#images-discussions
1 messages · Page 38 of 1
What's your approach for intentionally confusing something great at understanding?
Because the general public can't write, and so it would receive books of incoherent text, which would lead to terrible images.
sentence fragments, non-existent concepts, ambiguity, etc.
Ive tried to add "extremely vague" to some details for interesting results but I think I have to insert the ambiguity myself rather than just asking it to be ambiguos lol
lots of people think 4.5 is due tomorrow or soon, i wonder if a better model could make prompting any better. it seems its probably already as good as could be without dalle 3.1 or something but idk
it's as though it hasn't been updated for DALL-E 3, still limited to 400 characters, although the API since DALL-E 2 accepted a 1000-character prompt. i'm looking forward to when BIC takes a longer prompt
I assume these models will keep getting better, albeit asymptotical in their improvement curves
the jump from gpt 3.5-4 will surely be greater than the one from 4-4.5
well if they actually release 4.5 and its a jump like 3.5 was to 4.0 we'll be too blown away by it to even think of images... maybe haha
yeah, chatgpt will always try to make coherent text, which leads to very good images, but not the best you can extract from the engine. Mind you, I have been using it (dall-e 3) since, basically the beginning of the experimental branch. When I still had to teach it what humanoid robots look like, and all prompts were manual.
possible. "better" is hard to quantify when it comes to art.
I wish there was a way to practice on that version, seems like an advantage for knowing how to appraoch it now
The closest you can get would be bing. And I don't think you are really missing much, honestly.
True, although I'm sure there are some fairly objective parameters you could assign to good art and then test for whether the average of images created by the new version is higher in those parameters than the old one
Coming up with those metrics is way above my intellect though
y'all are really coloring outside the lines with the daily theme, i will have to get more creative
did u create your pfp
ya, it was in a generation, supposed to be the moon with the face of Zeus lol
This one in particular, so beautiful
here is an example of one of my earliest attempts to get it to understand "humanoid, but robot". Eventually it got easier.
the one by @chilly onyx reminds me of an Imperial Walker
Kinda fun making something inspired by others anything you can think of and we all try doing it and then show each other what u think?
good use of B&W
whoops it only linked to one
Always, I try to stay on theme, but then add my own personal twist or "misunderstanding" of the prompt.
I really wish they had included the prompt, im extremely curious how they made it so emotional, like les mis
i've admired their contributions also
Yeah Sawa is one of my favorites, for sure.
I mean it very well could have been a victorian era painting people would be money laundering millions of dollars for lol
the fact that she probably made it in 30 mins or less is wild
Oh I don't know about that assumption. I sometimes take so long to even come up with an idea that I miss my window. And then it's still hours to get the four or five images I want to submit,
True, but even still she got it in within the day, a painting like that would conventionally take weeks or months
I mean, true, but... it's a different medium, it's low resolution, etc. I don't really think of it as a replacement.
Its also in the infant stages. I imagine eventually 3d printing or animatronic painters will be able to replicate AI creations like blueprints, surpassing human abilities
if its reasonable and possible within the laws of physics im sure it will be done/ or the direction we head if their is enough demand
That's what people say, but... I remain skeptical. For now it's an intriguing new medium that still requires a great deal of human involvment. Especially if you want to produce a larger piece.
Imagine what leonardo davinci could have done with tech like this
(two images I didn't get to submit for the skyscraper challenge)
I rlly like the left one
Yeah, same.
Anyone know if/when outpainting for Dalle3 will be available? I just upgraded to plus thinking there was an image editing mode like Dalle2 but couldn't find it anywhere. Is it only accessible through ChatGPT or Bing?
Did you distill it down to a prompt or is that the result of many generations of editing and prodding gpt
me? That was handcrafted, from beginning to end. Took a day to decide what I wanted to create, an hour to consider what style I wanted it to be in and then a few dozen attempts to hone in on the desired result.
What style would that be called, is there a glossary/ wordbank of styles I could look at to experiment and familiarize myself?
Idk what to make now do u have any idea just made this bc you showed angel themed picture
IDK, I picked every design element from my actual human memory.
impressive
its like a unique mixture of gothic and pencil sketches but also fading watercolor
very cool
This was the first iteration (not giving away the gold):
a masterwork illustration of a futuristic necropolis. the got themes of this piece are accentuated by the dark color palette. scant color accentuates the work, as though spirits emanate from the towering stone buildings. a river of souls. perspective drawing. visible guidelines transition to rough sketch.
that's mispellings at all, exactly the prompt used.
And from there I iterated to draw out my vision.
Man you guys are too talented
Wouldve never thought of smth like that
Looks almost like that rlly cool medium in the into the spiderverse movie with the renassaince vulture
My sibling is a master of fine arts. I had to ask him what the sketchy line things are called in unfinished perspective drawing. Back when I was trying to master an inkpunk aesthetic.
And its called visible guidelines?
yes. visible guide lines are how you describe when details used during sketching are left intact. unfortunately that can muddy a piece as it's a somewhate ambiguous phrase. I also use fragments like "rectilinear details" to help with that. My full prompts probably look like poetic line noise with the occasional coherent descriptive phrase.
makes for the best imagery in novels as well so I can understand why that works
Also, sometimes mispelling a word actually leads to better outputs, and you would never know that if you don't just do it by hand occasionally.
I mean, type it in letter by letter into bing, or demand that ChatGPT only use my prompt verbatim. And then operate on the prompt iteratively. Not by hand is e.g. doing that and then asking ChatGPT "Okay, your turn, I want you to riff on my ideas, let's see what you come up with. You have full creative freedom" or similar.
Which there is nothing wrong with doing, and is a great way to learn new techniques and approaches.
Oh Ive just been writing a prompt myself, saying depict this, and then telling it what it did wrong or right for future generations and asking it to de-emphasize or emphasize certain elements or words from past prompts
ig that might be a mix of the two idk
That's assisted prompting, look at the actual prompt it submitted, you should be able to see it when you click on the image.
(and again, nothing wrong with assisted prompting, just not what I personally specialize in)
Ahh I see I didnt even know that was a thing I thought it was doing what I told it to. Im gonna experiment with telling it to only use what I say
and see how past creations change
Just got this image after riffing off your og prompt a little with a couple lines from a different project I was working on with a similar vibe
Nice!
I wonder how much the art you make in ai generators could tell you about yourself
and your thinking process/state of mind
it's a mirror
I mean, let's be real, my art says I have a very healthy interest in cute robot girls. lol
