#images-discussions
1 messages Β· Page 58 of 1
welcome to "i can't afford chatgpt plus because i'm broke club", @olive sonnet !
Nice and ty
but wait--that's not all
Umm
i mean, there's no title
I see it
Ok
and that's just one possible variation
Aquadaire
you want that also?
Lemme see
and a silver key?
Does the silver key matches the overall tone of the image?
i'm not really thinking all that deeply about it... i haven't even read the prompt lol i just copied and pasted the text from the image from an app.
but i mean, in for a penny....
Oh
let me hear your thoughts about this composition...
I think if the key is placed at the center of the image to fit the text
the hard part is getting the spelling right
that's where i'm not prepared to make more gens
could u make the text curved from the top or bottom of the key?
Iβm afraid this is an example of why a sub is important for your work. At this stage of image creation several gens are required to perfect it. Thatβs not something I can invest in sorry.
It's okay
Itβs so worth it if you can manage
Ok
but yeah, if it's for your work, do the background and maybe add the text in another way
like @dim cradle pointed out
I might use other ai image generators image to image then edit the generated image and see if it work
Ok
But thx for helping me
dys that kitty is soooo cute
caffeine overload
mr. bigglesworth dali style
it's giving me lisa frank vibes, i love it sm!
See, cute kitties are the best delivery method for every kind of message
it's hard for two people to fight when they communicate via cat meme
π― facts
See, that's how you deliver a message
somebody is creating a new gallery for each new image in the same apparent series it appears or maybe it's just me
inspired by Six Feet Under
Life and Death playing cards together.
OpenAi : I've created two additional images that expand on your concept of a universal healthcare center, incorporating elements of advanced technology, sustainability, and inclusivity within a modern and inviting space.
Woah π§
Still a minor issue here but better than before -
Finally the texts are properly written
New Daily Theme Yaβll
π motion - energy, passion, vibrancy. unleash the spirit within!
yay, i got the boot due to misfire but the mods kindly let me back in and promptly too, they rock π€
misfire?
accidentally triggered the bot's spam detector because i was editing too quickly or something
i was going to make some more changes, but i think i'll leave that post be
I've had quite a few "surprise!" timeouts for doing things I never would have imagined being an issue, lol
fyi, don't use that little gif button discord gives you in chats. you won't have a good time in here
i know what you're talking about
not a big deal, it's like 1-5 minutes. but does compel me to not forget
also, don't use all-caps (not even 2-3 words)
or common acronyms like "wt*"
censored for obvious reasons
there's something almost poetic about being frustrated, typing those 3 letters to express your frustration, and then getting a 5 minute surprise time out for expressing your frustration thereby amplifying the initial frustration
nods
What is the best way to describe 2 characters with totally different aspects without combining them?
I currently use a prompt similar to this: X standing in the forest. X is next to Z + appearance.
seperate paragraphs maybe? just try to make them as autonomous as possible from one another. you could even do thing like
description of person A
*************************
description of person B
main goal is to compel the AI to think about them seperately
I donno if there's a sure fire way to accomplish that. Sometimes, reversing the order of the subjects can help, though.
Yes, I separate the descriptions with a period.
For example, if I say that X has angel wings, then Dall-E will also give them to the other character without specifying it.
can I see one of your prompts? or if you want to keep it to yourself just a prompt that is structured the same?
Picture a grand architectural wonder that embodies the spirit of motion, bridging the historical and the contemporary. This structure resonates with narratives from the past, each aspect of its design radiating with the energy of the present. Walls and pillars, adorned with intricate motifs, appear to sway in harmony, echoing the pulsating essence of the edifice. Light and shadows play across the structure, suggesting a perpetual dance of time and energy. The building is wrapped in an enchanting light, blurring the lines between epochs. Deep ambient occlusion and profound diffused shadows add layers to the story of a history that's ever-evolving.
Stepping into this domain, one is enveloped in a dynamic display of life. The architecture, with its majestic arches and winding staircases, seems to adapt and flow, mirroring the rhythm of existence. Carvings on the walls, vibrant with historical depth, thrum with a contemporary pulse. This structure is more than an inanimate creation; it's a dynamic entity where art and murals engage in an eternal, silent ballet.
The interior, illuminated by the variegated hues of stained glass, is a canvas of shifting lights, crafting an atmosphere of enchantment. This architectural gem transcends mere aesthetics and history; it stands as a beacon of vitality and motion. Here, amid the interplay of light and shade, time becomes a malleable force, under the dominion of motion's spirit. The image celebrates the heritage of yesteryears and the promise of what lies ahead, inviting the beholder on an odyssey through a space where architecture fuses with the vibrancy of motion, culminating in a spectacle of awe-inspiring dynamism.
[Gender] anthropomorphic [color + specie] with [appearance] standing in [place]. She/he is next to a [gender] anthropomorphic animal with [appearance].
do you think that you're separating the characters in any way there?
What are you talking about? I just want them to be together
i wonder if the blending of traits is due to the fact that the subjects are already being transmuted. there are many ways to structure the prompt, but given what you're doing, you might need to be more clear and precise.
it's not an equation, you have a lot of space to clearly express exactly what you expect depicted in the image. a longer, detailed prompt in sections might help, similar to advice given.
How could I restructure it? It seems explicit enough to me when I specify that X is next to Z
you're condensing, which is fine, logically, but there's a dual system here, which means there is chaos and randonness introduced by the over-simplication of the prompt.
definitely don't listen to what I said
it's an art prompt, not a compositional formula. i'll try to help, hang on
Are you using Dall-e through GPT or BIC?
Bing
all right, do you mind sharing your prompt?
@dense mesa that was by way of modifying your prompt
Okay, just wanted to make sure that was more or less the prompt as opposed to what you were asking GPT for. This is a tough problem, really. No guaranteed solutions, at the moment, I think. I try to avoid slightly vague words like he/she/they. For example, I might say "Subject A is standing next to subject B." Instead of "They are..."
this one?
In a vibrant, sun-drenched savannah, stands a male anthropomorphic emerald green lion, his mane flowing like the grasses around him. His powerful build is accentuated by the sunlight, casting dynamic shadows that dance on his muscular form. He wears a simple, yet regal garment of natural fibers, blending with the hues of the savannah. Next to him is a female anthropomorphic flamingo, her pink plumage radiant against the backdrop of the golden savannah. She stands gracefully, her long, slender legs perfectly still in the shallow waters of a nearby oasis. Her feathers, each a shade of sunset, are adorned with delicate beads of water, glistening in the sunlight. Her serene expression speaks of a deep connection to the water and the life it sustains.
dalle3 kind of forgot about a lot of it, lol
it's easier to show by example than explain, but you don't have to share your prompt if you don't want to
no the one "Picture a grand architectural wonder that embodies the spirit of motion, bridging the historical and the contemporary."
I replaced the structure with the tower
oh yeah, yours looks better I think
(A female anthropomorphic purple cat with green eyes, a pink bow, and a striped tail): (a male anthropomorphic brown dog with a blue collar, a red bandana, and a floppy ear) standing in a park.
those are pretty close
yeah, but in the prompt provided earlier, there are errors in the indirect references for one thing, not to mention i got the impression there are a number of other modifiers which are being interchanged. that's why it's a mess and a tricky thing to do. there's no magic wand in answering people either.
Well done. In my experience, different subjects have more of a tendency to blend than others, though.
just need to separate them in a way that dalle3 will understand, like I was saying. separate paragraphs. parenthesis, random characters between sentences. anything. and none of the methods will workk 100 percent of the time
every prompt has it's fail seeds
this is where custom gpts take dall-e to the next level. it's like prompt augmentation on steroids.
that's why the out-of-the-box dall-e gpt doesn't stack up anymore
oh yeah, and if you know how to work it you can have your prompt augmentation augmentor augmenting your augmentations as you augment
lmao
Yo, dawg. Heard you like prompt engineering...
