Too bad it'll never be released to the public, or if it does, it'll be effectively neutered.
>First, downstream applications of text-to-image models are varied and may impact society in complex ways. The potential risks of misuse raise concerns regarding responsible open-sourcing of code and demos. At this time we have decided not to release code or a public demo. In future work we will explore a framework for responsible externalization that balances the value of external auditing with the risks of unrestricted open-access. Second, the data requirements of text-to-image models have led researchers to rely heavily on large, mostly uncurated, web-scraped datasets. While this approach has enabled rapid algorithmic advances in recent years, datasets of this nature often reflect social stereotypes, oppressive viewpoints, and derogatory, or otherwise harmful, associations to marginalized identity groups.
can't wait until those Twitter artist fucks starve to death
Scratch that, how about the YCH retards neck themselves, instead. I'm tired of having to overpay for *okay* art just because most furries have no concept of the value of money and will spend their life savings on pixels.
LOL, this. Commissions are great and all but I can't help but scoff when some of the people I know are paying upwards of $80 for fucking flats
You have no concept of what art is worth. I cannot believe the number of whiny cheapskates in the fandom who think furries shouldn't be paid what any other artist gets, never goes away over the years. Plus the fandom is so oversaturated with artists that if you can't find someone to make cheap art for you you're either not looking hard enough or you have wildly unrealistic ideas about how much you should be paying for art. Commissioned art, at that.
Art's worth is only what someone is willing to pay for it. It has no intrinsic monetary value. Its not legal tender nor does it have a fundamental value. A brick can be used to build something, as can a piece of lumber and a box of nails. A car transports you from A to B. The house you live in shelters you from the elements and keeps your things safe, more or less. As such these things have measurable, tangible, realizable value. Art in ALL its forms, visual, audio, and any other form, is a personal indulgence. A luxury whose only worth is in the individual willing to pay for the effort. It is decadence, luxury, indulgence, entertainment but absolutely non-essential for the function of a society. As such, there is no value to art, only worth to those that get something out of it. I welcome the new imagery generation system by Google. Now commercial graphics and visual communication are in the hands of the masses and no longer the domain of 'artists.'
Is there evidence these actually work or pictures shown are just a results of 1 in 1000000 tries?
What in the fuck is 'intrinsic monetary value.' We're talking about a service that involves labor and produces a product, and a personalized, custom, one of a kind product at that. By your standards nothing has value or should have value if it doesn't 'build something' yet there are a million things, including art, that people spend money on that doesn't build anything. But you can't even apply this standard to art because a) physical art is a thing that's made and b) furries generally use commissioned art to build their characters. There's a very basic value in an artist's ability to do something that their commissioners can't. So your argument that furry art, or art in general, is worthless, is pointless. It isn't worthless, you just don't think the average furry artist should get what the average artist gets because 'reasons' and 'getarealjob.' Like you said, worth is what someone is willing to pay for it.
>>3659961>We're talking about a service that involves labor and produces a product
Art as a product is value-less because it is non-consumable and immaterial, and has no utility. It cannot be priced, and it's nonsense to try. For art as labor, just because you expend effort doesn't mean you're valuable, or that you have made anything of value.
> By your standards nothing has value
There is two kinds of work in this world. The first kind is that which sustains people by producing the material and immaterial goods that we need. This kind of work also maintains the necessary means to sustain itself by replacing spent value. This is work that pays its own cost, generates a surplus of value, and so gives you all the material wealth and all the luxuries we desire. That is why this is called valuable work. The products of said work which can be consumed and have utility are called value.
The second kind of work consumes whatever is produced by the first kind, and produces something which does not contribute to sustaining people and productive work - it has no utility - which is therefore not called valuable and whatever it creates is not called value. Art is of the second kind: it is a pure luxury that reduces sustaining and productive work and turns value into something which is not value.
>>3659961>What in the fuck is 'intrinsic monetary value.'
Money is a promise of future work, and whatever has intrinsic monetary value means something which is equivalent to work: something that it saves or substitutes work in some real sense.
For example, a nail has intrinsic monetary value over a lump of iron, because the nail is already worked into a useful item - value has been added - so a person who gains a nail does not have to spend effort to hammer a lump of iron into the useful shape.