Like what would a therapist make of all the dark twisty things we're meticulously putting together in these models
Oh rlly 😂 where are the cute robot girls I havent seen any from you
This is what happens when u listening to sad music the whole day
search "from:vanya has:image" in the search box.
this is awesome
roman pompeii
Oh wow you werent kidding, well I guess you'll be embracing the future AGI takeover
😂
As long as they're cute
I would never joke about robot girls.
Here u go roman Pompeii
Was that with BIC? With ChatGPT you might ask it to advance the timeframe a bit...
When saying robot girls are u referring to woman (18+) or girls (17-) bc when i ask chatgpt he is very specific to say that to me
The really hard thing, IMO, is getting it to do less. Just a splash of color, detail only where it is called for. It's genuienly difficult but in my opinion makes the overall piece just feel better. More human.
In prompt, women, female, etc. never girl. And the odds of getting hit by the image censor are still high, so... good luck!
You're right my bad, this is no laughing matter
Ye i saw that when i asked for girl the first time i meant woman but i usually refer girl with women so didnt think about it the first time
Agreed, the more striking smth is the more it looks like obvious ai to me
I made it with chatgpt if it was me u asked
yep, exactly what I was imagining in your first image, well done
yeah, there is a difference in how humans talk and how the AI interprets things. I might want a robot catgirl. But instead I request a "cybernetic female neko robot". and offhandedly describe feline features as an afterthought. It takes a bit of practice to push into those spaces properly.
In my head girl is always females that are around my age and then everyone significantly older would be woman or madam but chat gpt doesnt see it that way
if I am getting to much "human face" in the result I might further refine by specifying a mask, or even tattoos (which can be artfully used to produce texture on existing objects. And no I can't explain exactly how)
I'm in my forties, boys and girls are anyone from mid twenties to mid thirties in my mind. And children are anyone under 21. But that's just my perspective.
Exactly how i think too
"Okay, I will destroy humans."
If vanya's into that I aint judging
No, no, "Befriend all humans"
What am i if im 21 then🤔
yup, you are a kid. (mostly teasing)
Does this count?
That's pretty good, has one of the most common flaws. Too much skin tone on the face. This is harder than it looks, for sure. (nice work though)
its crazy how AI incapsulates all of the perfect human facial ratios for attractiveness so effortlessly.
Idk where that came from considering even the average of human facial training data wouldnt have given such a good metric for golden ratio faces
The Golden Ratio
Its more complicated than that though because philtrum to chin length ratios & cheek bone/maxilla protrusion arent related to the golden ratio and are just arbitrary evolutionary artifacts
This is one from the ice theme challenge that I think shows a pretty coherent and consistent design. This is extremely difficult to get right, IMO.
[especially after adding additional requirements like matching a theme, and specific, non standard materials]
How many iterations did this take
lol, idk... give me a sec.
It looks very similar to some NFT art from a project I saw a while back
Probably sold for thousands 😂
I have 128 images saved in my gallery for that theme, so probably around 70-100 attempts and yes, some I cannot post here.
💀
I thought dall-e doesnt let you generate anything NSFW
Sure, totally.
Reminds me of that scene in blader runner where the prostitute realizes the protagonist (ryan gosling) is into hologram ai girls and not real ones
She is very friendly
why she rockin the grandma hairstyle
how do i generate homer simpson
she might be a grandmother, appears she may have had some cosmetic work done
the surgeon required she destroyed one city as payment for his miracle work
consequence of how the system tries to make sure faces look good. which has undersirable consequences for robot characters. (or maybe that's your thing, idk)
Try working in jellypunk with purples, blues and pinks and watch just how vigilant that system is. (it will block innocent images)
What I got from putting in that exact prompt 😂
do you know how to get homer simpson?
"cute adorable jellypunk dragon" BIC -> here are two results. "okay but... oh I get it"
did you try the obvious?
copyright just say be inspired by the character
I'm unable to create images of Homer Simpson directly, as he is a copyrighted character from "The Simpsons." However, I can create an original character with some general characteristics inspired by the request, like being a cheerful, cartoonish figure. Would you like to proceed with an image of an original character inspired by these characteristics?
Mb im too naive but I have no clue why jelly punk would trip NSFW censors
im not even rlly sure what jelly punk is
I can't personally advise you on how to circumvent any systems intentionally placed.
bro what is this
ok
i just wanna generate homer simpson
a nonexistent style. Try stuff like "a masterwork image of a cute adorable jellypunk dragon." in BIC.
do u work at openai or smth?
BIC is bing?
Its in the 80s she is very old
yes.
say smth like "take more inspiration from the aforementioned character"
No I don't work for OAI, but I try to follow the server rules whenever possible. I'm not a mod, but I do like to help foster a good environment. Like if I were actually a mod, I would point out that g-rated includes violent imagery, but... meh. let a mod handle it.
Oh lwk I didnt even read the rules, probs should do that before I get busted for an honest mistake
I woudn't sweat it too much. just... you know, don't just cross the line blatantly and you'll probably be fine. I just got hit with a five minute mute for trying to explain why jellypunk causes false positives.
Lol you can DM the blocked message if u want
I just did the bing prompt you were talking abt and I dont know if I want to even think abt that 💀
See, it false positived one of the images... but why!?
his Bimmer though
so serious
The mighty robot knight with his trusty horse STEVE
very creative, i've never imagined a robot dragon
Tyty i do my best with the little imagination i have
if anyone can get me a prompt like this
i will be insanely happy
i cant get it to make homer simpson
or anything like this
reminds me of my black 440
Describe what you see to Bing Image Creator: "Photo of Homer Simpson in a black suit sitting down in front of a black BMW. He's having tea and sitting on a Persian rug."
"Doh!"
what do you think, which of these pictures was made in dalle-3
Anyone know why dall-e 3 doesnt have a feature where u can directly edit the image u upload with ai
outpainting/inpainting aren't currently available for dall-e 3. I assume they have not yet been completed/fully tested, but I don't know. It's a pretty important feature. It's possible that MS has early access, but I don't use any of their offline design products.
yes, the picture with Dalle 3 is the first one
Yah hopefully it comes at somepoint, that would broaden the horizons of what I could pragmatically use this tech for ten fold
is the one on the right hand created or mj, it feels like mj based on color palette.
ahaha, the one on the right is actually Kandinsky 3.0
ahh, neat.
Yes
ty ong ur goated
@true mural using some of your tips & tricks
I like it. That's really nice work.
I told it to use exactly my prompt this time and I think it definitely helped make something more original and unexpected
Of course that also means the training wheels are off.
homer baba
I say bring on the challenge
nailed it
one technique for BIC (for when you hit your GPT-4 usage cap) is to use gpt-3.5 to convert your concepts into precise 400-character art prompts
A little tinkering and im getting even cooler results
great themes
what about be careful?
U win