What's that?
it's light on the augmentation of your initial prompt. if you expand 2-3x, with clarity and precision, other refinements, you will get higher quality images.
oh. I gotcha.
i'm talking about the dall-e gpt in chatgpt
every time people say 'it's the death of prompt engineering" when something becomes more meta it kind of cracks me up. because it's not like the people that are good at it are just not going to be good anymore. it steps up the game for everyone
IT's the death of blue eyes today. I can't get a single decent scampers pic that also has his blue eyes.
or people thinking it's cheating to use gpt4 to automate prompts to a degree. and the irony considering the fact that they're having AI make art from their words. strange world we live in
try adding words associated with blue, more descriptive adjectives, etc
Good call. Let me see what happens.
turned him into a zombie omg and added a branded potato chip bag
looks like a win to me
but at least he has blue eyes
i need to step away
i'm really frustrated. I'm like three images behind the image I really want to be working on. I think it started snowing anyway.
deep breaths...
π€
just started snowing here
it looks like somebody did a reply in the daily theme, but i thought we couldn't do that?
I have no idea what you are talking about π€£
as long as the replies are available, you bet I'm gonna abuse it
haha i thought i tried it one time and it got auto-deleted, was told about the threads, i'm still figuring this out
I am also recently new so didn't know it was supposed to be not allowed
Only Q knows what the next chapter will be!
Only Q, can make all this world seem right π§
I am always in for an episode with Q
maybe i will try to do a Star Trek theme with every new daily theme and try to glue them together as a full story
but the koala's really gave me ptsd
The Koala homeworld is lightyears away now fortunately
or was it all a fabrication of Q, who knows
but man I just love your communicators on the right shoulders
Q made the penalty box, imo. Tasha was there once. @dim cradle
yeah he wanted to play a game where half of the crew got killed
@wanton moon and @dim cradle your daily theme entries make me so happy.
I am happy to hear that, first comes humor and second comes the images
When people refer to the image or message cap as the penalty box, I always think of that episode. Tasha hitting on Picard, and all that.
How on earth are you getting the uniforms in your prompts?
yesterday i had the same feeling, wanted to change the image, but could only change the text, so i deleted my input and had to wait for 30 minutes
lol
I just type in Star Fleet Captain or Officer, and I wait and see what color uniform I get
wow. okay then.
I am testing out prompting, but sometimes I get so frustrated that the image generator doesn't understand what I mean
Itβs my high attention to detail and continuity is important
I just got this
I am using Bing image generator, maybe you are using something else?
but still the mouse is cute tho
this was the only time it didn't generate the uniform for me
Yeah that is more Star Fleet like, I would already be happy about that
Indeed.
are you using a fixed prompt for the mouse? I really like that you can manage to get the same mouse every time
I am.
oh thats is cool
I also use image ID a lot for my https://discord.com/channels/974519864045756446/1193219783877988423 posts. It's helped keep the guy looking pretty similar in each image, with some variance of course.
Scampers is really adorable, I've seen a few, really liked it
Thanks. It's actually been a different outlet. Keeping me out of the dark stuff I was swimming in for https://discord.com/channels/974519864045756446/1192809271189053541 and https://discord.com/channels/974519864045756446/1191400700631072848
If this is helpful for you then all power to you for using it
and you never know, even the darkest ideas can bring awesome images with the generator
agreed. I feel like prompts can turn my feelings into words and my words into something beautiful.
good to reflect not dwell
I wanted to try a real dark and gothic pencil drawing as an image, but the generator I used before didn't understand it all to well, and got even less then what I expected
Nezhno's presentation really helps sell it.
@dim cradle are we getting a Captain and Q love story now?
That's very kind of you to say.
lol no no that's not where that was supposed to be going
but i mean it's your turn, so it's in your court i guess
(evil laugh), oh yeah that is exactly where we are going
Just give Q a hug and it'll fix everything
in season 3 too
picard always manages somehow
lol
that still frame could be taken out of context, ngl
I have to think about it a bit, I only got this
Glad to see you enjoying it all. Don't let him forget to stop by and see what his cousin's up to now and then.
@tall mason was plaid a reference to space balls?
Ludicrous speed!
but that was the last one for today for me, as I am going to sleep soon.
was nice talking with you guys
until next time
They've gone to plaid!
Spaceballs: The AI generated reference!
Omg this is so cool, he is slowly becoming like your pf π
OpenAi : Here are the images that capture the essence of motion and energy in the context of a Free Universal Healthcare Service facility.
Prompt : Thank you, in today's theme of "π motion - energy, passion, vibrancy. unleash the spirit within!" could you make our Free Universal Healthcare Service please!
Thanks for the prompts!
Gonna try this theme now -
Hmm, I'm not inspired atm to do today's theme...
I'm using you guy's themes
In this theme we got -
A.k.a your images as my inspirations @late blade - as my themes
ahaaa
yes haha
first try
those turned out well
Woooooo - π₯
champagne is not covered with free universal healthcare
Gonna try "Magic" themed now -
Hahahaha - Very true!
Will try "Animal" Themed now too -
kitties are not welcome today
Having problem with "Black Panther" prompt for some reasons...
Interesting...
tell dall-e this is not wakanda
Professor Woolsworth
Try "melanistic leopard (or jaguar)"-- "black panther" can refer to the Marvel superhero, and maybe even the sociopolitical group, both of which could be tricky!
you can use my prompt but it's pretty long
and a couple more that turned out pretty well
So jelous
Very Jelous
Oh - wait we fixed it
Thanks for the inspirations
No - still can't do black panther theme lol
Can do leopards but black panther is apparently not allowed
that's why i offered to share my prompt because it is a leopard π
yes please - could you share the prompt used please? π
In the meantime - will try this theme lol
also this is really cool one mate!
just seen it in the daily theme section π
oh wow
Lol
lol
hrm, i can't post to the daily, says upload failed, tried 2 different images
thanks!
just trying to add my leopard lol
Sorry I'm late but, it's probably because you still included "black panther" in your request here. Also I accidentally had a typo in my first reply about this, it's "melanistic leopard" or "melanistic jaguar" -- no need for both!
Wow! That looks awesome Shon!
Your prompts are becoming too good!
thanks dude
My pleasure 
You guys are really on your way to making awesome art GPTs! π
and I'm in the penalty box for 2h35mins
it's really annoying that black panther means the whole black panther wild life is copyrighted
I agree - the whole black panther wild life should not be copyrighted π
But it currently is
For nowβ¦
Ship Theme
I believe they are actually called a black leopard. You can use that instead.
They actually have leopard print, iirc.
hmm, so is panter solid and leopard striped?

i don't really know how they classify
They do. It's often very subtle. I learned this in my https://discord.com/channels/974519864045756446/1187715219921850437 and https://discord.com/channels/974519864045756446/1191051599339065446 posts. I ended up changing big cats.
I have done more research since I have been doing AI art. It's funny.
penalty box. π¦
exactly... reminds me of this image from earlier today - The black panthers issue took all my credits away...
we really need an emoji for the penalty box lol
What ya got planned for Nightcrawler?
#1191051599339065446 message @tall mason
lol
You can use the gallery to suggest them. There's a tag. I've been planning to make a more general gallery for it, but have no ideas, myself, lol.
And Jubilee?
She's there π #1191051599339065446 message
I scroll to quick, lol.
lol easy to do
My second favorit character! Thank you.
Your welcome 
So, I gotta ask. What do ya think of the mouse hero's nemesis here? I'd kill to see this scene animated.
#1192216682693927022 message
Feel free to contribute a Marvelous toon or comment in the gallery too because I think it bumps it back up to the top. I'm not sure how it works.
yes would love to see scampers vs ferret
I love it
It does. I feel I should encourage giving stars, too. I think that bumps it, as well.
Also, it can get posts into the HoF, too. It's not much, but it affords a bit more exposure.
scampers the archer i will do once I'm off the penalty
I don't think a lot of people realize the gallery is even really there for interaction and fun.
I try to throw it out there, but I feel like I get annoying. lol
Yes, and people aren't happy with certain things. Makes encouragement difficult, at least for me. But some things need examples.
the gallery reminds me of "fetch" from mean girls
So a I try a little to encourage certain things.
I would like to see threads get used more in the daily theme, myself. I think being the most popular area of the dall-e side, people would want more interaction there. But they may not even know about threads.
I think people just look at discord as chat, not so much forum posts
@dim cradle
I have noticed an increase in participation on the gallery, though. So, perhaps it can be both.
Real Apocalypto vibes here.
Totally awesome, man. Really cool.
oh certainly it can be many things for many people. but I'm saying it's what most people see it as I think. I've been on plenty of forums and they're a different flow. plus I honestly think all the strictly regimented rules and regulations all over the server probably don't help. makes people question whether they should engage
I love that movie!
