Tumgik
#ganbreedings
ink-the-artist · 7 months
Note
so somehow it’s the first time I’ve seen your Dogs illustration from May. it’s very very cool and strongly reminds me of AI-generated art — did you use AI to inspire the shapes? or otherwise influence the piece somehow? interested in your process! very cool.
yes! some more directly reference ai images id generated than others lol
Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media Tumblr media
573 notes · View notes
kichijouji · 2 months
Text
I miss when ai art was just fun silly cursed shit like on dall e or ganbreeder, just seeing what the robots would create on our own, but like 4 years later, this current ai craze sucks ass
2 notes · View notes
ganbreedings · 1 year
Text
this blog is on indefinite hiatus
with the recent boom of ai “art”, i do not want my blog to be associated with such.
this blog was made due to the creator’s interest in neural networks and ai images in general and i do not want to be associated in any way with the recent spotlight it has been put in.
i do not support ai “art”, especially since i am an artist myself. i believe ai “art” is taking away from artists and the work and time we put into our pieces. 
my interest in ai images is nothing other than seeing it as an interesting novelty; never something corporate and never something to replace humans. i love science and technology; i do not love taking away from artists.
if you disagree, you are welcome to unfollow.
15 notes · View notes
Photo
Tumblr media
GANBREEDER, the beginning
2 notes · View notes
doggerell · 3 months
Text
uuuuuuuuuuuhhhhhhhhh I am cool and sexy and you dont want to beat me to death with hammers and you will hear me out. please.
0 notes
oetscop · 1 year
Text
Tumblr media
terrible aura coming from this temu screenshot
0 notes
metacinematics · 1 year
Video
vimeo
björk talks about her TV (2022)
found footage project for intersections media
first final cut pro x test
0 notes
aigeneratedstuff · 2 years
Text
Tumblr media
0 notes
tamasquish · 2 years
Photo
Tumblr media Tumblr media Tumblr media Tumblr media
Some old artbreeder photos of things that look suspiciously like strawberries
0 notes
medicalunprofessional · 2 months
Text
i miss when ai was silly and bad and dumb. Remember ganbreeder? I do….. I created so many fucked up looking dogs …. I want ai to be stupid again… Look at my son now……Hes gone down a life of crime…. Greed always wins it seems.
56 notes · View notes
person918x · 2 months
Note
hey, i would love to know what kinda AI art tools you're using that don't require a login if you feel like sharing. i've been playing with bing's dall e 3 and it's fun but the censorship is really frustrating. i miss the old ganbreeder site so much, it's really sad the way that one went. anyway, love your art, have a good one 💜
Most recently deepai.org (downside: there's a daily limit). Also Neuralblender (though they keep changing the algorithm and I don't like the newest one very much) and Craiyon. I also use GIMP to edit out watermarks and make other edits as needed.
18 notes · View notes
ink-the-artist · 2 years
Video
3K notes · View notes
my-reality-my-rules · 2 years
Text
list of sites and apps i recommend for designing your DR shenanigans
part 1 - avatar creation (DR self)
- - -
as it is in the title! I've been meaning to post this for a while now, so here it is ((: hope y'all have fun making new DR pics lmao
[warning: semi-long post]
˚✦•·················• ✼ •·················•✦˚
LIST INCLUDES:
Artbreeder
Everskies
Hero Forge
Lily Diary
Live Portrait Maker
Metahuman
Picrew
˚✦•·················• ✼ •·················•✦˚
Artbreeder
WHAT: Artbreeder, formerly known as Ganbreeder, is a collaborative, machine learning-based art website. Using the models StyleGAN and BigGAN, the website allows users to generate and modify images of faces, landscapes, and paintings, among other categories. [lifted from Google Search]
PROS: allows for realistic avatar-making. overall, has a pretty good range when it comes to the genes/features to be selected. everything's also free to use. images can be mishmashed at any rate. thousands of AI-generated photos are also free and available for use.
CONS: controls are quite difficult to maneuver, especially for people new to the site. certain genes are also inconvenient to use—offsetting another gene in the process of balancing them. additionally, also highkey racist when it comes to the cultural-based gene designs (e.g., the east asian or desi features). limited number of uploads.