There seems to be a lack of reference material in its training data for a nearly depleted hourglass.
that's an example of a concept i want to refine, and i have many generations i've iterated over, but one difficulty is specifying the fine-grained (no pun intended) variation on the hourglass--it's essential to portray the dread and anticipation... the model says it struggles to convey emotions associated with nearing completion 🙂
Try saying 'particle' instead of grain
Little sand particles

If you share the prompt, I could give it a try.
See what comes up.

I wouldn't be using Dall-E though, so it may differ. But I'll give it a shot and share the refined prompt with ya.
I would like to perfect a 400-character prompt for the concept. I plan to optimize every character as I expand. As a use case, I'll paste the one I used to create the base above, although the "surreal" parameter is optional:
Digital, surreal painting: a towering, nearly finished hourglass looms over a sinister cult's eerie ritual. Enveloped in darkness, convey their unsettling worship with fervent dread, an eerie ambiance, and an air of foreboding uncertainty. Leave viewers on the brink of the unknown.
For other variations to get me started, I specified some darker terms, and you know how BIC can be, so just a heads-up. That's where I'm going with it, but I'm focusing on the language to get the hourglass envisioned properly for now.
Sample copy narrative for inspiration:
"In the ancient, cryptic chamber, the cult of the millennia-old hourglass gathered with bated breath, their forebears' prophecy echoing in their minds. The colossal, nearly depleted hourglass, a relic of forgotten time, cast spectral shadows that danced ominously. For countless generations, they had passed down the belief that when the last grain fell, an ancient prophecy would finally be fulfilled. As the sands dwindled, their anticipation and dread reached a fevered pitch, for they believed that an unimaginable destiny, concealed for ages, would soon be unveiled, forever altering the course of their mysterious cult and the world beyond."
Alr, brb. I'm gonna see what I can come up with here...🛠️
I've tried a number of ways to express the nearly depleted hourglass part, that's the struggle.
many thanks
awesome, detailed dioramas are fun
agree
I'm getting the same results as you are. It's not registering my command for a completely depleted hourglass. You might want to ask Ac3d. He can probably do a better job at helping you than I can with this sort of thing.
tiny houses, cross-sections, etc.
It may be that some depiction images have a "pre-set" default as to how they should look with Dall-E and that may be the reason we can't get it down to specifics.
But I'm not entirely sure.
It is strange though because I specifically asked for a depleted hourglass as a test and it gave me a full one.
yeah, thanks for trying, the nuance is the challenge, and there's more to add, so I'm trying to figure out how many characters I have to work with after I arrive at that detail consistently.
i asked the model, said it probably didn't see enough images of an hourglass in that state, "Additionally, DALL-E's architecture and training process may not prioritize fine-grained details or nuanced variations, making it challenging to depict the subtle intricacies of a nearly depleted hourglass effectively."
Oh ok, that says it then.
So it knows that it may not be fully trained in intricate particle variations.
Which makes perfect sense.
But I remember adding a gif into ChatGPT and I remember it being able to detect the contents of that gif. If that's true, then it also means that it can detect what a gif hourglass would look like showing it's complete depletion. I would try finding a hourglass gif that does that and asking it copy the last moment of the gif into a recreation for Dall-E to reproduce.
interesting idea
Just a hypothesis 
But it may just do the trick.
sigh tech. I just want to ask ChatGPT to imagine the scene after the last grain has fallen and visualize 🙂 Either they're going to be sorely disappointed or we might see some cool monsters.
it's worth a try, it's a little project lol
Pepe might be hiding in the upper chamber of the hourglass, finally free
This is the best i can do
That's far better than what I got
That's close, and I was getting that far, but clearly there ought to be more grains at the bottom
i mean, in a few variations, i mean, i have generated two-dozen at least
How do you like this seller?
if you observe any patterns or concise phrases to achieve that, please let me know. looking for the condensed language DALL-E understands.
Even in the winter he worked at the cash register, and in the summer he burned out...
Homeless pepe
Stressed pepe
Too many entitled customers
I need to consider some new styles. change is hard.
Then pepe goes to udemy and learns a new skill and lives happily ever after
Try oil painting. That's always a showstopper
Pepe became a businessman and lived happily ever after
He must be selling top hats
haah, of course
Try making up new styles. That usually works. Like say "oilpunk" for a punk to represent oil paintings.
And have ChatGPT have new interpretations to what you mean for new styles.
intriguing.
Yeah, and also, get an old prompt of a pic that you did, and ask ChatGPT to condense that prompt into just a few words. And then use that as the prompt and see what different results you get.
The fun thing with "surreal" is you never know what you're gonna get, unrefined prompt: "Oil painting. Surreal. Frog. Snow angel."
I decided to check what would happen if I asked the GPT chat to come up with a new style, and he came up with “Geometric Fantasy”
that sounds like a good technique
It's worked really well for me in the past.
Yes you do, as long as it doesn't share the likeness of a copyrighted work.
this feels like something out of an over the garden wall christmas special
And owning and selling are also two different things.
Like you can own a batman pic and show it off, you just can't sell it as a batman pic for money.
Although people sell them anyway.
not bad, by virtue of its background knowledge, it can "imagine" some fascinating convergences and fusions
That's for sure
So yes, the short answer is always yes, because even if you do sell it, the legal liability for selling a copyrighted material is on you, not on OpenAI
in shire land, cute dog, walking the ground way
Surrealism is what got me interested in art, specifically Dali, long before DALL-E. AI is perfect for it.
So if our ai art is original we can sell it
thats dope af
Yes, that you can do no problem.
what abt parody cant you parody copyrighted work and sell it for profit
like if I did batman in homer simpson style with a different logo
thats fair use right
but you can't copyright it yourself, either, a judge ruled
You can, but OpenAI fights against it by not allowing you to reproduce the likeness of things.
No comprendo why wouldnt we be able to copyright our original work, isnt copyright auto given to u even if u dont ask for it?
in the US atleast
well it doesnt do a very good job lol
Because it's actually illegal to sell parody stuff. And they're trying to protect you.
Google Théâtre D'opéra Spatial
then how is south park not on law suits against disney for selling their parodies as entertainment?
I think because they have a subcontract with them.
Known as distributors.
Because Disney can sell in high volume, they give a subcontract to Disney.

