching onto their livelihoods. I could see MANY lawsuits on this, especially as this is done without their consent. Disgusting. Take this website down
Yes, by this standard all teaching art should also be banned and forbidden. Because when teaching art, we never got permission from Michaelangelo, Leonado, Van Gough etc.
hahaha you are naive @Concerned artist
@Concerned Artist Please provide your real name so we can use your style.
Thank you very much. This article is very helpful. It saves me a lot of time looking for the right modifier.
Pascal Blanché is not present in LAION400, you should remove it for now. Because the image is not relevant.
https://rom1504.github.io/clip-retrieval/?back=https%3A%2F%2Fknn5.laion.ai&index=laion_400m&useMclip=false&query=by+Pascal+Blanch%C3%A9
This is laion doofus
Isn’t
There is an Eiffel Tower in the generated images, this is also the case on LAION400. This artist has never put an Eiffel Tower in these images. Moreover, the generated images do not correspond at all to the style of the artist. It is probable that other artists in this list have the same problem ?
Many artists seem to have limited or dubious effect on the output. Many names produce their own quirks despite not representing the artist accurately at all. This is one such example – Pascal BlanchĂ© is not represented, but the outcome is noticeably different than the “control” prompts without his name. I think Harmett plans to post these here soon, for comparison.
As two great contemporary artistst, I suggest:
Jonas Burgert
and
Glenn Brown.
Both have a very special ductus…
Added to the list – it may take a while to get to… We have a lot of artist names.
What parameters are you using: steps, cut_overview, cut_innercut, etc?
Would be great to publish a txt with all the parameters using in the experiment, because I’m getting quite different results with the same prompts.
Thank you in advance!
Thank you very much! Very interesting.
A beautiful painting of the lost temple of the underwater world by ross tran and Mobius, Trending on artstation.”, “dreamy blue color scheme
I found some Manga/Movie/Game “Artists” that I pretty like.
HIDEAKI ANNO (for Evangelion)
Katsuhiro Otomo (for Akira)
Miyazaki Hidetaka (for Dark Souls)
Togashi Yoshihiro (for )
This is absolutely amazing, I’m blown away. The explosion of new creative potential this enables is unbelievable, styles and techniques lost to time and practiced by few alive can be generated, this means new art coming out that draws from almost the entirety of art history, just mind blowing to think about, there are often artists whose styles and aesthetics you love and admire but they are long gone and no one devotes the time to study and create art in their style – but AI can get you an amazingly good approximation its better than nothing and the results really speak for themselves here…
A few things I’ve noticed- artists that specialized in or often featured an element in the majority of their works will have it show up unprompted and morph the result, like Masamune Shirow, Chriss Foss, H.R. Giger, Simon Stalenhag, John Harris, Huge Ferris, Bouguereau, etc so this means that prompts more fitting to their niche in art (mechanical, spaceships, aliens/gothic, nudes, skyscrapers etc) will generate images that feel more like their works which is incredible
And another thing I noticed that’s really blowing my mind and needs further investigation Imo. With regards to Rob Gonsalves the AI seems to not only generate an image in his visual style but also utilizing his clever surrealist technique!! The snow in the distance turning to ice ponds which resemble the shape of water lilys. And in the other prompt there seems to be clouds in the distance which seamlessly shift to packed snow on the roofs of the building in the foreground – this resembles Gonsalves clever surrealist style and technique, so the AI can learn and copy and generate its own attempt at it surprisingly well, its remarkable.
I have a few artists suggestions I think would yield interesting varied results-
Ătienne-Louis BoullĂ©e
Paul Chadeisson
Lawrence Alma-Tadema
Frederick Carl Frieseke
tokyogenso
Thanks
Is there a way to write the prompt to ensure it looks for the right artist? I tried with paleoartist John Conway, and can’t get Disco Diffusion to use the right references…
So the “Trending on artstation” is throwing things off a bit, and handing it subject matter directly is also throwing things off in evaluating if DD can recognize the artist at all. While the results are cool.. several of these artists are not even trained into DD.
The way you can tell is to first type “Artwork by nobody” which will generate a control image. Then you can do “Artwork by (artist name here)” ie “Artwork by Tyler Eldin” and see what you get. If the result looks like the work of the artist with their subject matter. The training model is strong.. if however it looks similar to the control.. then the training model is weak or non existent. (basically the further from the control image, the stronger that training model is)
This will also indicate to you what kind of training it has by the colours it uses and the subject matter it picks which will tell you how you should expect it to behave when given new things to create. Does it have any colours of its own? Does it just hold textural detail? or is there unique shape and subject matter? can it even adjust composition? Stronger training models can do more.
Weaker models will only work well when the subject matter is similar to the kind of work the artist normally does. I did this test 100 artists from this list (plus a few off my own) I found 17 that failed miserably and a few which were only partial fails. Hopefully this helps someone else.
All said and done, this is still a cool list.
Amazing :). I’d love to see H.R. Geiger, Alberto Seveso & David McLeod as well!
This is an amazing resource that has done more to boost my appreciation for visual arts (and a great deal of the artists whose styles have been rendered here) than anything in the 36 years I’ve been alive.
It’s also an amazing resource for my future creations of art in collaboration with these amazing AIs.
Thank you very much to all of the contributors, and every one else involved.
Great resource, thank you so much.
thanks for the great collection and references
some suggestions:
Akira Kurosawa
Alejandro Jodorowsky
Christian Riese Lassen
David Lynch
Fra Angelico
Franz von Stuck
Fujiko F. Fujio
George A. Romero
Georges Ferdinand Bigot
Giotto di Bondone
Hajime Sorayama
Hayao Miyazaki
Hideaki Anno
Hishikawa Moronobu
Isao Takahata
Jan van Eyck
Jim Davis
Jim Jarmusch
JirĆ Takidaira
John Carpenter
John Lasseter
Katuhiro Otomo
Kuroda Seiki
Machiko Hasegawa
Ogata KĆrin
Osamu Tezuka
Ridley Scott
Shotaro Ishinomori
Stanley Kubrick
Takeshi Kitano
Taro Okamoto
Tite Kubo
TĆshĆ«sai Sharaku
Utagawa Hiroshige
thank you!
You can check the artists you propose may already be in there database > https://docs.google.com/spreadsheets/d/14xTqtuV3BuKDNhLotB_d1aFlBGnDJOY0BRXJ8-86GpA/edit#gid=0
WOW!! thank you VERY much!!!!!!!!
Any Indian artists? Nice article btw, thanks.
Who’d you like to see?
Cool but not so much. No way on this webpage to select all names to paste them raw on a word or excel file. One’s gotta type them down one by one.
Sounds nice, but pretty useless at the end.
Thanks. You can find the link to the database (under Contributors heading) which has all the artist names and you can copy them all from there.
I propose you Camille Pizarro and Victor Vasarely