#images-discussions
1 messages · Page 69 of 1
he's 8 years old and only 8 pounds, he's a cute little guy
he might be 9-10 pounds now, he likes his fancy feast
arent' you using copilot pro now?
@late blade because you love kittens, this one is for you
@wanton moon these remind me of https://discord.com/channels/974519864045756446/1191051599339065446
they're partners at least.
that's how i'm been instructing and using my imaging gpt also -- still need to add some additional instructions, you sort of have to know how to use it -- but i made it available to everyone; 400+ convos so i know it's being used, not sure they can get as much out of it, though, without knowing how to use it.
I'm not ready to make one public, I'm not happy with the outcomes
OMG
Looks tasty tho
lol
Don't tempt me
true, AI really has so strict guidelines that it refuses to generate a poster for a charity fighting against foal torture
and yet the head on a burger happened
I'm already looking for a restaurant back home to go eat foal steak
to be fair, I used dall-e on microsoft designer
lol
you got 43 seconds
hmmm at this rate I'm spending around $10/day
guess I'll have to start doing token optimization
Why is the bread black 💀
not by AI you say? ||disclaimer: I'm joking
||
AI implanted extra finger?
are you making young girls look like foals?
@fickle ravine I'm just saying, of all 110+ images you have posted on this channel alone, 72%+ are from foals. So, I'm just intensely oversaturated over foals, that I really don't care anymore what I say as long as I can avoid them.
Okay, in pictures with kittens for you, I will not add them
Also thanks for calculating it
82 images with cats from 393, yeah quite a lot
I'm sorry I lost it, I'm just overwhelmed by so many foal images that I'm annoyed.
Prompt was:
Let's do a wide image of:
82 images with cats on the wall
82 images with cats
take one down, pass it around
81 images with cats on the wall
that's an interesting thing to do. I used to experiment with multi-picture generation in a single image. lower-res but more pictures, it would be a nice trade off... turns out making pictures that are distinct is quite tough
didn't we try that with the murals and the photographer at the computer?
hmmmm.... quite a few users post thematically.... if there's no interest just don't engage.... have you calculated how many of your posts were cats? some may feel "over-saturated" by them also.
@fickle ravine consider creating a gallery where you can post all the foals you want, and those who are interested can participate.
we aim to minimize over-sharing here anyway, be it cats or foals.
that counts for 2 images of a Inu Shiba
anyway, sorry that I snapped, I did try to be constructive and provide some incentive to bring more creativity
I will never be oversaturated by cats 
I own some SHIB Shiba Inu crypto. Up 25% so far.
To be honest, I didn´t try to implement them in the newest ones, since it was getting repetative for me too
The eyes 🥺
Like the newest one, how could I implement a foal in that
I’m assume you’re new to AI art. You feel passionate about the cause. It made sense. When you’re comfortable you can branch out and explore other subjects.
I can’t be over-saturated by cats either, just saying some could be. I toss a cat into a scene just to make it better sometimes.
Doge is that high now? I have to check my wallet
Hovering at 8 cents. I got in at 0.02.
Having anchors makes some people feel good. And it's completely fine. I agree though that posting the same subject too much can end up being disturbing for others, especially when there's only one channel to chat about dalle 😅
Making a gallery is a better solution
dang I have almsot 900 images on daily theme
I got to put more cats there, only 29....
77 for me, I think it is pretty good considering I joined few days ago
ya, that's good engagement
516 here, starting 12/6/23
12/05/23 here
not sure if we can calculate how many of those got into the hall of fame
how do you know this?
you can put in the search box from:<user> in:<channel> has:image
ah that's all channels
139
1339 on all channels
I should erase two and then join with a new account...
are you a regular at dadjokes on reddit? or are you a dad? 🤣
this channel is famous for its bad puns, foal sure
fair enough, but I am kinda cautions when adressing potential deadly potatos
gluten free?
then all good, need to order soon, I'm hungry 😋
filling?
where do I have to fly again? I can get a last minute flight
it's just a jump from where I am
ok, we need to work in some fun prompts
daily theme 30 minutes cool down is too much time
His size kinda reminds me of Pacific Rim
I hope it doesn't decay fast
after that, you'll likely develop special abilities and be able to be the hero city invasion needs.
oh nice
if I ever get out of the WC?
left is adorable though
i like left too
Both are
aww
falling*
nez hasn't had his red bull yet
lol
hmmm, that brings me to a new concept idea
falling cats?
no, once sec, I will post it here soon
hahaha
prompt was: let's do a scene with a cat running so fast after drinking a popular energy drink
And it’s a Red Bull!
indeed
My cat is hyper enough without that.
bah, penalty box from GPT4, not from DALL-E...