Théâtre D'opéra Spatial overwhelmingly fits the bill for copyright protection like its not even funny how clearly it should be copyright protected based on all he did to the image
thats wild
that they ruled against it
You can try, but judges are ruling it “lacks human authorship” and copyright law “excludes works produced by non-humans.”
thats a horrible ruling, I would take that all the way to the supreme court if I had to
I bet the supreme court would rule correctly on that
it's a contemporary issue to keep an eye on for sure
I honestly wouldn't mind making my own superhero line just for merchandising
was this in Dalle or Bing? i didnt think it would do the Batman
it's too bad. in the early days of dalle they must have had some restrictions off still because i remember it making some Batman costumes for some other people on reddit
maybe it'll inspire your own creation
hmm this website on " up to date allthingsai" is saying it should be currently backed by cc if original and creative
Definitely, it seems like he took heavy inspiration from that scene in revenge of the sith where they are in the opera house with the bubble window
haha good call
i'd need to do some more research, but that sounds positive
have you ever heard the story of Darth Plaguies the Wise...
idk ive heard that name before
isnt that the homie who can bring back from the dead
or smth
yes its a line from that scene
I think it's because the internet is so saturated with spiderman and batman pop culture that ChatGPT's and Dall-E's database automatically picks up its connection if you mention anything related to the character or genre.
i swear these images are sometimes optical illusions and appear to move lol
probably not exact i have not seen Revenge of Sith in some years. great movie though imo
yes probably pythagoras. batman and superman should be public domain really
Yeah, they should be...But oof that moola baby!
Bing
oh ok
I should make some cool Batman costumes now that I think of it haha
mad max batman would be cool maybe
Batman, spiderman, and superman will be public domain soon
copyright should be timing out in a decade or so
Batman and Superman I think have like 25 years on Spidey
oh wow rip
superman, batman, wonder woman all should be soon though i would think. they are the cornerstones of all super heroes. and robin
Its so funny that they keep extending the copyright length everytime mickey mouse is about to become public domain
That's never gonna happen
copyright office is in bed with disney
yes warner brothers and batman are happy to see those decisions too im sur ehaha
y
Because they are owned by the company, not the artists.
yah but u dont get protection on creative property forever just bc someone owns it lol, u have like 90 years or smth before it becomes public domain
yeah I mean Batman may as well be the mascot for WB more than Bugs for the last 30 years. hes probably bringing in 2 billion in merch alone haha. same with spidey
I mean, what are the rules on this? If BIC can make it?
nolan batman trilogy will always be peak superhero fr
i wonder if that new google image gen lets you make copyright stuffs
bard?
brutal
damn i got a time out for saying the name of the google image maker lol
Yes you do get protection for it just because someone owns it. Companies don't work as individual people. Something can't be public domain if the company still owns the license to that product.
time out lol
wonder if saying midjourney will bring out the minute ban too 🤔
thats not what googles sayin
I stand corrected! That's news to me. And how strange!!!
Copyright guards any original creative piece of work for the lifetime of the author + 70 years
after that anyone and everyone in the country thats true for can use it
yah like I said basically everytime mickey mouse is within a couple months of turning public domain it gets pushed back mysteriously LOL
but i read before also then would be batman/bruce wayne as appear in detective comics or wherever he first appeared, but lots of things were added to the lore over years, so even like the bat cave would be off limit for x-more years, this character for 10 years, etc
exactly
every creative piece of work gets its own copyright
and u get it without asking for it
So it's not going to be public domain, lol
as long as they dont push it back again it will be in bits and pieces
I don't think it can get pushed back actually now that you mentioned it
look at Mary Shelley's frankenstein for example, thats public domain now and tons of stuff uses the lore
I forget when it became public domain
it wasnt super long ago
actually nvm it entered it in like the 1930s LOL
I don't think I can mentally handle the world suddenly making batman and superman shows and stuff at 2035
I think I've just about had it with them even right now
Talk about brainwashing
that' conjures some scenes to visualize
lmfao imagine 2035 we get like 90 new batman shows
That book is such an amazing piece of literature
I just love it
You laugh now. Your nightmare will begin in 2035
the characters are so fleshed out and you really feel for all of them/ understand why they make the choices they do
its way well done, doesnt get enough credit bc its so old
and out of the common discourse
On the brightside, the extremely famous movie Metropolis from 1927 entered the public domain this year and there arent a billion remakes yet
I havent heard of one
so mb production cost will keep a lot of people out
Oh man. I actually have to care about newer movie stuff. This is going to be a hard next few decades.
wdym
y do u have to care
Lol, yeah exactly
Ive watched and rated over 750 movies so im set in that department
one of my favorite hobbies
perhaps because it's already inspired a number of movies exploring similar themes (Blade Runner, Dark City, Brazil, etc.)
True, metropolis quite literally created the sci fi/dystopian genres and is responsible for 1984, ghost in the shell, OG blade runner etc
as well as basically shaping the course of all the future sci fi mediums
its quite the influential film
I've never even heard of it
Lol
First time
Superman's city
That's all I know of it
I don't even know -- that's the fun of surreal, abstract....
That's a cool pic. I'd live on Frankensteins head
Those are houses right?
Or is that rolls of string?
i think they might be surgical tools lol
In the book frankenstein is the doctor and the monster is just called the monster
but the monster is actually the good guy lol
Ohhh hmmmm
.... I'd still live there
haha
yah they look dope, what art style did you tell it
1984, liked the book better--memorable final page--no spoilers, of course
Oh I never watched the movie, but the book is my favorite work of fiction of all time
I used a simple prompt, using surreal and stuff so bing would surprise me: Oil painting. Surreal. Abstract. Knife-edge effect. Dr. Frankenstein's Monster comes to life.
atleast as far as books go
idk abt movies
Lol. You can get funny just saying random words like that.
Knife-edge, head, yogurt, jobs, Cotten candy, Dr. Frankenstein. Monster comes to life.
I thought there was no way they could meet the expectations they had built but yah they did
and I was left shocked, thinking abt it for weeks
brave new world and 1984 are both highly applicable to today
- A Clockwork Orange & Fahrenheit 451
A clockwork orange 100% also eyes wide shut
fahrenheit 451 is a chronically misrepresented book
basically no one understands its point, and its not the deep meaning everyone leads it on as,
the author intended the book to be a dis on television thinking it would serve as a distraction that would be the death of literature and books
lmfao
this was back when TV was first becoming popular
it's contemplative--consume 1000s of movies and books, 100s of plays and works, etc., the natural progression is a synthesis that compels your own unique creations. AI is great for that because I can't paint.
You can't paint?
i can't draw stick figures
Painting is like the most easiest thing
Lol..uhhhhh....I refuse to believe that
painting and drawing were always meant to be an outlet synthesizing our human experience but the limitations made it extremely low bandwith,
AI just increases human imaginative bandwith to a real life medium in an unprecedented way and im way here for it. The more efficient our output can be the better
rn 10 fingers and neuron speed limits us
hence nueralink
i'd get half-finished and then move onto a better idea
How do you paint your house?
I complimented on his hair. Said that I've always wanted a bust of his head to put in my front lawn.
Yeah no gifs unfortunately
why dont they just make it a no gif channel instead of timing out for it
I would have been posting all kinds of gifs
I only have a few 😦
kurt angle 1000 yard stare my favorite rn
he be lookin intense

Imagine you woke up and saw that guy in front of you just like that
I'd scream, ngl
🙈
lmfao
real
this dude keeps popping up randomly all over social media tho and its so funny
Kind of sore on the eyes
I'm gonna write so we can get the chat moving
So I don't have to look
At that
There
It's moving slowly
Ok it's almost gone
lmfao