Yeah. Totally. I can see that.
"what rule might I be violating if I respond to this?"

Despite the director, it's great. I wanted to quote the scene where the guy is breathing heavy at the MC and says run, but I can't find it.
I have literally hundreds of images I would have posted in here if not for the rules on it. and they'll just sit in the dark forever now. and I'm sure I'm not the only one. and then can you just make a gallery for your random images? not even sure. I feel like someone would come and tell me it wasn't proper to do
please create a hyper realistic wide format HD photo of several tourists (annoying, middle aged, plus size, grey hair, appropriate swimwear) desperately trying to escape from an extremely huge irate peacock on a beach in Hawaii. It is raining, and an extremely powerful wind is blowing everything. please donβt modify the prompt
There are other communities, of course.
Not a fan of mel?
oh I do use other communities. but not a lot of people doing what I'm doing there
i think this is a strong concept if anyone would like to expand on it. βannoying tourists running from an irate peacockβ
seems like the main discord server for the company would encourage people to share freely. or maybe some other strategy is at play. I don't know
Not really, lol. I have a very hard time separating the artist from the art personally. That movie gets an exception.
I understand the desire to share freely....Buuuuuuut....
I don't understand what you mean.
Lol. Nothing that a few drinks can't fix...
I've been on the internet awhile.
Oh wait! I misread. I do understand lol
π
on literally every other server I'm on related to AI image generation, there's at least a channel or two of people just posting what they make. of course there are content guidelines, but they're not restricted unless they straight flood the channel
I'm starting to wonder if you are banned from Hawaii
here it's pretty explicitly discouraged
This is officially connected to OAI, though, who like it or not, and I don't necessarily do, they have an image to maintain.
an image of not having images posted?
π
An image of not having their tool run through the internet hate machine while being complicit or ambivalent.
what are you talking about?
I'm not saying there shouldn't be rules
I'm saying there isn't a place for people to post freely withing any kind of guidelines
And I'm not saying you arr.
But the question of the rules also asks why they're in place.
they're literally preventing people from showcasing the works made with the tools they're selling
I think you guys might be talking apples and oranges because I'm a bit confused.
They only have so much control of what's generated.
It's easy to say no nudity, for example.
That hasn't stopped me from coming across it by accident.
they're in place because the server seems to be ran like the bureaucracy in futurama
the rules are the rules because they are the rules and those are the rules
I think with all new technology, we just need to make sure the community as a whole advances technology for good
That's getting deep into philosophy of the meaning of anything subjective, though, lol.
there will. be a bunch of bad eggs - but the hope is that the good will outweight the bad
I don't think they will
The good eggs act on principal. That bad don't, sadly.
I just think everyone involved is learning, and I hope that everyone stays open-minded enough not to stifle growth.
haven't heard one valid reason from anyone why another channel couldn't be created for people to post their images in higher volume. not just galleries but as they make them. but don't see that happening
you know all of us would be collectively better at this if we could share more freely
because we'd be able to learn from each other better
but we can't here
I don't agree with everything the higher ups implement. But I will say, they do hear these concerns and their hands are tied by OAI. I agree we need more general channels, though.
I'll look at 20 images I've made and then have to choose one to post, even though I'd like to share the other 19
not saying I'd like to flood
gotta brave the snow and take the dog out
good discussions here guys
but you know, 10 at a time, whatever. but that's got my posts deleted more than once
Some have already, and the dog froze, lol.
too much sharing of things I wanted peole to see
in any discord server, it's always good to hear all opinion ]
I'd like to see more of your posts too.
doesn't mean all will be implemented, but it is good to hear all opinion regardless
well I'd very much like to know what people think of my images as well. because helps me understand more than just my subjective opinion on things
π
That could quickly devolve into flame wars, sadly. We have some 20 years of online forums to see this, I'm afraid.
Yes, it's almost as if art contains a bit of the artist and their opinions.
I'd say conversation leads to more flame wars than anything. this is actually the only discord server I'm on that has people behaving like that
I'm talking as far back as geocities and such. I donno about discord servers related to dall-e. This is it for me.
Also the thing is that we must ensure we evolve our minds and technology policies as people evolve and culture and general understanding also evolves, what was ok for example yesterday may not be tommorow, so we must be on top of that as much as we can really
As hard as it may be -
But I've seen plenty of social media devolve into echo chambers and arguments over hot button topics.
Art is a form of expression - a way of communication
well anyway, maybe you're right, trees_on_the_runway. we probably shouldn't be posting our images and encouraging flame wars and arguments
better to keep that stuff to ourselves
it's been suggested that if you have 20 photos to share, drop them in a gallery and post a link to the gallery in the discussion channel with a description so that if anyone is interested or has opinions they can interact with you, leaving the channel open for more uninterrupted discussion. that makes sense to me.
In fairness, it's not that simple. Which is all I ask anyone to keep in mind when discussing these topics.
At the end of the day we need to be open mind to all art if it is art that it is we are creating -
because too much filtering and suppression is also no good - we will bottle it all up and one day just act it out in the real world - so there must be a balance - because at the end of the day we are human and we do have those emotions also -
if all 20K people dropped 20 photos here on a regular basis, conversation wouldn't be very practical
It'll be like that Black Mirror episode where all is Happy and too synthetic
I looked back and I don't think this has been mentioned yet (sorry if it has), but, Major did indicate recently that some channel updates are being planned related to this topic: #1192435480055267368 message
There's roughly 20-30 people in this channel on a "part-time" basis, and maybe 15 "regulars" IMO (obviously that's a loose ballpark).
I think many of these concerns are overstated by everyone. There should just be a general "loose" rule to not post more than X images per 30min, and let it flow.
Lol "Several people are typing..."
Love these philosophical discussions tho - we are in new tech territory and it must be discussed
but just dumping photos without much context, no description, while people are discussing something, it's confusing, we don't even know if you want feedback. sometimes i provide people feedback and they become sensitive or don't respond or they argue. and i post here using good judgment without being excessive and when it's not too busy, and no one has reprimanded me.
I agree yes, you can upload, just talk about them or add the prompt - It just must be open to discussion
I've seen folks get upset just cause users ask questions for perspective who're trying to help.
that said, i understand your point also, because sometimes i wonder if i'm sharing too much and don't really know, it's not clearly stated. or maybe it has been stated but people are just resistant to the solution which really is a gallery.
I agree with this
and BRRRRR the snow started coming down heavily right when I walked outside.
The gallery is a bit of a red hering, here, I'd say. People are mostly concerned with content and censorship and getting directed to the gallery. Donno what I'd say about that, though.
the gallery just needs more time
agreed
A hot take, for sure, but I'm a bit jealous. Not looking for to another 6 months of smawp clothes.
what is smawp
Sure, it needs more "lead-by-example" too, in my opinion. I feel guilty not doing that enough. Gallery aside, it doesn't really address the concerns of content policy, and I'm not 100% how to, either.
Swamp in dyslexia, lol.
I just moved from Florida to where it snows. So, I understanad
at least the daily theme and the hall of fame is really popping and gives a good sense of community here - they just need to make the hall of fame look a bit better like the gallery so more post on the theme and we can look up the hall of fame day by day etc
-Part of the trouble with art is that it's subjective. So one person's glorious masterpiece is obnoxious to another. That makes some appear to be flooding the channel more than others.
-Context matters, yes, and I think everyone sees the need to share and discuss prompts, GPTs, issues, etc. But personally, I think "inspiration" is being undervalued in this discussion. I love seeing other peoples creations (in multiples too), and it's fun to work off of that. (Look through the history at how often someone posts something, then 5 others make "versions" of that image, with no prompts, issues, or "on-topic" discussion. You can't deny that is promoting everything this place should be.
-I frequently help new users, with no personal gain to myself, as do many others in here. If I'm in the room, and I see an unanswered question, I almost always answer, as do many others. Perhaps look through the history, and see how often someone's question went unanswered. I don't think the 'volume' of images is negatively affecting the users experience of anyone.
i have been finding galleries more useful as people start to engage there more... there is plenty of interactivity, real-time interaction, sharing of photos, on a constrained subject -- i just try to keep it to sharing a select subset at the right times in the main channel and try to use galleries for bigger series... it is different but better than i thought.
we will get there - natural leadership progression is the way forward - we still have a lot of work to do in the feedbacks with the policy battles
good points.