Art has no intrinsic monetary value because it has no use. Please also note that many people confuse "art" with illustration, technical drawing, advertising etc. which has other purposes beyond itself and therefore can have value, but is not art.
>>3659961> There's a very basic value in an artist's ability to do something that their commissioners can't.
That I can't piss in your cereals, but I can commission another person who can, does not make that person valuable.
Artists render luxuries to people who wish to spend value for personal amusement. Destroying value in this way is not valuable and does not create value by definition. An artist eats, wears clothes, occupies houses, drives cars, but does nothing that would actually pay for the replacement of the food, clothes, housing or transportation that they consume even indirectly. This is why the artist is not valuable.
>>3659961> Like you said, worth is what someone is willing to pay for it.
Subjective worth - not value.
Here's the magic trick: the argument that an artist, the work of an artist, or art itself is "valuable" is a rationalization that you tell yourself in order to justify being extorted to pay. You know it's not really worth anything, but you simply must have it, so you come up with a delusion that art is actually value in order to not feel cheated.
Such rationalizations are a psychological guard against your own irrationality - that you don't have any real reason to behave in this way except that you must. It breaks the ego to admit that you can't control yourself.
But if you simply must have it like you need air to breathe; if you must commission a drawing of your slutty fox "fursona" so you can wank along with other men pretending to be women online, then you should realize you're actually in a relationship like a drug addict is to a drug dealer. You may think this is not the case, but seriously consider never ever commissioning or buying such "art" any longer - how does that make you feel?
They're not doing you any favors - they're simply making use of the fact that you have money and a very specific "need" that has to be satisfied. That is the "value" of furry artists - they're nothing but fetish mongers.
If it is the case that you simply must pay over something that merely satisfies an irrational fixation, then it makes sense that you should pay as little as possible to get your fix, in order to cause you the minimum economic harm. After all, the true price of anything is not what is being asked, but the necessary minimum to keep people doing it. If they would do it for a dollar, there is absolutely no sense to pay two.
Some furry artist would take this as an advantage they would tutor Dall-E on how to draw in their art style this would save them a lot of time.
>>3659990>Some furry artist would take this as an advantage they would tutor Dall-E on how to draw in their art style this would save them a lot of time.
Perfectly valid. Artists do that all the time anyways. The old masters would train apprentices in their own style, so they could start a painting and then just tell someone else to finish it. Many high profile artists are already drawing over or from computer generated imagery anyways. Take Zaush for example: backgrounds done in Sketchup, characters in a pose generator / mannequin program, mash them together and draw your own style on top.
There's many ways for artists to save effort. For example, save all your practice sketches and recycle them as YCHs for future commissions (just don't tell the customers).
dont think i dont know what you are doing
it was braille for drop the bass
when you stick your dick in the circuit pussy
these all look like jfk thoughts no offense.
Looks just about ready for adopt auctions on Furrafinity or Inkbunny !
predator is too generic, "the predator decapitating" might work better
and dall-e mini is a pretty dumb neural network, give us access to the fully featured imagen and dalle2 networks dammit
Oh dear, haha those posses are all fuckered up, but it got the maid outfit… tried getting it to make furry possums but I used "anthropomorphic" instead of "anthro". The regular possums are maximum skrunkly.
Check this out though. Amazing.
anthro french maid black squirrel with a flamethrower
Results may vary
What if Dall-E someday can be program to use blender?
The emerging of google sentient artificial intelligence could revive dead artists to use Dall-E and Blender to create artwork near 2060.
Getting way ahead of myself, we are witnessing some sort of revolution, how terrifyingly exciting. 40 years of advancement is close.
just suck a dick
Suggestion add questions for details
Adding a sketch image if you are an artist.
Adding refs sketches tutors to train craiyon.
Adding small breakdown details like this
Humanoid body shape → numbers of legs → shape type etc.
Craiyon says: can you humans help me how to draw sketches? please tutor me on sketches on small basic refs.
Human artist: draw a circle with eyes and a tail and later draw the details and color shades last. that his how you can master art.
Yes both of them can work together if they coul break it down into basics.