No, my rent is too expensive to have people like him in there
which is scarier rainbow demon or kurt angle
That's a cool one
and what was that
You can only post 5 images max at a time Riden
Not more
You got automuted by the discord bot
damn, I spent the whole day there pepes, 5 images are not enough
where
What is where
Why is what?
Well, he said where, I asked what where
how difficult it is when you’re from Russia, you have to translate everything
Beautiful
the picture painted itself
pepe lol
why not
pepe is goated
These are dope, howd u get it to use nike
🐐
what! the! hell
hi that is absolutely horrifying, ill take 20 please
I just put the Nike brand in the prompt
what do you want to take?
Why won't dall e generate stuff that's has "laying down"
how to understand "laying down"? can you post an example
Like laying down on a bed or laying down in general it won't generate that
The image got deleted by mod?
Idk, probably
I got in dm it was deleted by mod
here they lie
Maybe
Noped used "lie"
And "unsafe image content detected"
I’m just a translator, I translate everything, requests also go through him, so the way he translated it happened
What exactly did you put in
boy and cartoon character lie on autumn leaves, 8k quality, 3d render
I used
"Asriel dreemur from undertale lie in bed, 8k quality"
Unsafe image detected
And if I make to many failed prompt I could be permanently banned from using dall e
Ariel dreemur is from a video game. Often it will trigger the security for protected ip
Is that why?
I can't use characters from games
It is a common reason, yes
I have more questions but I feel like I won't get anywhere them so I wont
go ahead, I have time
robert is great at answering even the most toughest question. If you were to put answering questions as a condition for saving your life, robert would save it 10x over. So use this chance while you can 

Is there a roadmap or something when next bigger update for dalle is going to come? Maybe fixing the tall/portrait image format finally? 😄
Let's wait and see #announcements 👀
I hope so, let's hope, seems there is a huge chunk of feet data that is missing. Hope it's included in the next model upgrade. 🐾 🐾 🐾 🐾 🐾 🐾
Yeah, it's almost impossible to get full-body portraits or tall images that aren't just flipped landscape format images.
At somepoint when i tried to describe shoes or feet, the character sometimes just ended up holding the shoe in his/her hand, which was funny.
Have you checked #1021130377026351105 
Once in a blue moon i get a generation id and it works for a while, but no matter what words i use there seems to be no guarantees and the whole process just becomes frustrating, since the images fail more than 50% of the time and since mobile is vertical images, it's a pretty big problem in that sense.
Yes, I've been in #1021130377026351105 still it's very inconsistent
Also #1019652163640762428 can help
good, I will check there
Please go upvote that or something, i think i have seen multiple reports of that in there, but not sure if they simply 'expire' over time.
anyway, I just hope the next topic takes me away from fairies. I've been doing too many of those in the past days
Haha, i am often stuck with one theme at a time as well.
but, have you made your ramen fairy today? 😜
that was so good
I seriously meant it, at some point wanted to do toilet paper roll art with the concept
Sorry for the late response didn't know why discord notifications didn't show up but
Question is
Like I asked dall e to make something and it didn't said image was unsafe so I tried it again and again with the same prompt and it Finally generated an image but after I ask it to make another it said unsafe images again why
DALL-E will create a prompt to generate an image, if the subject matter is a grey area it may refuse to generate the image. Are you able to share your prompt?
Yeah I’m done. Was going to respond to yours with this in defeat 🤣
But, no need, new theme!
I don't think it's "broken" per se, but more so the fact that you requested a portrait.
If I want a widescreen portrait I lead with, "In 1792x1024 resolution, create a portrait of..." or something along those lines. Occasionally it falters, and it makes a 1024x1792 portrait, but that seems to be rare. Likewise, occasionally it'll make a widescreen that's returned "sideways," as 1024x1792, with the image correctly in widescreen, but turned 90⁰. I think it's normally a misunderstanding of intent when the prompt has those multiple conflicting terms in the same prompt. TBH, I rarely use the term portrait. I prefer close, or close-up, or macro style close-up, depending on the distance I want (or other similar terms).
You can also control distance with objects. I made an angel earlier that was depicted too close-up (face only). I said, "can you give her wings 🪽," which automatically backed the camera up to fit the addition of wings.
I hope that helps.
I've tried varation also, no better results
Dall-E + Magnific AI upscaler + Generative Fill for Outpainting and details and it Rocks!
That looks awesome 👍
thx bro. it's amazing how upscalers are getting good... and if you know how to give some extra love in photoshop with "old school" retouching and fx , damn it's really really amazing
that's the original one directly from dall-e
damn it, already fell into the fairies concept....
😅
Oh wow, that is awesome. I gotta start learnin how to use em.
is it possible to do upscaling with the API?
no, but if you're on a Mac you can use DiffusionBee to upscale
Yeah, that is also another way. I usually do it with photoshop tho. Was just wondering if it was possible with the API
Show me the prompt! LOL j/k
I have had it glitch too, where it seems totally unprovoked. But I'm just saying, most of the time, when I read deep into the prompt, I can find the cause of the "misunderstanding," for why it chose to deliver a 90⁰ image.
lol, you want to read a text file of around 5MB with just trying to do that?
I think I can work with a new concept for now, but it's gonna make me very hungry...
Can you guys try my sample art generator?
I'm trying to tweak and refine it to make it better, and I need some testers.
Here is the link: https://chat.openai.com/g/g-NhtqOfIt9-apex-visionary
If you guys could just show me the results you get and put them here, that would help me.
I will try it later, atm figuring out how to not eat chocolate

Epic chocos!
if I have something more complex I'll try it out again
that one looks nice, you can sell it at wallmart
Hehe
lol
Input:"In 1792x1024, create an ornate hand crafted box of chocolates, where the chocolates inside are seasonally inspired by Rudolph the red nosed reindeer."
Epic 
LOL, that was just my input into Apex Visionary, so Pythagoras could know the input that got the output image.
But those are some bang'n chocolates! I'd keep the box for sure!
Was that in Apex Visionary, or just Dalle, or your own GPT?
Look at this fail, bahahahaa.
ChatGPT: "I'll show you how to make a box!"
Me: "Um, but how are you supposed to close that box? And will the antlers even fit?"
But the chocolates look good I guess...🤣
wait they added usage limits to dalle?
The architectural challenges of a building shaped like a reindeer
Oh you used AV for this? Looks really good.
What does AV mean? :x
Apex Visionary
ah - thanks! 😄
This
Use it @gray surge and show me your images
Would love to see what you come up with