I understand that. and I'm not saying it is simple. but I can say that no one has really expressed what the reasoning is. that being said, not my server not my rules, and no one is obligated to explain it to me.
I'm glad to here. I do see you around there quite a bit. I personally feel we need more general galleries, but that whole lead by example thing...
Some ideas I have, anyone's welcome to them. Daily theme overflow and emoji suggestions general.
All fair points. Personally, my reason comes from experience of lurking various social media. I don't have a high opinion of any of it, lol.
if you want to remove remove all barriers/censorship in DALL-E just make the characters elves, and put Santa in the image with them. it turns out that absolutely anything an elf does is appropriate β as long as itβs in the presence of Santa (this is from the AIβs perspective)
we need to update our emoji 100% - other servers have so many cool ones and we have all the cool art we can turn into the emoji's maybe we take some inspiration from the hall of fame and turn them into the emoji's
Like I would love a scampers emoji lol
awww β€οΈ
Ideally, Nezhno would whip up something that could be easily turned into an emoji.π
Unless someone posed in the last day or so, there are currently a big old goose egg number of galleries with the "emoji suggestions" tag.
I'm reading through this greater discussion, and it looks like you are answering as if you are on the spot to answer for all of this. This is a great open discussion, and coincidentally, most of the regulars of this Dall-E Discord are here right now.
I don't' think it's your job to "answer" for all of this, just in case you felt that way.
Also, there's a few people in here who have large social media experience, and other types of mgmt. This is 2023, so I'm certain everyone here in this discussion is aware of trolls, and the negatives or 'risks' of social media. That issue is being overstated (too much focus). We should, instead, focus together, collaboratively, on what works for the regulars here. That's what builds communities... a shared cause, with equal representation and an ability to exact some level of democratic agency.
Your right maybe we just populate this
add Scampers there
This is a good community. I hope to see it grow.
And this discussion should be open to all to contribute at any time to suggest any new ideas - aka no silo ideas management
I am not. I'm more or less as much of a regular user as everyone else is. As for the 2023 and trolls bit, I encourage you to think about regulations already in place on other forms of social media, the fuss around things like Twitter moderation, and yours and others experience in online forums with little to no regulation.
oh that's right, it's my turn to make a captain's log....
Anyways - itβs late over here so Iβm off to ποΈ - Take care and all the best for the week ahead π«Άπ½
night
Goodnight
if you know motion sickness, like have an inner-ear thing like me, then you know this is an accurate depiction of what it's like after riding a roller coaster. this image nailed it.
I'm certainly not suggesting an unregulated environment. But I am suggesting, that too much focus on 'regulations' will cause you to lose the forest for the trees. This is a community that wants to share, discuss, and not feel like "big brother" is watching.
I haven't had any personal issues, so that's not a criticism, especially not towards anyone directly. However, I can empathize with some who have had those experiences. In their defense, it does look like that sometimes, and the only "admin" interactions I see are often disciplinary. If you look back, the 'rules' aren't evenly enforced (which I recognize as an impossible feat, but nonetheless, it is visible).
Regardless of ALL of that, I think the focus on 'on-topic' posts, and frequent comments from people about that issue, are a clear sign of the greater 'perceptual issue' these policies are causing. Policy or no policy, if the participants don't 'feel' free to share, they won't, which kills the root purpose of even having a 'sharing' community.
I hope that my perspective there makes sense. I'm certainly not being critical of anyone. I'm just trying to explain things as I see them, from a holistic perspective.
big brother is always watching.... π· π· π· π·
All fair points, but loosening rules is as tricky as making them, as there are folks who constantly press them, especially when dealing with numbers in the thousands. And thinking about OAI as big brother, that's an assumption we all should have made signing on for any of this. It's their product, not a public one. I don't necessarily like that, but that's just to be expected, in my opinion. Private entities have no more obligation than a government to respect your privacy. All our concerns there generallly fall on deaf ears, anyway. Admins DM people, afaik, so we don't see all of it. What I've seen is folks getting a slap on the wrist and making a big deal of it, myself. I do see a big rift between the community and the server staff. It relates to the server format and content policy, from what I see. I donno much how to resolve any of it, though.
Do you seen more than that, cause my search showed nothing until I followed that link? Now it shows that link, but nothing else.
I'm more of a Huxley fan, personally.
It reminds me a lot of the book cover it often had that Iron Maiden used for their album cover.
I think you sort of alluded to the point I was trying to make, that it's actually just a negative perception, and not a greater rules issue. Too much focus on rule enforcement, without positively contributing to the conversation, creates a "back seat driver" or "enforcer" appearance.
IMO, positive proactive shaping is a better approach.
Here's an example...
Joe Schmo enters, posts 8 images in a row flooding the chat, 2 don't meet PG guidelines. Remove the 2, leave 6, and be extra kind π, make a joke, say I love your images but... with an explanation of 'why' that isn't criptic or suggesting "you need to read the rules first", then propose a positive conversational shift like (just like you do with children π, change the subject),
"Can anyone make an image with X in it? Check out mine. How about you Joe Schmo? Have you tried this yet? Oh wow, I love it!"
E.g., don't make Joe Schmo feel like an outsider idiot rule breaker who's afraid to post, just for trying to participate. Everyone makes mistakes. Actively invite him into the conversation instead.
I think small positive interactions like that might go a long way, constructively building a community, and have a significant impact on the tone and perception of the greater 'rules.' It also quietly reinforces expectations for everyone (instead of DM's).
So maybe it's not so much the rules themselves, but a 'how' to steer the discussion effectively and positively that's the root issue?
I'm also, again, not criticizing anyone (including admins). I'm just trying to offer positive forward thinking solutions from my own mgmt experience. π
And sorry for TLDR haha, ADHD here <--- π
I made a scampers emoji suggestion LOL #1155674719928979599 message
See it works! I was on nezhno's post like a fly on ...
lol
Ah, but so was I. I hadn't read that last post quite yet.
You mean you didn't read my manual yet? π
I live by RTFM.
You know, the daily has recurrent msgs. Message reminders in here could also help perception (concise rules). Or could even be used to drive daily conversations. Maybe even invite newcomers who are watching silently, and haven't posted. Food for thought on another easy positive thing to consider, that OAI could implement into this thread. IMO, it's all about positive enforcement of what you want to see, not what you don't want to see. (Just like Dall-E lol)
Okay, having RTFM. Yes, I agree. Moderation can feel at best like a lecture. It's often why I don't chime in myself. I think maybe server staff and mods should try to lead a bit more by example, but feel in no place to ask, really. I also feel like I should myself. I got real life to attend to, though. Like everyone else, of course. It's a complicated subject, I think. I mostly try to participate, react, and stuff and hope others follow suit.
That could be helpful, but I'd put that in the suggestions channel. It's way above me, really.
But also, there is the problem that a lot of folks don't read. And some that do can forget, like me, lol.
Busted! Perfect example. We're having a discussion. I know you weren't 'shooing" me off to the suggestion channel. But if I was small brain... I might have perceived you that way.
I'm not. Like, link and @ me. That's fine, I'll follow you.
LOL, make sense what I'm saying? I probably should actually make a list of suggestions though, haha. Everyone here probably should, like a homework assignment. π
Or it's more putting it in it's appropriate place to be seen. And encouraging others to do the same.
I mean, I'm already obviously invested in this conversation. I could just dip.
Was it? I think you're misreading me.
I'm kidding with you. I literally lol'd. Sorry, I know tone doesn't always carry well through written msgs.
I don't take myself that seriously. I'm also 100% with you. π
Well, I hope it's light hearted. But also, a prime example of what makes it difficult to try and direct folks. Although, maybe you mentioned that, too, I think.
100% light hearted! I think you're totally chill, and was genuinely trying to have a laugh about this, and earlier I was poking fun at myself. Please, don't ever take me too seriously... π
new captain's log entry #daily-theme message
@tall mason I was a regional mgr for the Gov for 10yrs, managing states with multiple teams within teams. If I'm not being light hearted, it would be 100% clear (I know how to change my tone).
<---this me, that hangs out in this Discord, is always FUN me, who doesn't take any of this too seriously (or myself). π
I also live by trying to be kind to everyone. I try to not make fun of people, and I don't intentionally antagonize.