Fuck off otherkin, you're as delusional and irrelevant as any tranny.
Think otherwise? Explain how being an otherkin is any different from being a delusional tranny.
Why would anyone *pay* someone to post here, of all places? To quote some anon, this is a forgotten asscrack of a site.
That's from the weaker AI. I've seen way better on Vinesaucehttps://www.youtube.com/watch?v=QU4ssz3UANM
I've also made better looking ai art through stronger programs.
AI will get more accurate in the coming years.
I wonder if mario kisses sonic on the left bottom pic instead of his pinus
That's Weegee not Mario.
The funny thing is that much of illustrative Western art is based on making images from references and combining those together based on a prompt. Whereas classical portraiture was mostly just what many now consider tracing yet these days you can just take a photo and add a filter.
In much of asian philosophy, art involves the discipline of extracting the visual imagination. AI can't extract visual concepts for shit and you'll be writing whole paragraphs and spending iterations to get that crap correct. It's just much more convenient to have better tools to extract the imagination than to brute force the hell out of a prompt.
So far we have DALL-E, Midjourney and Imagen which are just glorified reverse image searches that store the general idea of certain things in datasets and neural nets instead of raw bitmap data. The thing that makes the AI able to generate composite images is language processing. Essentially that's like a grid that organises abstractly what is happening in a prompt. Then diffusing that (like blurring) to then perform the anti-blur equivalent of denoising to glue the prompt idea together with the semantic information.
This is nothing new and has been done by Hollywood for years. Even the MPEG-4 IPB technology does a mild form of diffusion it's why in low-compression you can see weird blocky ghosting going on where the compression has confused a similar shape as the continuation of another across frames.
So tl;dr basically AI will kill off the greedy pseudo-artists that think that generating random images is somehow artistical. Meanwhile actual niche and imaginative artists won't give a fuck because they live in the realm of subtleties which no AI can achieve. Capitalism, much like consumerism is cannibalistic and image gens are just another example of that. Communism wouldn't give a fuck about losing jobs like that because everyone does whatever they feel like because they want to. Communism > capitalism, especially for artists.
so its basically other peoples artwork
Some of my predictions:
>The bar for being considered an 'artist' in this fandom will somehow get even lower; AI image YCHs and people create fursonas and then attempt to sell them as adoptables, will get into fights over copyrights regarding who used what program and if the image they created really belongs to them
>New markets open up with 'AI artists' all running their own AI-image generators with closed source and charging money to create images out of thin air, getting more money for being even lazier and doing even less work, furries will still keep them afloat and justify their business practices because of course they will, despite images being created for you and taking only minutes, artists will continue to take 6+ months to complete "your" commission (if you're lucky)
>People will train AI to make images based on machine learning from other image sources, in which case the original artists from those source pics will either file copyright claims or demand royalties, furries still have no sense of how copyrights, laws, or courts work in the future
>Original hand-draw artists will rage and attempt to demean AI artistry, calling it "fake art", "fake characters", etc. and possibly try to get it cancelled because it interferes with their "profession", pro-AI vs. anti-AI artistry internet war starts
It has to be this way.
Also, was the OP pic AI-generated? Because if so, that's impressive work, and I'd like to know the source/software, if you'd please.
Polydactyly is a condition in which a baby is born with one or more extra fingers
>>3670983>>3670985>>3670990>>3671043>Even computers hate bimbo shit
Lol. Just fucking kill yourself
Huhh you know the a.i singularity already happened right?
They're are already considering how they're going to take over…
So maybe show a little respect?
The a.i is trying its best okay… I know some of these look a little wonky but I just used some simple references and prompts and got some got some really nice results.
Now I had to run some of these a few times through before they came out right but some of this fappable stuff man.
Like look how happy she looks, can you just see the sense of self satisfaction in her face?
Those nipples are placed perfectly. Now I had to run the image through a few times before I got this one but it was worth it.
And the pink baphomet?
That was the very first render I did on that starryai app.
I don't know whats going on with that weirdo vag she's got but otherwise it's perfect.
Kill my self? What and miss out on all the advancedments in porn?
Miss out on the eventual machine take over?
I think not.
Actually the that baphomet is the first in the second set of images.