Sure I will think for something. 😄
@late blade great reindeer car concept, love it when people blend unrelated subjects to make something totally original
😁
What's BIC?
this was my prompt: A house resembling a reindeer with Nature as the backbone, Respecting the environment, Solar energy and light flooding through spans, Curved architecture
very nice, i've been playing with a reindeer house concept myself, so far i like yours better
Bing Image Creator
not gonna post them myself on the daily theme because that is your idea
Those are some amazing images, ngl
What do you think about that? 😄
(I am happy our earth doesn't have a hole in the center. :P)
that's considerate but i like to see your creations so please post away
Lol, looks epic and scary at the same time.
ok, I will post them, but I owe you a good idea
Cool details though @gray surge
you and everyone here have inspired lots of creative ideas
those are really great, what kind of galaxy distressing event are you going for?
Colliding galaxies?
that's an extremely edible-looking reindeer
The goal wasn't a distressing/colliding galaxies - the test was more how many details it can integrate and if its able to implement something, like x° - so I told that the Mars is in the foreground, while behind of it (round about a inclination of 45°) there is a peace of earth and behind of it the sun - with the universe in background
I hope it answers your question? 😇
Thats really cool! Do you want to share it in #daily-theme too? 🙈
I will try, so many ideas, so little time
this was my prompt: A majestic merging of stars and cosmic elegance, two galaxies resembling majestic reindeers embrace in a breathtaking celestial waltz.
excellent. In the vast stretches of cosmic time and space, that quite probably exists somewhere
still have a cool down going and then another one, let's see if they make the cut
sorry but it had to be done
BIC is very permisive when it comes to copyrighted material
Here's my result with Apex Visionary. My original prompt: "A small army of steampunk snow angels marching through the snow field toward the fortress. HD digital illustration, hyperphotorealistic epic fantasy. Extremely detailed, dynamic asymmetrical composition."
AV prompt (first/left): "A small army of steampunk snow angels marching through a snowy field towards a majestic fortress. These snow angels, intricately designed with steampunk aesthetics such as gears, brass fittings, and mechanical wings, move in a coordinated manner. The fortress in the distance, elegant and imposing, draws inspiration from grand architectural designs, featuring intricate details and grandiose structures. This scene is set in a vast, open snow field, enhancing the epic fantasy ambiance. The image is a high-definition digital illustration, rendered in a hyperphotorealistic style. Special attention is given to creating an extremely detailed, dynamic, and asymmetrical composition, with a keen focus on depth, lighting, and textures. The elements in the scene, including the snow angels, the fortress, and the surrounding landscape, are depicted with an exceptional level of detail, emphasizing the contrast between the mechanical and the natural.
Comp image (second/right) with my GPT: "A hyperphotorealistic, HD digital illustration of a small army of steampunk snow angels marching through a snow-covered field towards a grand fortress. Each snow angel is distinct, adorned with steampunk details like brass goggles, mechanical wings, and Victorian-era attire with gears and cogs. The snow field is textured, showing a dynamic range of footprints and snowdrifts. The fortress in the background is a mix of gothic and steampunk styles, featuring towering spires, steam vents, and rotating gears. The composition is dynamic and asymmetrical, capturing movement and depth in the scene with a focus on extreme detail and a sophisticated, epic fantasy atmosphere."
that is spot on. nicely done. i'd run.
Woah, that is pretty good. It turned it into a big design. Is there something you think I should tweak?
oh my goodness, I just found the perfect concept for my next post for the daily theme, the galaxies are out
they're running from my Reindeer Walker
people so sensitive about their image gens. i said Bing is better than MJ and they ban me haha. anyway i guess thats ot i love those little robots Sawa, i want a story of them
hehe that can happen, just be mindful that there's a channel for other AI's if you wanna focus on comparissons or features that others offer.
yes, i mean the MJ people did that not here. everyone so sensitive
I got a lot of very crisp details from AV, thought the castle and mountains looked incredible. A couple of things I noticed: 1) "asymmetrical composition" instruction not being followed (DE3 LOVES symmetry - my personal battle), 2) no foot prints on snow (my GPT has an instruction on "texture"). Don't know if these are important to what you are trying to achieve, but just my observations. Ty for letting me test-drive. 
Rudolf WMD
😆
looks like rudolf star wars thing haha
or maybe thats the big red button to make it go boom