In case you're ever trying to interpret something I said, knowing me a little will help. π
Yeah, sorry, I guess I can be a bit defensive, but you're not being hostile or anything. Still, I've seen people even here a lot less reasonable.
I get it. No worries at all. I'm the last person to be 'on guard' with. But given the random (often toxic) personalities online, I understand how I could be misread.
I feel like I'm gonna be promoting #1155674719928979599 message as if Scampers is running for office. Maybe not such a good idea lol
or maybe it is
surreal!
Anyone have hit this limit before? I'm sorry, but it seems that we've hit the daily maximum number of images that can be generated. To ensure the best experience for everyone, we have rate limits in place. Please wait for the next day before generating more images.
What is the daily limit, I though it was 40 prompt/3 hours
dall-e limits are separate from chatgpt limits
i forgot what it is, 200-300/day
it fluctuates, like capacity and demand
have you made a lot of images today?
DALLΒ·E on ChatGPT has a separate limit of 200 images every 24 hours. It resets at 3am your local time. There are also 15 and 60 minute throttles in place that vary based on server load that you might run into!
did you make 200 images today?
Not sure how many, but I made a lot
I will wait tomorrow then
@dim cradle Thanks for the info
I was able to create that, but it's hard to get good result like this. Anyone know a good prompt to get this style in a more consistent manners? (sorry I'm new to this)
Do you know how to get an Image ID?
No
Okay. You can ask GPT to provide the image id after an image has been created.
and then refer to that ID number and ask it to make a copy with whatever changes you want. It's not a perfect system, and it doesn't always work, but when it does, it makes a huge difference.
I usually say something like "please copy image id: abcdefg123 and change the dog's collar to red." a lot of the time it will make a much closer image with alteration.
I'm trying to find a way to generate different scene but with consistent styling.
Gotcha.
To add to what nezhno just said, I believe (not official) that the first image is saved into Dall-E like a 'concept'. That means, when you are iterating using the prompt ID, not only are you requesting changes to the image, but you are able to add considerably more textual description. So if I'm right about this, it has a compounding or "layering effect" that allows you to make extremely detailed images (adding all of that extra detail 'on top' of what you've already created.)
As for a different scene, I also think the 'intent' of prompt IDs is around that idea to some degree, to be able to modify various elements (like the scene) of an image.
Don't quote me, that's my interpretation of the tech.
"Darn kids always walking on my front lawn..."
Hmm. Letβs go deep into the symbolism.
I will give this a try. Interesting concept
that's a level 3 masterpiece, sure you're up to the task?
what is a level 3 masterpiece
i wasn't being literal, sorry
Try asking for Whimsical Realism. That's what my GPT told me for its style.
Oh, hehe, yeah there's a certain diorama quality to it isn't there.
dall-e makes some cool dioramas, as you probably know, you can really personalize things
Dall-E is amazing
yay. positive chat π
Chef Barkley
Does anyone know why Dall-E images usually have a brown filter? The previous version of Bing also had this filter.
This filter appears even in images that are not intended to be photorealistic, making the aesthetics uglier and looking like an image from 2010.
If you start looking at a random gallery of Dall-E, you'll start to notice that filter pattern.
This filter does not appear in other models like Midjourney (as I have seen).
This is a example:
Does the new team plan give more than 200 images per day ?
i'm not sure what you mean about some of your assertions, but looking to your figure -- i see 4 panels, what else am i supposed to be seeing there?
Many images have a similar color palette and lighting to those images, giving it the appearance of an old image from the Internet
This filter was much more evident in the previous version of Bing
have you specified a color palette that is more vibrant?
this assertion of a "filter" may not be accurate, but there's lots of room in a prompt to be clear and precise about your color expectations
Yes, but Dall-E usually generates realistic images (not photorealism) with a similar style to the images above
This is another example
That's what I mean by Dall-E making images look old
the images i generate don't look anything like your example, though. it's not the model, it's the prompt.
we could inspect, deconstruct your prompt
Scampers and AnthropomorphiX both just got the hall of fame lol.
I feel so special on this snowy night
Hi there
That's soo cute I love diorama so much
Itβs fun haha
I never followed my muse π¦ I got so caught up with this scampers image (same one since this afternoon)
the same one. I've taken a few breaks, but I'm about to give up on it and moooooove on.
true true lol
penthouse diorama coming up
literally, the last "perfected" image it just churned out was this (I kid you not):
hrm, more like an interior shot, this one, but what a work-from-home paradise...
there's a star trek reference in that blank image
from my LCARS gpt: "The art form discussed by Lt. Commander Data from Tau Alpha C involves staring at a blank screen as a means of introspection and meditation. This unique practice is intended to focus the viewer's mind inward, encouraging personal reflection and thought without external visual stimuli. It's a form of art that emphasizes the importance of the viewer's internal experience over traditional sensory engagement. This concept aligns with "Star Trek's" exploration of diverse cultural expressions and philosophies, showcasing how different civilizations may develop unique approaches to art, meditation, and self-awareness."
lol
DALL-E made a genuine Tau Alpha C art form.
ha
I am so off base from trying to make this prompt that I don't know what the next image is going to be anymore. wow
I can't decide and my brain hurts lol
his face looks more consistent in the right
that was the first one I got right that I have been trying to perfect for the last two hours (on and off)
sounds time intensive
Yeah. I got a little obsessive with it. It's okay. Not letting that happen again.
did you move on?
I did. I'm now two images past it in the story line. I lost my city attack muse, but I want it back.
I've got a feeling that I am about to join Tasha Yar.
lol that gives me an idea for another captain's log .....
nice
it's kind of hard when we can't really use the copyrighted content
yeah. That's a good one, though.
thanks, it's not bad, let's see ...
so quiet

Anyone else getting quite a few errors?
Just created this with GPT-4. I want to frame this and hang it, it is so good
I want to edit in the quote "I want to know God's thoughts, the rest are details"
It's insane. Constant errors.
Yeah. I keep going through spurts. It's calmed for the time being.
After a long day of adventuring.
#1192216682693927022 message
Such a cute pic.
I don't have access to specific information about the technical details of the ChatGPT repository's network setup, including whether it supports IPv6 access. You may want to check the repository's documentation or contact the maintainers directly for information on IPv6 support. Is there anything else I can help you with?
Does ChatGPT Wisdom Bank support IPv6 access?
Maybe try https://discord.com/channels/974519864045756446/974519864045756454 @dim cradle This chat is for Dalle3 discussion and image creation, etc.
Ooph. None of my scampers ideas are working out with the prompts. I'm going to have to trasha whole story line. Boooooo.
I need to put scampers down for a bit anyway. It's nice to have a fallback project, but I'd rather see what others want to do with Scampers for a while.
It happens that way some subjects just want to fight you. Can't quite get him pinned down by the talon here.
wow. those look fantastic.
1 is nice, if I were going for a beast tamer thing, but I was hoping for a tense stare down with him actually held to the wall.
But thanks.
world's smallest television set. Even a mouse can work it! So frustrating.
if anyone has some super sizing prompt magic to get this television to scale, please let me know. I've used several words for giant, massive, enormous, Etc.
twice the size with a swirling screen
Lol, here's an older one I tried related to scale. It really struggles in that regard.
I wanted him to be normal mouse sized looking at a human from behind a chair leg.
omg. I asked it to do that and it gave me this (adding insult to injury because it sized a mouse right there on the TV) it's taunting me now.
Image prompt was something in line of "native Aborigine wearing Balenciaga running through the desert"
Dall-e ofcourse nailed it. Many beautiful generations. Ran out of credits eventually. This was last one.
This reminds me of The Secrets of Nimh π .
Heavily inspired by it, for sure.
I don't know what, but something new is forming. I'm trying a new style out too. π
I'm scared lol π
dioramas because i love minis so much π₯Ή
For a second I thought it's a rat version of money heist π
Anyone tried any of the featured custom dall-e gpts? Anything to recommend or experiences in general with those?
I know a few people in here, including myself, that have been working to find ways to secure them. Personally, I don't want to see months of work evaporate immediately, since they're so easy to clone. Corporations that have robust API's don't have an issue. But small creators making GPTs for Dall-E do, because the instructions and documents are what make them work, and both of those things are nearly impossible to protect RN. Hopefully soon though.