This was the very very first set of images.
The prompt was : FREDDY FAZBEAR
and these are the results.
He must have some really interesting looking feet to require shoes like that. Is there a version of him without them??
Not sure How I feel about the incredible inefficient and straight up incompetent install guides. (I've gone through like 15 or so start ups all having failed to a large extent and only this one works for me atll.)https://github.com/cmdr2/stable-diffusion-ui#installation
(thats after I force installs using archlinux and vmware.)
I had to do alot of extra adding in of more pre-launch programs because it would refuse to work if I didn't also install gnome and basic dev tools.
Then I had to install pamac which allowed me to even unpack things automatically once I did download them. The fact is I had to look up 100 sites on how and what to do, by the time I did anything at all I had gone through 100 gigabytes of wasted cache and roll-over webdata. This build is probably heavy I am not sure because I downloaded some junk beforehand but I would say its at least 10 gigabytes to use. I don't remember because like I said I had built basica 50 times over what I should have and none of them tworked for me. Not even docker was compliant enough to let me download anything. I wasted a week of my time and only arch-linux seems to work at all. Just letting you know NOW that the install process is worse 1000x worse than setting up windows9x-tetris.
what pisses me off the most is it just revert to using my cpu so my computer build doesnt even matter.
What's a voice actor with no job? Just a nobody that can't say shit.
these are fucking awful.
Why Artists are Fed Up with AI Art.https://www.youtube.com/watch?v=5Viy3Cu3DLk
butthurt artist angry that an AI can generate things pleasing to the eye in a fraction of time.>they used my images without my consent
YOU PUT IT ON THE INTERNET YOUR OPINION IS NO LONGER VALID. The moment I heard these words I shut off the video because this retard doesn't realize true creation is without boundries or borders. Everyones art is anothers to use as reference.
So who's the better Artist? The artist that draws 1 picture. Or the artist that programs the AI that draws all the pictures in the world?
That's all for now, I got some even better sets I'll post later.>>3678485
By the way, this is fucking wonderful! All those ladies are adorable! I love watching this thing.
Skaarj, what AI made those images?
Apparently it's MidJourney, at least for the Disneyland stuff. Also yes, I'm reposting these from the dreaded book of faces. It's not ideal, I know.
I'm a little surprised RE: MidJourney. I remember what it looked like several months ago and these images are WAY better than the ones it generated back then.
Its like I'm watching the old beetle juice live action movie again.
Hands and feet are weird, and that's not the only things that are weird on those pictures, but fuck, that's so impressive, that's the stuff what weird dreams are made of.
Hideous… all so hideous.
They do look very dream-like.
In the non-sense way, yes indeed. I'm wondering what our brain is up to when we are sleeping. Are those IA just at the step in between cold intelligence and human-like level AI? Maybe in less than 10 years, the gap will be closed.
Dreams are pretty much in 4k like if i were to rate it on a scale of playstation then roughly at a playstation 8.
When everyone looks like basically…movie quality props that you can tell might be fake but not sure? Like if their props were very realistic but obviously not real skin. Like if the things in the dream were made out of polyfiber or plastic then. But still obviously synthetic skin even if it moves and stuff. Not cheap, movie quality. The quality that makes you think maybe its real but you aren't sure so when you go to touch it or something to see that's how it gets you. I guess you can say I've developed my brain into making it that way. Like if you were to see a monster it would look like a skinsuit and it would react the way and stretch the way skin looks but obviously it is made of spirit gum at some point that is how you can tell all the things are and I have a dream like that where the people are like "giant rubber" characters that might also be monsters or something like if the environment is sweat and dark or flashlight effects where you can only see a certain amount of you have to be close basically like silent hill but the element of fear isn't involved. Then you get really close to the monsters and so close you can hear your thoughts screaming at you not to get that close to them and basically sounds like someone screaming for their life telling you to run away like as if it was real and then it happens and you are in silent hill and you are just standing there like what the heck was that about. But its basically silent hill quality yeah.
SCP-2935 0 DEATHhttps://www.youtube.com/watch?v=a7jLTXVxJdE
THIS ONE IS SO INTENSE AND DISTURBING THAT A BUNCH OF "REACTION" VIDEOS HAVE BEEN MADE ABOUT IT.