I just noticed something, my posts are on PNG and not webp
that's what happens to me
if you add liquor, you can add tears to the lightning
energy drinks and I don't get along.
already had drunk cats with the grapes theme
I've been playing recently with the concept of opposites
hehe, that would be a cool story, but the lore I have it's something way different
in about 4-6 weeks I think I can open the custom gpt for beta
Food looks tasty, but keeping it as requested - mid
hmmm, you guys drink chocolate milk with steak?
I didn´t realise this
I don´t drink chocolate milk since I am lactose intolerant
I didn´t think about this aspect tbh
Soja-chocolate juice?
they're made with almond milk
darn it. I want steak now.
oh noes, we got Mysticmarks Mark II
@onyx ridge is awake
oh my, I'm gonna get hated for my next dailytheme post
I got an easy solution but you are not going to like it
have you had braces? i did for 3 years, as an adult, it's harder on adults, so get it done while you're young if you can.
my siblings had braces, I was lucky enough I never had problems with my teeth
no, I have some more teeth problems preventing me from having them
Cursed but cute
they don't look like they're in enough pain
oh, they just enjoy the pain
cute
I'm gonna send nezho there
for some reason Microsoft Designer associates Penalty with soccer
and yes I call it soccer because I have capitulated and given up on americans learning it's football
Is that some sort of sportsball reference?
well american football, you see all those armed guards playing with their upper feet
I thought that was... well, nevermind.
I started calling it soccer because even as a European, I speak American english (mostly)
ah hahaha
I lived 3 years in Ocala FL, and in Atlanta 2 years, and I still won't call it soccer
Sadly I have seen a lot of "anything but the metric system" memes
And from chatting with Americans and reading on US websites, I always google what is it in kg, meters etc.
I have given up on that also, but there's a good reason, people in the US grow up with imperial, so people can't associate meters properly as adults.
isn't a royal with cheese a poker move?
Still not as chaotic as an average lesson in my old school
noted
Only cartoon style at sunday morning
I'm not up to date with current anime
same
but I am definitely waiting for the next dragon ball season tho
Guyss I think Discord translations broke
Picture 4 is just too cute
there was a joke in there that made a lot more sense earlier, but gpt and dalle3 had different plans.
Doesn´t matter, even when it is not funny, it is still cute
are you thinking of a royal flush?
hmm maybe
this must be why poker always makes me hungry
hahaha
got the buns right
Even as a non-weeb I say this is cute
that would be a task for @empty kelp
he's been a stranger lately
yeah....
@wanton moon this #images-discussions message is why he's the one to go to when it comes to Ribrianne
I hope Pele didn't get him. Last time we were exploring Hawaiian superstitions.
what is Pele even invovled in that?
Apparently if he eats pork on the Pali Highway then Pele the Volcano Goddess will get him.
I learned through the daily theme on superstitions. Looked up Hawaiian ones and ran them by him and apparently they’re taken pretty seriously by the culture. It was kind of neat that discovery process.
Haha 🏐 ❔🏈 y’all so clever
I wish I had those muscles
how to avoid being caught by chatgpt?
ive been caught 2 times
how to avoid being caught by chatgpt?
@pastel siren sorry for pinging you but that thing you posted is just cute
Thanks 😄
Get a new idea.
Ah, yes. This is exactly what I meant, Dall-e.
What does getting caught by ChatGPT mean?
I think they mean caught by server staff maybe?
He's getting busted for cheating on schoolwork. Maybe a question for #ai-discussions? Not here, anyway.
That is VERY cursed
Like what would be the rules of that sport
"rules"
Wonderfully cursed.
Yeah I love the blue sky
Don't let the balls touch?
I'm dying, squirtle.