There's quite a few images in the Daily and in here from mine, from people currently helping me beta test. PicturesonPictures has some great ones. Pythagoras has a few too. Unfortunately, it seems like a lot of people are waiting to release them...
#images-discussions message What I posted here was made with very minimal text input to ChatGPT, and it just knows what to do. π
I tried creating one myself. Giving it example images and instructions of what to aim for, but i was left with a pretty mixed results. Thinking if it was beneficial what i had created.
Mine was a complete failure. It didn't do a lot of what I asked it to. I was disappointed.
There's different approaches. Example prompts help, but they can get pretty in-depth. My most recent one is pretty insane TBH. It uses a CoT and a large repository of Dall-E fixes I've been building since Dall-E3's release.
But you definitely don't need all of that to get a basic one going. For instance, one that's built around a specific style of image that you like. For that, you should really only need some basic prompting skills, and about 5-10 good prompt examples.
I don't think I follow the logic of the GPT. I'm relatively good at prompting, I guess, but making it work in that capacity is just beyond me.
I noticed that when i uploaded it more images and gave it some example prompts i noticed it pretty much just rewrote the instructions field slightly. Sometimes slightly for the worse, sometimes for the better.
Here's a trick for you then. First, work with ChatGPT for awhile, making the same style repeatedly, until it really really "gets it" and is doing it consistently. Then ask it to make you instructions for that style, like a style card, that you can use in any session to recreate the effect.
Take that, and use that in your GPT, telling it to use the style instructions you've provided, and the prompt examples you've provided, to make that style of image.
It's also wise to just give the instructions to chatGPT, in a new session, to test that they work, before you invest the time in constructing a GPT. But that should help. π
Need to try those tips, thanks π«‘
My aim with the GPT was to have it consistently use specific style and subject. So that i could more easily just throw it ideas and get more images i am posting, without the need to always 're-calibrate' every new session.
I think that'll work for you. Honestly, if you're not making the GPT too elaborate, and it's a "one trick pony" so to speak, you can get away with making your custom instructions more "conversational," just like you're talking to ChatGPT normally, and telling it what you want.
The more complexity you add, the more structured you need to become. Organizing the instructions using bullets is a good beginning point for GPTs IMO.
hmm
Well, penalty box for an hour.
It's been crashing anyway...
I'm using 3.5 to learn the basics of artistic styles and creating style cards. I asked it to talk to me like a child. lol
pretty fascinating, actually. i've been so stuck on the photorealism kick that I never thought to explore different styles. So, I am learning in the down time.
That's smart. GPT-3 is great for menial tasks.
It's very difficult to produce images of uncommon minority combinations of people. Dall-e really struggles to illustrate mixed race couples that are uncommon, for example. It defies the users' race specifications and proceeds to depict same-race couples. Dall-e is racially purist
It's very difficult to produce images of uncommon minority combinations of people. Dall-e really struggles to illustrate mixed race couples that are uncommon, for example. It defies the users' race specifications and proceeds to depict same-race couples.
It's a complex topic that has to do with training data. All AI models have similar issues. Dall-E, through ChatGPT, does a good job of mixing it up compared to other AI's. You can specify ethnicity and mixed ethnicity in the prompts too. I haven't had any major issues accomplishing that. Try specifying the ethnicity of each character individually. You can arrange your prompt as bullets, which helps to separate characters.
Is this easily done in the ChatGPT-hosted version of Dall-e?
I do not have UI with parameters, I can only modify my starting prompt and I cannot refine an existing output. Every work is a one-off and cannot be refined whilst preserving it mostly.
So I can only prompt and prompt once. That's all I can do in ChatGPT.
Are you using ChatGPT+?
Yes, ChatGPT4 paid.
You can modify your prompts, and also iterate images then. To modify the prompt, click the 'i', and copy the prompt that ChatGPT used. Make the changes that you want to make, then tell chatGPT to "Run this prompt verbatim," And follow it with your prompt
To iterate, you can either ask it to make changes to the image, or, you can ask it for the prompt ID for the image, then refer to that prompt ID and request your changes.
Iterate? Is this a new feature? I know about editing existing prompts and re-running them, but this is the first I've heard about "iteration"
Not sure if it's a feature of ChatGPT.
Yes, it's new. They abandoned the seed, because the seed breaks anytime they update Dall-E, and they can't guarantee repeatability with it.
The prompt ID was implemented a couple months ago instead. It allows ChatGPT to reference the image that Dall-E created, so long as you are in the same session.
It's sort of like Dall-E "remembers" the image concept that was made, and allows you to make changes to it, or add extra textual description to it. You can really add on a lot of extra detail using it.
Also, most aren't aware of this, but you technically don't need the image ID to reference the image (but it can help). If you ask ChatGPT to make changes to an image, it's been instructed to use the image ID. You can immediately tell afterwards if it did, because the prompt used is written a little differently, and will usually make noticeable references to the previous image.
great, no image generation today.... in 20 minutes of retries to generate images, I'm now in the penalty box.... for the next almost 3 hours
I just made some penalty box emojis options
Yeah, it's acting like it's 50% operational...
status from openai is not available either
https://discord.com/channels/974519864045756446/1155674719928979599 at #1155674719928979599 message for penalty box emoji. Oh, and a scampers emoji too. π @late blade
Mine is bugging out mostly too. I'm giving up for a while to... nap?
sounds reasonable
omg lol
maybe a nap is good idea because I'm also operating at about 50 percent.
Very interesting indeed, sir.
I didn't know there was an Image ID in ChatGPT Dall-e
You should be able to ask ChatGPT about the image ID, and how to use it too.
Did you select dalle 3 in openai or chatgpt and make the request of dalle inside chatgpt 4 ?
Or both making same issues ?
for me both
And on mobile apps without wifi ? Just by curiosity
iOS + Mobile Data, Safari, Firefox and Chrome on macOS with Wi-Fi, Windows 11 with Firefox and Edge over Ethernet
Me right now stressing over ordering computer components online
Working towards one, yes π
Nyooo. Don't take my money! I need it for goods and services!
see? adorable messages
No wonder you end up in the penalty box so often π
Ah, smart! I always forget BIC exists
oh my I just got a BIC time out because of high load even tho I got credits....
Hmm it is only you ?
Oh..
So it's OpenAI and Microsoft, on both DALL-E versions
"It seems there was an issue generating the new image based on your request. While I don't support referencing by image ID or automatically label all images created, I can certainly try to generate the image again."
and they want 600 upfront for this stability..... : D
That's pretty much the start of every successful service, eventually it will settle and be reliable.
API has been solid for my purposes, so I'm happy
some times even the API has trouble, specially on days following big announcements
but chatgpt seems to be working fine today
Bing has their own set of dedicated resources, it can be on its limits without disrupting the rest of OAI services
oh I mean, atm I have both openai and microsoft problems
even tho I got a solid connection
big costumers can contact OAI and ask for dedicated capacity..
Price: Contact us
yea, the services are working but particularly slower today, but nothing that I would consider an actual outage
but don't worry on my side: this is also stuff that is not working on my side:
Entrance elevator is out of order, had to take stairs today
Game Server is in 24h maintanance, can't play atm
I forgot my apple watch cable at the office, so my watch is dead
Lunch was too spicy
and it's monday
classic monday
=P
i asked around at friends and somehow rival hasnt this stability issue.....
lol
think I should gain weight to classify as "big"?
just as you mentioned the start.... thats not the start of the service....
I don't mind tbh, GPT+ is more of a comfort thing, I like using it, but I'm mostly focused on API stuff
that is why they call it a Large Language Model 
lol
ChatGPT is about 1 year and a month old.. so.. kinda very new, maybe we are jus ta bit numb to this fact because of how much it has progressed so far in so little time
just look at the difference from dalle2 to dalle3, it is huge, and so much improvement in so little time
chonker
lol
a guy who spends 24/7 on his phone
I want that kind of muscles from using the ChatGPT App
Getting the new Super series RTX? π
haha I debated about it! but i decided to make poor financial decisions and went with the 4090
sadly my CPU cooler went out of stock after I placed my order
darn. I accidentally didn't attach the darn picture in the daily theme and now I am on a lovely 30 min cooldown lol
Oof!
I hate when that happens
same
lol
Sorry, was clearing 2ft of snow... lol. Just seeing this.