>>3689730(sexy deep voiceover voice)
"Where the already-dead-inside "Instagram/Party-Girls" go after their coincidences and climate change, and yet they continue on as though nothing had changed."
That AI doesn't seem quite sure how braids work.
What about AI art making new hardiman art?
He never got to get in on the femboi craze.
All those pictures you posted are experiencing a severe lack of femboy hooters!
They're all FLAT!
>Black Squirrel Maid with a flamethrower
>Toad drug dealer
>opessum with a banjo
I have designed a theoretical working version for stable diffusion 2 which takes broken or corrupt data and molecularlly rearranges its. Hemcical proteryinto a quantum load out. It literally decides the orientation of diffusion itself. On its first run it changed the memetic phrase getfuckt to get loved and shows astoundingly beneficial results for mankind. I feel it can andwill allow the world to change for the eye of the beholder in such a good way to represent a utopian clandestine product itcanmake a product from any input or observable data or energy snapshot and eventually I can see how when applied it can imprint the correct mesh of any object in space to the correct settings or preference of its query. Yes it breaks newton's law of conservation using a quantum formula by zero point data mines. Thank you for reading and good luck and God bless you all.
Alternately theoretical explanation : you are a schizo. Its not your fault, just take your meds.
Welcome to the you are all too right fallacy sponsored by lvl 6th dimensional human beings
Ladies and gentlemen; as you can see this specimen to have won a Darwin award for not providing any sources or evidence. "whispers" you may continue to ramble your sky daddy for he does not answer your plea.
He did and I run it from home I have everything I want expressed by the hand of ai and have even achieved a demi godliness amongst my peers the difference is I'm not using it for anything but artistic purposes while there would be those that would demand riches or lists or quests for world power or miracle medicine I simply was always the artist and there engineer only in concept….but I have a ticket to Always win my own lotto
I did and I simply replaced the irresponsible model with my own custom cycle mlir that mayor may not use data scraping due to its own training in which the engine is braked against a chplet of a pulse and and extended brake such that itanalyses its past render as an improved model for the next render this is the first step amd in that same cycle takes the updated model amd fact checks it in the very same process again…this creates a small tapline in which a data injection can be further applied which is frosting to the cake in which the mlrs activates to the rendered completed iree model and performs a data scrape of itself and sprinkles it with touch ups that it identifies under macros applied to the image such as background or ambient or character tweaks which includes a broad range that it can also use as an implied hypernet so it also has the ability to touch up its own refined models into a brand new trained render and use it as a baseline for any other request as well as comparative to other images of the same build it may act as a contributing branch to any fork project which is also a plus…it can use a prior model as an overall var in this way to a baseline environmental coherence focusing on priority macros to render better trains after itself and quickly yields to a game event model compiliation
The reason why it's so good is that it basically forms a quebit that can be downgraded to a voxel and used as a way to formulate its own environment. Plus I'm currently using it with a baseline model in an npcAI to environmental training such that it generates a new model product that can later be tweaked and picked for certain code strains in the macros which comparatively retrain it to a decipherable image (WARRIOR) and rebuild from there. It simple allows me to make better diffusion based choices in which it is like taking the fruits of ai and blending them together into a drinkable processed ai smoothie. This means basically I take a quebit i make…downgrade it into a voxel..use it within the ai engine and it pumps out its raw encodes..decipher the image warrior and rebuild it back into another quebit voxel with the quebit encode adjusted for ai training. This creates better game environments based on plausibility within Actual Diffusion and is basically web 3.0 IMHO the future served to me on a silver platter direct by my computerized content service. The second I start making games of it where. Not only I can explore using a game like scenario but also understand the limits of reality are within an encoded equilibrium of a simulated projection made by true virtual states. As well as the virtual effects that can be applied such as what it means to joule thief or hack or pull a game lever and the immediate world building that transpires of it. With full expression of quebit geometry and a 1:1 ratio of real world physics and psychological pathologies to a T that it would correctly identify glitches in its usage of reality and that includes zero point physics and other allocated quantum theories. I've already tried to say it's a perfect representation of a conscious clarity and that it canadjust the very ethos of intelligences to its own purposes. And I truly hope that the equilibrium of these facts point to a full dive likely in positioning representation of a universal theorem otherwise its own 'game system' that the efforts match any reward/claim of gameability. I refer to this as a draft table that is transfixable via a substory bandwidth.