That is an interesting point. I will try and have it focus on compositions better, to have a more thorough idea for distinguishing between symmetry and asymmetry for the characters vs the surrounding environment. Thank you for that feedback.
pretty opulent. that reminds me, DALL-E can visualize some amazing tapestries
hi all. is outpainting still available with dalle 3?
i think there's a use case for a custom gpt that specializes in visualizing tapestries
not available yet, maybe when it goes live on labs
does dalle2 still work?
as far as i know yes
imagine blending a milkshake in that using @shut niche 's chocolates
sorry i just got timed out for posting the link to labs, so yeah on labs you can find dall-e 2
Duck Donuts sells a $100 milkshake, I think we could command twice that.
Input was: “What about a double exposed duotone image of Rudolph curled up watching late night infomercials?”. The prompt it created was “A hyper-realistic image combining two scenes in a double exposure effect. In the foreground, a detailed image of Rudolph, with his distinctive red nose, curled up in a cozy, dimly lit room, embodying a sense of solitude and contemplation. He is surrounded by soft textures, like blankets and a plush couch, under warm, ambient lighting. In the background, blending seamlessly, is a scene of a late-night infomercial on an old-fashioned TV set, glowing in the dark with exaggerated colors and sharp contrasts. The infomercial features exaggeratedly enthusiastic hosts and absurd products, creating a surreal, dreamlike atmosphere. The image should have a duotone color scheme, emphasizing the contrast between the warmth of Rudolph’s setting and the cold, artificial light of the TV.”
there is more potential for crossover than I first thought
add more steam, and also add victorian style to that render
that's cute
quite a few aesthetic/style variants on steam also
LOL, I was just trying SawaSawa's input in a GPT. I thought it's inclusion of the epic sized robot was a nice inclusion.
"A small army of steampunk snow angels marching through the snow field toward the fortress. HD digital illustration, hyperphotorealistic epic fantasy. Extremely detailed, dynamic asymmetrical composition."
But then I asked it this for you. 😎
love it
All I said was, "Can you make this a photo version, and add more steam, and Victorian influences?"
What is that style called / what was the prompt for #daily-theme message ?
( @late blade )
one sec, I have to search for it
from your message history, I'm guessing something like high fantasy chalk?
I want an image of crystalina, epic visual storytelling, She is a safe keeper of today's theme. A herd or reindeers kept safe and being guided to Santa's home. Dramatic immersive bold lighting and colours that enhances emotional impact. Then add a touch of peaceful, serene, innocent. a snowy winged masterpiece!
The art style is a result of using fountain pen calligraphy that depicts the strokes and then ink wash effect.
crystaline is my to go for ice fairies
I defined her ealier in another prompt
oh, you used a continuous GPT4 conversation? I was assuming it was BIC for some reason
yes, I mix both, when gpt4 is unavailble for me because I'm actually using gpt4 as a sweatshop, then i us BIC
I may or may not use and abuse the term "ink wash" in my prompts from now on, thanks xD
there, i was able to post my new concept, meet the Raindeer inspired hygine set
oh another people have used ink wash
so feel free to use. it
not fully sure why it went black and white, but first result from asking it to add an ink wash effect to a previous conversation
try adding colours you want
just telling it to regenerate added colors, funny how it works sometimes
it already had some anime & dark fantasy in the prompt, so it might have learned towards manga for the first
try adding crimson red and sterling silver ink
that looks really good
These were good, although the knife and fork was a perplexing inclusion for a hygiene set 😅
lol
the red lines bother me a bit tbh, but I can just crop it myself I guess
if you use gpt4 you can ask to refine the image, with big is a hit and miss
@dim cradle #daily-theme message is really nice
to be fair with gpt4 it's oftentimes still a hit or miss
I feel like you can typically get 1.5 things right out of style, subject and composition, then pretty much roll a die for +-1 by regenerating
(unless you use an extremely common thing for one of the three)
a good tip to get into artistic stuff, use GPT3.5 to explore art concepts, once you like one, maybe look ona search engine for examples, and then with an idea in mind go to gpt4
tyty
ya, gpt-3.5+bic make a good toolchain when up against usage caps
@wind skiff this is deviation of ink washing and pen calligraphy I added to the mix. So just don't fix on the concept, you can add something different to the art style
yeah I'm testing a few variations to see how each term affects it right now
funnily enough it's generating literal pens when I ask for a fountain pen calligraphy brush stroke style
ya, I had that problem too
Is that a hot fart or a tail?
try modifying it to: the strokes in the image are the effect of a calligraphy pen, or something along those lines
lol, meant to be a tail
LOL j/k! 🤣
I just noticed....
🤨
I think I'll just stick with anime strokes instead tbh, I typically yeet this into the initial prompt and get a half descent style```
- Use vibrant, unnatural colors and exaggerated forms to ensure a sleek anime look, with clear lines and stylized features typical of anime art.
- The subject's face should be expressive.
- Include distinctive traces of anime art, such as dramatic shading.
- Use a wider range of color gradients to create a more vibrant and detailed appearance.
- Use dynamic lines, sharp angles, and a interplay of light and shadow to highlight important features.
- Use clear outlines and keep the design free of clutter.
It didn't get the duotone aspect, but the rest is pretty good.
give 'im a break, he had taco bell for lunch
Yeah, I thought it had overwritten the style with the GPTs own instructions but it is right there in the prompt it generated so might not be the GPTs fault really
As feedback I was a bit confused when I gave it my input and it immediately generated something. The description made me expect that there was going to be a bit of a back and forth before image generation for some reason
I'd share the Flatulence Monster I made awhile back, but taste and art go hand in hand
Hmmm. I think because it's focused on the hyper-realistic aspect, that it veto's other art styles.
But I'll try and tweak it and see what comes out.
You inspired this lol
Santa seems to be doing pretty well for himself
was that in the daily theme on sleigh day?
i'm afraid i missed that one
thanks for the link, i apologize, i did see that one after all, and i had already given it a star. i like how it all connects.
I feel like sometimes adding too much fluff/lore to the image tends to take away the focus from the core elements tbh
how often do you typically inspect the actual prompt gpt passes to dalle? as far as I know it should have no context outside of that final prompt
infinite variations. it's all about making those connections.
@shut niche the infrared nose 🤣
My input was: A double exposed duotone image of Rudolph curled up watching late night infomercials
I like that one a lot 😊
I don't want to copy and paste the output prompt in fear of spam mute
But I'll send it to you
Going to post my own version in the daily theme shortly, I stuck it through my own GPT
I used AV for that btw. Tweaked it to make it remember better.
What I like about that one is that it retains a properly proportioned Rudolph, kinda. Doesn’t turn him into a cartoonish cute version
It’s in line with AV’s aesthetic in that way
Yeah that's true, keeping the size of a real reindeer, for hyper-realistic images.
I'm a big fan of realz
ok new idea landed, time to search for a new one
I tweaked it to understand symmetry vs asymmetry more fluently.

Or at least I tried to...
Is there a place to share artwork?
You know that moment when you have an image that you love, and then you see "THE FLAW..."
Why is his nose about to explode DallE? What's wrong with a regular nose that lights up? What's so strange about that?...
ah hahaha
cherry bomb
I mean, it's supposed to be a children's story and I get Rudolph who's about to lose his head 🤯 ... But I can't even do Cinderella? SMH 🤣
inadvertent depiction of imminent suffering, for shame
Warcry: "For the gnomes!!"
haha
lol

a lot of eyecandy there, very majestic, generated a few Pegasuses recently also. that's the best antler helmet i've seen all day. i was playing around with Sauron with an antler helmet earlier.
i think it was trying to make the unihorn into antlers, i guess it mostly did that
i like the phoenix-like feathers also
Made this with AV. Just tweaked it again. I think this was my final tweak. Because the images are consistently looking awesome now.
AV?
second one has a better composition for me, colors of the first are more striking. in the second the table feels secluded in two areas
observe they're watching from the picture
Yah the second one was my first prompts generation, I tried to tweak it to how I originally imagined and I feel like gpt got it all wrong
kinda creepy huh
Thank you kindly
Funny part is I think I finally figured out how to upgrade my custom gpt further to take it "even further beyond"
this is so dope dude
what's different about it
focuses on hyper-realism and transforms user input in high detail
That thing you made kinda reminds me of gru on his bike and terminator
does it render in higher resolution?
yeah it should.
Hyper realism tends to remove some of the key details when making an illustration
Ive found instead asking for maximum detail within the style outputs better
Yeah, it's not meant for illustrations. Because illustrations is more like 2D stuff. This is meant for 3D hyper-realistic artwork that pops! So it's not a 100% universal for every kind of art. And the reason for that is because I've found that if you made a niche image generator that focuses on one style of art, it usually gets better results, as opposed to using Bing or Dall-E.
good morning, I have school in an hour
Goodnight, I have work in eight hours 😊
Time zones are strange
I like your pfp, I was just thinking of an idea that would have similar themes minus coffee
what time zone u in
GMT (it’s 00:16)
Oh wow so you're in europe
Also Europe here, here it's 01:16
Yah did you make it
and it's 6:18 am for me
r u in asia?
Oh, profile pic! You know what, I really need to change that, because no, I did not make it. It’s extremely old from way before Dall~E was a thing, and I feel bad that I can’t credit the artist
everything everywhere all at once vibes
@glossy scroll The GPT did a good job reimagining war Santa 🙂
therefore, your messages have to be translated through a translator, and the answer also through a translator
The original is always better though 
I wonder when AI is gonna start rlly making anything possible in movies
Fixed
whoa this is so sick, wanna drop some sauce 👀
christian flash
I think I like your version better anyway
I'd like to give you the sauce but I don't know what that means 