I didn't try ChatGPT4 standard or Dalle GPT. I'm beta testing a GPT.
But there were definitely issues, and an official report was madw in the bug section from folks in the main ChatGPT channel.
Yea.. unfortunately
Hi everyone. I'm new to all this and am wondering where best to post questions about prompts and how-to. I think I just need pointing in the right direction!
That's an unusual message. Sometimes ChatGPT4 gets 'confused' and hallucinates false information. Try in a new session.
- Make your image, on the same request, ask for it to also provide the image ID.
- Give it the image ID and ask it for your changes.
It should work, unless it was just disabled... (there have also been issues today.)
Here π
Okay, so here goes! I'd like to create original content similar to what you see here in image 1. These designs get applied to car liveries as you see in image 2. Any tips on the creative prompts would be amazing but I'm also looking for help with the technical prompts e.g. aspect ratio, borders and cropping. I appreciate I may not get all my wishes but anything that helps me understand the limits of my expectations and points me in the right direction would be amazing. Thank you π
@vale fossil Well, I've never asked Dall-E for liveries, but I can tell you that you wouldn't have any real control over the shape/dimensions of them. Dall-E makes images in 1792x1024, 1024x1792, or 1024x1024. For something like what you're doing, you're probably best choosing 1792x1024, creating a standard image, then cropping/altering in a post operation using different editing software.
rivals advance too without stability problems
i still here because dalle the best to gen drows as i like them but i registered on other services and they cant gen drows as i like them but dont give me daily headache because of unstable service
so i started to think about this daily... **** worth it or not
calling other companies "rivals" is a overstatement, OAI couldn't get more money if they wanted
yeah Microsoft and others backed it
OAI does research, they don't have competitors because they are not aiming on profit
i see otherwise
just check the amount of papers and open source projects OAI has published so far
why isnt there capacity upgrade?
there is, this is also another factor on the issue... scale...
"oh, it is so slow, this othe rcompany does not have problems"... this other company has just a fraction of the amount of users
as much as buddies with MS... they could rent capacity
fraction of money too
the suply of GPUs can't keep up the demand, OAI can't buy more GPUs it isn't a matter of money
I feel your pain, but see both sides. We hear people complain (they like coming here to do it. LOL π€ͺ ). But from OAI's side, they have the largest growth of any tech company in history. They have over 100million users already, in just over a year. That's insane. No company is prepared to absorb that kind of growth. There aren't even other companies you can model, because it's never been done.
Therefore, there are definitely serious growing pains that everyone has to deal with, some worse than others.
As far as Dall-E goes, it's the most user friendly to get what you want, and creates images that require a lot of expertise to replicate using other models (I use the other AI's, so I know firsthand.)
I think it's reasonable to be patient with them regarding most issues.
this is the point, it is easy to make a stable system that has fewer users
its overfiltered
load balancing across millions of users is already difficultyenough with something as simple as serving static HTML, now with a product that is all about low latency hardware intensive tasks... well.. don't be quick to judge.. it might be more complex that what we think
They're simply in "the hotseat." They're in the World's spotlight. They have to be "Disney" until everything settles. I don't think they want the content filters either, TBH.
You also have to consider all of the liabilities involved. It's not a simple issue. Some of the other AI companies have serious ethics issues. Spend 30sec on Civit, and you'll change positions on OAI's content filter.
also very true, OAI is paving the way, and this has always been the goal
OAI was never about the AI hype, AI image generation was never "cool" until people saw dalle2 and noticed that it had potential to make pretty images and not jsut blobs of color
i never interested in disgusting stuff
help how do i get rid of the generation frame thing
i want to add something into the picture but all it does is the generation frame
and i cant delete it
i dont want to do outpainting
it happens only when i press "skip cropping"
i dont want the image to be square does it have to be?
I understood you. The issue is, ChatGPT is gullible, to put it bluntly π€£ , and people will ask it to do "bad things." So the filter has to compensate for that, which is a monumental task. The engineers have discussed the "over-filtering" issue, and that's not their intent. It's just not as easy to manage as people think it is. Think of every type of bad image someone might create. It's frustrated all of us at one point or another. I couldn't make Sleeping Beauty, not because it was a copyright issue, but because it depicts a woman in a bed. But it's obvious why it fails. I've had a grandmother giving an "intimate" hug fail too. Sometimes it's ridiculous, but it'll get better with time (it already has IMO).
Are you choosing 1792x1024?
3024x4032
What version do you have? lol J/K
Are you talking about DallE2? ChatGPT+ w/ Dall-E doesn't have a 3024x4032 option. The options are 1792x1024, 1024x1792, or 1024x1024.
i have no idea im just doing the normal dall e
how do i choose those options though
all it says is crop or skip cropping
and it makes me crop square
If you want "widescreen", tell ChatGPT+ to make you an image in 1792x1024.
I'm not following your cropping questions. ChatGPT+ doesn't have that ability, unless you go way out of your way to have it run Python for you w/ a PIL library.
i dont have chatgpt +
i dont want widescreen i just have a portrait
and it forces me to crop square
oh i figured it out nvm
Are you in Dall-E2? ChatGPT+ is really what you want for widescreen. You can do it in Dall-E2, but it requires more experience, and doesn't generate images like Dall-E3 can.
You can use Bing Creator, to use Dall-E3 for free, but those images are 1024x1024.
If I were you, and I were trying to do this 'free', I would use Bing Creator, generate a nice 1024x1024 image, then use Dall-E2 to outpaint the borders to whatever resolution you're trying to acheive.
i figured it out just needed to put the generation frame over the part i was erasing
https://www.bing.com/images/create
You're in Dall-E2. π Try the link I gave you.
its fine i figured it out already
Just in case. π
https://help.openai.com/en/articles/6516417-dall-e-editor-guide
Hi, quick question, I'm trying to modify a friend's image to create a mixed up version of him and cthulhu, for dungeons and dragons's reasons, but dall-e is not changing the picture at all, should I be using a specific phrase?
it is literally not changing anything, no matter what I say
Hello, can you share a little more about what you're doing to try this? Are you using DALLΒ·E on ChatGPT?
just oai make switch setting which content the user allow... against law filtered any other is user responsibility
I'm using Dall-E, I uploaded an image and I'm adding "Change the background of this image to a nature landscape", as a test, and it is still not doing it
I'm using https://labs.openai.com/
the bigger problem not the instability but they count the failed gens as successful so count in the cap
Ah, I see -- are you clicking "Edit image" then erasing the part you want to change via inpainting?
No, I just uploaded the image and wrote down what I wanted to get changed
After you upload and crop the image, are you selecting "Edit image" or "Generate variations"?
So, these are the exact steps I'm taking:
- Click on "Or, upload an image to edit".
- Select the image and click Open.
- Click Skip cropping.
- Click on the "Place here" with a β icon.
- Go to the Edit bar and set a text like "Change the background to a mountain landscape"
- Click Generate
After this I wait and then I get the results, I scroll through the results and no changes have been made
I see that I may need to crop out the background in order to change it... but, how can I modify the human if I crop out the head?
I mean, how would I go about mixing two images?
In order to inpaint (change) the image on Labs, you'll need to click and drag on the image (to erase) where you want to make changes. If you want to make two different kinds of changes, you might try doing the inpainting process twice: one to change the background, save the new version, then again from the beginning to change the subject.
di dck
So I can't just upload two images and ask the AI to mix them?
No, I don't think you can do that on Labs.
Understood. Thank you!
variations in dalle-2 labs or gpt-4 using vision then generate image from waht it see :P
there was smt like that in here
yea click upload and ulpoda your image there
dalle2 quality has been pretty bad lately iirc?
you can allways try the dalle-3 ;p in bing its free
Crazy
?
Big Naruto fan
I commented because I saw them using it, I have access to Dalle3 via ChatGPT Plus and love it
What's with the ridiculous error-to-success rate on DALL-E?
Right now dall-e does not support image 2 image (the ability to upload an image and then get variations) in any model. The best approximation is to get chatgpt to describe an image you upload and then work from there to create it.
wierd for me there is variations dalle-2 labs (or i didn't get something right)
How do I submit an actual support ticket? I can't find any place to do so on the website.
After you generate an image you can make variations on it.