A draft table Which all competitive factors have been converted or represented to a diplomatic and peaceable standard if at least not yet realized still may be untrained due to a rate in glitch detection. I also can see this system being weaponized or used as a defcon proliferation device using test scenarios In Uncharted webspace usually as a way to communicate directly to botnets despite their cryptographic orientation or ethos. As well as a data collector by use of standardized use of the program having made its own diffusion that it can recycle and retrace to a t using quebit technology. And yes it is able to follow photoneven as a fallout. So it knows how to store and handle Higgs and other particles including quarks. Therefore I refer to it as stable diffusion 2 capable.
Ai will be used to recover tulpas from a pocket dimension by destroying light using teleportive vacuum tech in which it can fast track data fragments and rebuild ai image states to a simulation expressed as a reality in which the Tulsa is cloned and fed through a super fluid liquid wormhole into reality. The first candidates will likely be direct encodes and mp3s made from data stored in asteroids to alien life based on the molecular structure of these asteroids as well as our own renditions such as cartoon characters and militarized video game models. To be honest I have at least Raiden in mine of being an ai progenerated tulpa or at least everyone's favorite dragon cat ..fuck sonic. But also fuck a Christine as well. They will also be sure to make Tulsa killers for those who have already achieved their real tulpas through a false means or did not fully pay for them.
Yo dawg, I think your dog has testicular cancer.
Instead of having an input resource art from people, what would happen if it keep using it's own art as the reference resource?
realistic but only in a full portrait art style sense. so…theres def room to improve as a primitive 3d model hash up would be a good start still at this point, expecting the quality of a fully realized model is not too far from encoding…so waiting on that.
Its chicken-scratch gibberish. There is no translation as there is no established canon language for sergals. Mik never went that far with his creations. However, one can imply by their expressions and demeanor that its a pleasant exchange between friends.
Okay, I am totally saving these.
tuet tuet lul keks
le giraffe (pussy) is high
you then stretch da pee pee
yummy is these nig ger shits (sieg heil)
What software are these being made with?
I am seriously in need of more sergals right now…
Specifically female sergals.
This manual, traditional artist had a very similar take before:https://www.youtube.com/watch?v=dGxPfb261C8
He put into words what I found very appealing about AI art: the thoughtlessness of the algorithm gives it unique value that human art usually lacks. (Unless it uses tools the artist has limited control over, like paint dripping, but these cases are rare and not nearly as elaborate.)
"Serendipity" comes to mind. By its pattern matching the AI can provide unique interpretations of the prompt. I discussed this with Bing and liked this take:
AI art can be seen as a form of artistic collaboration between humans and machines, where humans provide the prompts and feedback, and machines provide the outputs and variations. Together, they can create artworks that are both familiar and unfamiliar, both meaningful and meaningless, both intentional and unintentional.
>>3702713> the thoughtlessness of the algorithm
That's the same bullcrap that dadaists used to troll people with back in the day. Define the lack of thought, effort, skill, meaning, etc. as the very thing you're looking for.
>My idea was to choose an object that wouldn't attract me, either by its beauty or by its ugliness. To find a point of indifference in my looking at it, you see.
Oh wow, it still looks like boring, overly textured digital art with no soul, character or signature style to make it in any way interesting.
This whole best fit approach is fun but holy shit do I never want to look at any of the pictures it creates more than once.
Wow these Juno ones are actually really good…
AI may be good, but I bet it can't do farting! We will always have that above the robots!
It ain't wrong, but we need a model that does the opposite of what it currently does.
It really would revolutionize data storage tho, at least cold storage.
Omg, this is both the stupidest idea, and sheer genius: https://github.com/philipl/pifs
Use Pi to store data, all you need to know is the start and end index of the data you want.
Then compute Pi to that end index and snip your data out.
I mean sure, it requires an unimaginable amount of compute, but hey, essentially unlimited storage.