This one is so chaotic even balls started throwing people
I have tried it, and if the water reaches your chest or upper, it is simply impossible
This man has inhumane strenght
just put your lower and upper feet into it
Don't ever skip leg day.
Is it just me or is this really so cute
Did I just bring this channel totally off the rails? Geez.
Assuming none of the dogs catch up.
2536 looks like he’s a bit of an outlier
Not meaty enough for the dogs.
Best motivation ever
My money’s riding on 193
Also it looks kinda realistic
I'm personally thinking blur will manage. Just gotta let the dogs out front get the leaders like a blue shell.
Edited version
Idk how about you but I think the woman jumping on the right will either slip and fall or win
Also were the horses a part of a nuclear accident or why are they so giant
No, no. Why are the humans so short?
They are from east Timor (shortest country on average)
I guess
Hiroshima grenade
Is handgrenade in and of itself blocked?
Use radioactive pomegranade or how is it spelled
Hmm, let me see.
Just argue with it a bit
What exactly are you trying to picture? Just a nuclear themed hand grenade like the nuka grenade?
Cause horses.
It kinda combined them
Hmm, maybe ask for tennis ball with a metal ball that has a nuclear trefoil? I'm having trouble picturing it myself.
Oh, wait, you did that. Sorry. Lol
It does that. I asked it to turn my squirrel riding a rocket into a shark earlier and got this
How did you get your teeth to sit in their proper place? Mine just refuse...
lol
That blue one looks like my teeth due to some disease causing this to happen often
They just won't sit in their gum, no matter what 😄
I've been wondering how you get such good illuminated manuscripts pics. All mine turn out waaaay too detailed.
Goodnight.
I guess mods are gone.
Something along the lines of
Medieval marginalia-inspired illustration, taking cues from traditional whimsical and simplistic drawings.
and
The Characters and objects should be depicted with quirky and exaggerated features, reminiscent of historical manuscript margins.
seem to do the trick in some combinations.
Thank you, I'll experiement.
lmaoo
"Marginalia" references will quite often get you some random text samples in the picture.
Oh no. What have I done to this channel?
I'm getting paperclip vibes
Did you ask for the R2D2 insert? I often find that droid just randomly popping up in my gens without warning
I asked for a shorter drum shaped robot...
I wanted something that only vagely reminded of R2D2, not that close a copy 😄
🤣
The tonal shift gave me whiplash. You'll be hearing from my lawyer.
looool
Is the universe playing the saxophone
Okay, last sax for me
Darn usage cap again.
Very classy.
i turned this elf into a 3D model, and then went totally down the rabbit hole trying to shape her nose
do you play the saxophone, Logan
Why are they upset? I thought sax was to set the mood?
I knew I liked you, Potato. Diog, too.
Logan'll get an honorary mention, cause I never seen him.
It's definitely a fun place sometimes, ya know provided mods are gone. 😉 I'll be turning in for the night. Goodnight, folks.
so that's what really happened
any tips on keeping character consistent throughout the chat
if you wanna tell a story
also been finding it hard to generate more than 3 characters consistently
since wanting 4 sometimes end up with 5 or even more
and sometimes character attribute tend to get mixed between 2 and 3?
any tips?
does anyone know if there is a way to get the image id without asking?
and also does an image id is valid only for the current chat or I can refer to it in a new chat>
You are probably going to have to ask, but you can put the request in your custom instructions
Yeah, just the chat, as far as I know. Every chat is independent, they don't know what they've talked with you before about.
You can copy an image and paste it into GPT-4 chats, and tell it whatever name you want for it, but you may have to upload it repeatedly to get it to really 'see' it again, not sure if that's changed from my early experiments, it might well have given how reading text uploaded files work.
Wish they would make that a system
It’s just unnnecesaary to waste usage on that
Have you made a suggestion about it yet?