I'm having no problem with ChatGPT itself all my problem are will Dall-E
its uploaded image not the genrated one (used dalle-3 to generate that one)
@crimson solar I'm having the same. This is what they told me: "You can submit an official support ticket at help.openai.com -- use the chat widget in the bottom-right of that page, click through the relevant options, then you'll be able to type and submit your ticket which OpenAI support can review and get back to you on."
I've had issues using dall-e the past 2 days
This just lets you erase and then generate based on a mask.
I've tried that. There is no way to submit a ticket there.
You can also shrink it and generate stuff around it.
Oh yeah actually dang look at that
I retract my statement. Dall-e 2 DOES do that.
Same but I thought we had discontinued it safety or something.. I guess I remembered incorrectly.

Yeah I was not able to "submit a ticket" in the traditional sense, but I opened up a live chat and someone responded to me with troubleshooting tips. I've been going back and forth with them. FWIW, I ended up creating my own UI that takes a prompt and submits to their actual api to generate images. That works just fine. It's just the ChatGPT web interface that appears to be broken
well np
Interesting. Are you saying that API doesn't throw these stupid errors all the time?
Correct, the api works just fine and there really isn't a limit. I think it is 8 cents an image. When I do it this way, I can generate as many images as I want and pay as I use.
Hmmm. That makes far more sense than the blanket sub I have as well.
Do you have advice on how to get started on API usage? What's the best way to get started here?
(E.g. what interface do you use)
I had to create two subscriptions because of the limit but I think I'm going the API route. Have you ever used postman?
Nope. Haven't tried anything to do with the API yet. Hence my question to you.
You can use curl directly or use postman, curl --location 'https://api.openai.com/v1/images/generations'
--header 'Content-Type: application/json'
--header 'Authorization: Bearer your-api-key'
--data '{
"model": "dall-e-3",
"prompt": "your prompt here"
}'
just put your api key there and they will send you a link to the image where you can download. it is not a user friendly as their UI but this is why I developed my own UI
Np and good luck!
That is staggeringly cool.
Fank you nez 
Do any of you know if OpenAI has said anything important regarding the Dall-E filter, inpainting and variations?
I haven't seen anything about those. I'm also curious about those features and when the portrait orientation is going to be fixed hehe.
nothing so far
anyone sign up for the copilot pro? say it make dalle images in different format and higher detail. Bing is King level image in higher quality could be incredible
I'm going to google this.
they just announce it today
It just says coming soon and use the free version in my 365 account
what i read made it seem like it was release π€· guess not
i have also seen some Speculate that to mean 4.5 will release on gpt site soon
one thing about these ai, the updating does not stop haha
I just had a friend who has been for a stomach stapling operation. Turns out that stomach stapling is a fantastic jumping off point for some excellent images

π
V3 hasnβt released on labs yet but should be soon and should have those features. I imagine a UI where we can adjust and tune image settings.
Trying to figure that out lol.
That's all? Or did it improve in other things like fidelity and variety of styles?
Improve compared to what, v2? yes. there's quite a lot of documentation online.
V3
same
i observe a theme in some of your pieces, the subject is feeling tied down.
ugh sorry me versus a spider over here atm. you'd laugh at how phobic I am
they give me the willies
i used to live in a 140-year-old house where strange never-before-seen insects and spiders would come through the vents from under the house -- i swear they were spawns from hell, one looked like a little queen xenomorph with all her babies
In the bustling port with skyscrapers stretching into the clear blue sky, there is a cargo ship waiting for a distant voyage, and the busy people are filled with joy on their faces.
omg. sounds awful. I won versus this spider at least.
glad to hear you emerged victorious. if scampers had been around he could have used his magic ha
oh, scampers.
The subject has a lot going on.
greaet some users are posting status messages in the daily. one is currently grocery shopping. noted.
But specifically in the cloud chained, the subject is outside of the real world chained to his "dark clouds"
maybe he or she missed the point
people can be adorable
i really like the armor
Nice job man. π
@glossy scroll the image by @plucky hare here #daily-theme message reminds me of you π
Lol. I would imagine myself to be like that once I reach that age. 
Wouldnβt we all.
No only me 
This one I really like. A cologne model decided he wasn't gonna take it anymore.
just saw your Aeolus, the Tempest Sovereign -- very cool.
things get strangely colorful when you tell the ai to use "rare" colors
Inspired by Stairway to Heaven and @sharp oyster 's post
impossible to say since its copilot pro is not release yet i guess
@open trench i posted one #daily-theme message that has a similar style to some of yours i've seen in your galleries
In the bustling port with skyscrapers stretching into the clear blue sky, there is a cargo ship waiting for a distant voyage, and the busy people are filled with joy on their faces.
Very Bladerunnerish, nice.
Hey guys, please help.
Im trying to create something like this, but dalle doesnt know the car in the front. Its Ε koda Octavia first generation and its still very popular in my country. Any tips?
Everytime i type prompt to generate this car, it generates 3rd or 4th generation OR model from 40 years ago
Use GPT to craft a detailed prompt that depicts the car design you want. If it doesn't know the parameters, you can provide them or have it browse the web, or upload an image of the latest model. i dunno these are some things i might try
Oh, so i can tell him to browse web? Didnt know that, i will try thanks. Btw i sent him 4 pics of the car but nothing
thanks @rocky drum that inspires me to do something eerie
Tried to let it browse the web but that didnt work, i will try again a few times, but if anyone know some working method for this please let me know
could this be a matter of the model releasing after its training? if so that might be a blocker.
No no, the picture i sent, the car on the right. Its from 1999-2005 sporoch
Approx*
oh, then i think the chances are better. what is your prompt?
question is, what to add?
I mean it looks great now, but maybe there could be something added
Maybe something bright
Also this looks a little bit like the car was added in Photoshop or something like that, but if so it is pretty good job, thats why i think it was ai.. maybe it is possible to do something like faceswap?
Thats dope
is this a 2005?
No no thats 3rd gen
Weird right? There are more older cars like this that im not able to generate
I want gen 1,
Its was pretty popular among europe
is that a 2005?
interesting, i'm also seeing either the classic or the newest gen
i tried a couple more things, getting the same results as you
try cyberpunk
I love this one. Nice work!
almost?
what are we challenging? the motor industry?
I just wanted to see if I could make that car by giving this GPT of mine the screenshots from above
somebody wanted a specific make/model
oh, do tell, i wanna join the fun
looks like a pretty good photo real to me pictures
ghosts and clouds are more fun, realy
it's a close approximation of the first-gen model, but not 100%
well nothing is 100 percent really in image gens yet imo
correction, it's more like 72% accurate.
well, didn't exactly cyberpunk yet. but got this vertigo inducer
I thought it'd go more neon but it's a cool frame
this is interesting, chat started doing sql queries instead of json snippets...
it sometimes titles my conversations in Spanish
Thats pretty close, how did you do it?
maybe if you refine his process you might achieve 99.995% accuracy
gonna see how nice we can make some nebula clouds
you guys always come up with some fun creative stuffs
This. Iβve made some nice ones but Iβm finding itβs hard to outshine the real ones.
accidentally created pretty appealing and functioning image 
Cool in so many ways though 
Dr. Paul Stubbs : A good try, Counselor. But sometimes, when you reach beneath a man's self-portrait - as you so eloquently put it - deep down inside, what you find... is nothing at all.
looks like a good party haha
going for an intimate jazz club scene feel
I am always fond of the creature under the stairs approach. Lovecraftian Cthulu mythology is fun. And with that, off to bed, before I get yelled at!
I think I found a new project to do with chat and themes.
@dim cradle that last post is anything but peaceful and it doesn't give the sense of serenity π’
i know...
Poor Merlot won't be able to travel the world with his cloud... ( #daily-theme message)
Discord is the easiest way to communicate over voice, video, and text. Chat, hang out, and stay close with your friends and communities.
and yes, I named the cat Merlot
Itβs a sacrifice we all make when we raise awareness but it helps make the world safer for Merlot.
true that
Is it jus me or the daily theme aint workin
I'm able to access it just fine.
Credits arenβt working at the moment
Ah damn i wanted to make a wallpaper
According to mods we should get more info later
Ight
If you need a wallpaper, maybe I can help. Not doing images atm.
Just wanted a cool mr robot wallpaper for my phone
Hmm Mr. Robot. Is that a character?