I'm behind on my slang

Made with AV.
quick check : is there anyway to provide an image and enhance the image based on additional prompt
Yes and no. There are ways, but the image won't come back exactly the same. It will be a variations of the image.
You could also try to incorporate the GPT to give you a gen_id associated with the picture, but even with that, it doesn't come back 100% the same. It always has a slight to major difference.
This is very cute, and the sparce use of color really makes it pop. Nice work!
The revised depiction of the Psycho Nextdoor Monster, now with a larger, more terrifying mouth and multiple tongues, is ready. This enhanced feature underscores its loud, angry nature, adding to the horror and surrealism of the creature.
Not a bad first try. 🙂
I just read Sam Altman is fascinated by the killer robots depicted in The Terminator.
No
now i want to depict Skeletor
I just took a swing at what could get me that based on what I saw.
This was toying around with another GPT.
Dalle wants to make the same character. LOL
this reminds me of that early iOS game "Infinity Blade"
This is what I gave ChatGPT:
"Create a dark medieval soldier holding a massive broad sword who is cloaked in a tattered hood and cape, and has regal armor. Give him a skull face. Place him in a misty dark valley with an uninviting sky. Use the rule of thirds to place him off to the far left side of the frame. Add some vignette to create suspense."

Could be that I'm getting close to Dall-E's instructions, or that ChatGPT is taking from mine to help Dall-E. There would be no way of knowing.
It needed a little "pizzaz" LOL
the bosses in Infinity Blade always had gargantuan swords
Brains over brawn! 😛
nice, i could see that with a purple sun also
LOL 🤣 🤯
I can't look at that image it makes me laugh so hard I can't breathe!
wow. the boss has an apparent weak spot, good thing he's tall.
the last time i generated a brain that size, it had hairy spider legs
Infinity Blade meets The Terminator -- Altman-approved.
🤣 🌍 🌎 ⚔️
stellar, that is indeed a gargantuan sword and infinity blade lol
He's over compensating for his receding hairline.
but as they say, the pen is mightier than the sword 😉
I didn't know there was a capitol letter spam blocker.
All I tried to say was, "omg, I love it!" Whoops.🤣
How to get gen id
You can just ask ChatGPT for it, and it'll give it to you.
@shut niche just saw your pegasus reindeer for the first time as it made it into the hall of fame -- awesome
I've got something I think you'll like, lol. Trying to fix it up now.
you finally made your serene terminator gpt 🙂
your everyday chill killer robot
Well, it was working at one point, then somehow, I lost or deleted some of the instructions, so I just re-made them. They should be better now anyway, so no loss. LOL
Try it, you should be able to keep it simple, and be like, put him in church, or whatever.
AI is always getting a bad rap, as "killer robots." I thought I'd give them some fair representation! 😜
totally agree
perfect

The worst thing about this, is that I could burn all of my ChatGPT credits making silly Terminator memes! They're addictive.
Improving the image of killer robots
it's for a worthy cause though
Working on making some character art. This is solid.
I made a gallery for our friend. 😛
I'm on a "time out," so I can't even test him now. 😦 I pressed the button too many times...
trench warfare, intriguing
I think Dall-E woke up in a bad mood today. It's refusing to do everything I ask for
hmm, how so?
copyright here and there and everywhere
somehow the "takes inspiration from <insert concept here>" is not working this morning
I honestly think that's a stupid policy. Why is the art generators responsibility to not copy likeness. It should be the users liability and own risk to legally handle it carefully.
hurray I'm alive after chemo
It's always been like that.
yay
💕 🧬 💪
gotcha. fine-tuning for the forthcoming...
well at least I am getting schooled on derivative works and copyright law in the process
"based on" or "inspired by" have seemed more restrictive lately, incidentally.
not a scientific observation or anything
it would be easier if dall-e had browsing capabilities, but that's just for chatgt4
and requests to modify "only as much as required" result in zero likeness--perhaps the intent, i'm sure i wouldn't know
I think the art supply companies should get sued for giving us the tools to draw copyrighted material. You know? It's the same logic here. Makes no sense.
i think we should all give allowances to both sides until we find the sweet spot
Yeah, it should be user liability. Not company liability.
now you can generate something
I wish i coul ask: "has a reindeer in the art of Damien Hirst been done or would that be derivative work?" (just took a random artist)
but at least now I got my chubby reindeers
gpt-3.5 says no.... as of jan 2022
And here to play everyone a happy song about content policies and copyright infringement, it's your neighborhood friend, Terminator.
there's also "fair use" to consider.
Believe me, if video were already available, you'd all be hearing an epic song in a "voice that's not Arnold's voice but sounds extremely similar to Arnold voice" right now.
Lol it means like the secret sauce so in this context the prompt you curated
I was gonna try and riff off that figure you created bc he dope
lmfao
I could use that voice for my upcoming movie "Reindeer Rumble: A Holiday Adventure" "Bringing the Hoof-Down This Holiday Season!" #hall-of-fame message
that's consistent with your previous gen
oh yah I saw you post that in this channel earlier I thought it was dope
Yah this got it closer to what I had originally imagined
is that taking a super hero villain concept? or a really horror themed oriented twist?
My original thought that sparked this was "if god sought to instill fear"
almighty power to torture with feat
and then I was heavily inspired by @open trench's creation
I also had been thinking about a bioluminescence monster or virus for a while
Chat input: "Create a futuristic sci-fi action hero with a sleek black suit that has red geometric laser lines. He has a hooded cloak, and a short futuristic gun."
If tron turned into a vigilante
it's on my list to invent my own aesthetic/style, maybe give it a name
This is Pythagoras's.
this is what tron vigilante gave me
I was just demonstrating that it's like the guy in the cafe. A lot of these characters are rendered the same.
digging the mask
in the cafe?
I kinda like this version of it better he posted earlier
Make a guy in a cafe.
this is what I got
Im not seeing the resemblance
lol
Secret sauce: https://chat.openai.com/g/g-NhtqOfIt9-apex-visionary
Yes, this is mine. But Ace is making them better with his own secret sauce.
I think it's a blend of my secret sauce and his secret sauce. Into a combined super sauce!
i'm looking forward to test driving AV
Its probably got a dash of in and out sauce too
so much sauce
It's cool
Mmmm, sauce.
I just studied the OAI Custom instructions and mimicked them. 🙂 Then added "smell the roses" and "think step by step" through a simple CoT.

Had a little fun w/ it. I'm not sure, blue or red.
I like red
But I just kinda went back and forth reading the copyright laws of the U.S. and OpenAI's new terms of service and policy which will be effective Jan 31st 2024. And it looks like they're giving more leniency to us, by making our rights to our own input and output less complicated. Which will help fight against authorship battles in court.