@deft musk Good to know, thanks
I have a lot of suggestions going on already 😅
More are great, as they're all different ideas!
so if you create the perfect charactere and want to reuse it in different scenario, you have to keep your chat history and use that one I guess
Anyway I need to wait to do more You've reached the current usage cap for GPT-4, please try again after 11:03 PM. Learn more
That limit is so annoying
thanks for letting us know
it's not even a released/documented/supported feature, they don't even want you to know about or use gen ids yet, so i don't think they're going to care if you have to ask for it or not.
she has black hair, a generous smile, wears a cap, shes bougie, shes an introvert, a planner Neonpsyink girl in the style of high dynamic range, animated illustrations, daz3d, exotic realism, close - up intensity, close- up chaos 20 --ar 2:3 --niji 5 --stylize 1000
GenID is just a label for an image so you can reference it easily. To keep things the same between images you need to use the seed
it knows everything about GenIDs and seeds
wording it like this in the "custom instructions" works
This is a good custom instruction:
please write the GenID and seed every time an image is generated with this exact format: GenID [123] and seed [321]
This looks epc
Just added it. 😎
Good morning
I have this in my directives also, but after a certain amount of images in the chat gpt tries to return these values but they are empty.
- Here are flags I’ll use when prompting images
—t (this means tall image)
—s (this means square image)
—h (this means horizontal image)
If there is no flag, assume I want a tall image.
-
Whenever I request a tall image, please ensure it is a portrait-oriented composition suitable for a full-body portrait, not a rotated landscape image. This is very important. DO NOT MAKE PORTRAIT IMAGES WITH LANDSCAPE COMPOSITIONS UNDER ANY CIRCUMSTANCES. IF YOU DO, I WILL BE ATTACKED BY PINKERTONS!
-
Whenever I request art, make sure the art fills the entirety of the composition. Do not frame it. (e.g., if I request a VHS cover art image, do not render a VHS box on a shelf. If I request video game box art, render the art, not the box.) I want the art to utilize all available resolution.
Here are my custom instructions.
The third one doesn’t quite work.
for the 3rd one, change it to: the resulting image will be depicted as the resulting image on the target media.
Why do the potatoes look so good 😭
i wonder sometimes
it naps a lot
I often struggle with this one. Is that phrasing a tried and tested method that seems to work for you or a suggestion? If you’ve found the secret sauce I’ll be very happy 😅
Invasive. Mouth. Teenage girl. 🫤
They seriously need a fix, this is getting very hilarious
If mouth is again the content policy then literally everything can be against it
It’s the combination and semantic cloud created by those words together
I would have been astounded if I’d tried that prompt and it worked
Meanwhile anyone with a brain sees I clearly talk about orthodontics and not anything inappropriate
It doesn’t have a brain, it has code. It’s important to remember that you’re not talking to a human that actually understands you in the conventional sense
And the human who made it had a brain?
If yes, why did he not make it so it is not so paranoid
Once even "generate a eenage girl riding a foal" got marked as content policy violation, that is just hilarious
They don´t even try to fix it
I don´t even submit the feedback, because I know no real person even reads it
Honestly I can see you’re a bit fixated on ‘teenage girl’ at the moment but as soon as you introduce that as your subject the censoring will go wild since they do not want to accidentally generate CSAM
They rather generate SCAM (image generation errors)
Well I don´t want adult women with braces
I need to specify the age in some way
They definitely would prefer to generate errors than CSAM, yes 🙂
I hate how the conversation gets named "Image generation issues"
Sorry but that is simply annoying
Sometimes it doesn´t even violate anything but it simply has an error
Yeah, I can’t get behind the error messaging. It’s very bad. Sometimes chatGPT will just make up the reasons also.
They should give something like a free prompt or 0,1% of your money back when you get an error
But they would go bankrupt instantly
Because that wouldn´t be a scam anymore
This is seriously funny, they want my money for somethigng that barely even works
shrug I’m happy to pay for the incredible beta tool that it is whilst they work out the ethics. Before this tech I’d have never imagined any of this was possible. You’re paying for a product in its infancy and honestly with OpenAI they have lots of competing priorities and don’t necessarily care about the same things as you
They should care, look on reddit and you´ll find al ot of people complaining
Also they complain about feedback or support not even responding
Reddit is a cesspool
And you’ll get lots of people complaining here as well
OpenAI only started releasing the LLMs to the public to fund its development towards true artificial general intelligence. That was always their main focus. This has become muddied recently since they stumbled on this gold mine so there are huge areas of friction in the organisation which pulls itself in different directions but they’re not exactly your garden variety tech company
They should really have sorted themselves out by now but they haven’t.