PERSONAL EXAMPLE:
Tumblr media
[*photo enhanced with Remini]
≿❈≾
Everskies
WHAT: Welcome to the Everskies mobile game, an avatar maker and life simulator game that combines the dress up concept, where you can create your own character to match different clothes, hairstyles, shoes, and much more. You can also meet new friends, chat with people, and design your avatar fashion style as you want. Your second life with the virtual world is in Everskies! Join it! [taken from the mobile app description]
PROS: has a very wide range of clothing options. pieces from every style can be found—from punk to e-girl fashion, to forties clothing and little space. includes mini-games and competitions, as well as gifting and tipping to other users. outfits can be bought as a whole, or pieces can be bought separately.
CONS: not much, unless you like to splurge and immediately spend the 10,000 starting stardust for buying clothes and/or magazines. some features also require in-app purchases.
PERSONAL EXAMPLE:
Tumblr media
≿❈≾
Hero Forge
WHAT: Hero Forge® is an online character design application that lets users create and buy customized tabletop miniatures and statuettes. [lifted from the official Hero Forge website]
PROS: allows for interracial character creation (e.g., fantasy creatures like tieflings or gnomes). includes preset outfits that can be edited. little to no boundaries when it comes to avatar-making. custom colours and posing, down to all the littlest details. additionally, the site has save slots, so you can check your characters at any time. you can also order to have these printed out (which means you can even have a miniature of your DR self).
CONS: none, unless you're unwilling to actually subscribe to the company.
PERSONAL EXAMPLE:
Tumblr media
≿❈≾
Lily Diary
WHAT: Lily Diary is a dress-up game where you can decorate avatars and backgrounds. You are free to put your saved avatars anywhere you want! [lifted from the App Store]
PROS: includes a wide collection of preset outfits to try on, both free and paid for in-game. clothes can be layered and flipped. has a good range when it comes to facial features and colour palettes. also allows for stage and background editing for making screenshots of avatars.
CONS: a bit limited when it comes to the outfits to be worn. for example, there are traditional outfits whose poses cannot be changed, and some clothes are challenging to layer.
PERSONAL EXAMPLE:
Tumblr media
≿❈≾
Live Portrait Maker
(version 1 - girls), (version 2 - guys)
WHAT: Create your own fashionable, animated 2.5D girl. | From the creator of Live Portrait Maker - create your own animated 2.5D guy. [lifted from Google Play]
PROS: has an animated avatar whose line of vision follows the movement of the user's finger/cursor. facial features are editable singularly; includes options like tilting and resizing, as well as widening and lengthening.
CONS: includes some in-app purchases, if you're unwilling to spend for the more interesting items to be used.
PERSONAL EXAMPLE:
Tumblr media Tumblr media
≿❈≾
Metahuman
WHAT: MetaHuman Creator is a free, cloud-streamed tool you can use to create your own digital humans in an intuitive, easy-to-learn environment. Using MetaHuman Creator, you can customize your digital avatar's hairstyle, facial features, height, body proportions, and more. [lifted from the official Unreal Engine website]
PROS: includes a collection of faces to sculpt and combine, as well as provides specific measurements for smaller body details. editable background and lighting. also includes animation for the avatar, both in full body and in face, complete with 3D camera rotation.
CONS: some of the presets are limited when it comes to having a wide range of options to choose from (e.g., clothes or lack of eye colour variety). also requires high-speed internet in order to work at full functionality (including fast, ray-traced graphics).
PERSONAL EXAMPLE:
Tumblr media
[apologies if the resolution looks a bit bad—lmao i was running on 9 FPS at the time]
≿❈≾
Picrew
WHAT: Picrew is a layered paper doll-style avatar maker website. It was initially developed by the two staff of the Japanese company TetraChroma Inc from July 2017, and officially released in December 2018. [lifted from Wikipedia]
PROS: perfect for making animated pictures of characters. incredibly diverse when it comes to design—from skin tones, eye colours, to the most outlandish facial features. also caters to representing people with disabilities and doesn't shy away from expressing minorities as well.
CONS: none that i can immediately think of. the only real trouble I've faced is a situation wherein the artist creating the picrew hasn't finished it yet—but that's not entirely a problem. or perhaps unavailable designs in a picrew (e.g., red eyes), but there are usually substitutes.
PERSONAL EXAMPLE:
Tumblr media
[LINK: The Lady of Hera]
˚✦•·················• ✼ •·················•✦˚
193 notes · View notes
ganbreedings · 2 years
Text
HOW TO IDENTIFY AN IMAGE FROM MIDJOURNEY, DALLE-2, OR STABLEDIFFUSION & WHY YOU SHOULDN’T CLAIM IT AS YOUR OWN
people are pieces of shit and pretend that their “art” is actually theirs and not from AI. i like AI and find it fascinating and as someone with any inkling of a following, i want to let people know of this.
i am not affiliated with any of the AIs i talk/post about and i am not being sponsored by any of these to put down one and prop up another. i am trying to spread information on how this works and why it’s not okay to claim AI “art” as one’s own.
all images used in this post were either generated by me or were publicly available.
---
WHY SHOULD PEOPLE NOT USE AI TO CREATE THEIR “ART”?
the way AI pictures such as the ones from these websites and apps is by taking images from around the web to combine into one image based on prompts. this is accomplished by finding images attached to descriptions and putting them in a GAN; i have more information on that here. 
so essentially, by generating new images, you are ultimately taking someone else’s work and smashing it all together to create something new, most likely without permission.
this is also how artbreeder works, which was more evident in its old name of “ganbreeder” (which contains “GAN”).
since this is technically at the end of the day someone else’s work, this is also why they shouldn’t be sold as NFTs (not to mention the complete lack of effort put into them; you just come up with a prompt and type it in and the ai does the work. at least with stuff like bored apes someone had to draw all of the initial variables individually so an AI could mash them all into a single image with no rhyme or reason).
---
HOW TO IDENTIFY AI-GENERATED IMAGES FROM MIDJOURNEY
midjourney images have a very specific look; they tend to look very painted with sort of... muted colors a lot of the time. youtube channel SolarProphet has many videos that feature midjourney. they all tend to look like paintings and i’ve found that the AI very much likes to make portraits with a character front-and-center.
Tumblr media Tumblr media Tumblr media
---
HOW TO IDENTIFY IMAGES FROM DALLE-2
dalle-2 images can be harder to identify because the AI is very advanced with many data points.
it is able to generate images with multiple art styles, but also may generate things that look like photographs.
the best way i can describe dalle-2 images is just looking... off. if you look carefully, something will look wrong. in the picture below the top eyes are mismatched, etc. something would be subtly off. you have to look carefully.
Tumblr media
in this one: one of the eyes looks smudged and the spoon handle doesn’t seem to be leading to anything.
Tumblr media
dalle-2 can also generate rather convincing clipart at times, but they also may have something off about them. this picture of a frog looks normal at first glance, but if you look carefully, you can see that it’s missing a toe, coloring on the far arm is wonky, the lineart on 2 of its fingers is bulging out, there is an extra line on the top shirt pocket that cuts off a yellow band, the legs are oddly shaped, etc.
Tumblr media
yes, some artists may draw like this, but most likely they would keep proportions consistent, no? (saying this as an artist myself btw)
---
HOW TO IDENTIFY IMAGES FROM STABLEDIFFUSION
stablediffusion is a lot like dalle-2. due to lack of resources i can’t really give the best answers on it. it is much more customizable than either midjourney or dalle-2. the web apps can go up to 150 steps on dreamstudio and 50 on the huggingface website, which affects quality (unlike midjourney and dalle-2 where the number of steps is fixed (to my knowledge, anyway)). 
my computer struggles to run stablediffusion locally due to my graphics card (i CAN run it but it basically eats all my memory). i also don’t know how many steps midjourney and dalle-2 use to make their images, so i can’t make pictures with exactly the same amount of steps to compare it all to.
i got these using both web apps and as you can tell, they are off and not quite right, though i expect with more steps, they would be better. i expect the way of telling that they are AI generated is similar to dalle-2?
Tumblr media Tumblr media Tumblr media Tumblr media
---
it’s hard to explain how i personally tell when an image is generated by ai, and of course my method is not foolproof and it will never be. humans are imperfect creatures, after all. these are just patterns i’ve noticed personally.
7 notes · View notes
Photo
Tumblr media
GANBREEDER, the beginning
1 note · View note
doggerell · 3 months
Text
I looove seeing ppl in the ganbreeder feed who are clearly trying to make weeeeeiiirrd shit. horror tool!
2 notes · View notes