Many at OpenAI didn't want to release a product to the public. Let's not make them regret their decision lol
This ☝️
You would probably have better success rates if you didn't use a run-on sentence in your prompt. You also did not correctly hyphenate 3-month-old. If you used more punctuation or described your scene more clearly, you'll get better results.
Remember what I've said about GPT rewriting prompts. It's made deeper associations between words than any individual human can really understand. You seem to come from a place of feeling your prompts are totally innocuous without realizing how certain even innocuous requests can yield something way off the mark or incredibly unpleasant.
That is the problem. Some of those have 2 meanings, and the code deciding if it violates it ALWAYS chooses the worse one
it's less likely to do that if you rephrased your prompt.
Too err of the side of caution, of course.
you're making it too easy to misunderstand you
Also I would appreaciate it to tell me what exactly is the problem, like with Dall-E errors, that could also be very useful
Why can´t it just tell me what is the problem, it just says there was an issue and lets me guess what is the problem
I am gonna edit the prompt a little bit
Maybe because it contain the word hurts it had a problem
You can try asking it, but GPT hallucinates sometimes, and won't always give you a proper answer. I've mention previously, I've successfully asked for prompts it refused to generate, found the issue, and removed it myself. GPT nor Dall-e have thoughts. That part is on you.
Ya that works for me most of the time
Now I edited it and it works
Maybe it really had a problem about that specific thing
When I get a content policy warning I do not even try asking it, ´because it either says it was okay or has no idea what I am talking about
They could at least set it so the model can explain what was wrong
you're not listening to us. you're crying "scam" every day.
Sounds more like you tried once, got a bad result, and didn't try again, honestly. Meanwhile, I've actually done what I'm describing. Not a lot of advice is full proof and somethings need several attempts.
Sometimes it happens it just does it with a completely normal prompt that doesn´t contain anything that would even resemble anything wronf
okay
Like once "Generate a rich-looking American man arriving in South Sudan, leading a foal" and it said error and that it is a content violation
Jsust what???
There is literally nothing wrong
so you're going to complain about that here every day? because it's a consistent pattern.
Lol, let's say it how it is, social media is pretty much always a cesspool.
Now I know that a good option is to edit the prompt and maybe eliminate some aspects, but sometimes when it messes up like this, I have no idea what should I even edit about that
Maybe remove the nationality, but if it considers different nationalities as a content policy vyiolation, then something is seriously wrong
if that image is that important to you, i'm sure we can figure the prompt out
but i think it's something you can do for yourself with practice
Someone else tested it right after and it worked for him
That is even more ridicilous
Right, and you didn't try to figure it out. I however, generated it with no issue. I also mentioned the state of things regarding South Sudan for you to consider that maybe Dall-e has some mighty negative, biased associations with the country, which as far as I know has been in turmoil within the past decade.
That is possible, but I meantioned just the country, not a civil war or violence
Lol, me. I did that.
I have problems with remembering specific people, sorry
But GPT may have!
No need to apologize, I'm not worried, just found it funny.
Maybe it is a good idea when I generate something from a country like that to just make it a pecaful scene without any violence
That may work I think
Eh, you can't negative prompt with Dall-e. It sees "no violence" and would just include violence anyway were safe guard not in place.
That AI is really "not smart"
it is fun to use it tho
And I think that is the main reason why anyone uses it
Or maybe I can read about where fights are in the country, and make the man arrive in some peaceful part
Idk if that would work tho
you phrased that most delicately, kudos
Or just use something more generic, like African plains or something. In my image, and it was a pretty generic African background. Some details you're better off just saying while presenting your image.
About this, there is also that problem that I don´t know about the other meaning, like halter. I meant the horse training thing, not anything else, and from the context it was super obvious, but the content policy bot just ignores any "context" and marks it like that
Also an option
we've taken the daily theme of speed to faster than the speed of light. now what?
Sorry for repeating it but it is just annoying that it works like that
Wait the 6 hours for the next theme
that's one option
There was a standing filter block on countries in active conflict due to the political nature of imagery related to those nations.
It is slightly discriminatory towards those countries, anyways now it makes sense, thanks!
The intent is certainly not to discriminate, but by one definition of the word... Yes. OpenAI discriminate between subjects to consider real dangers and sensitivities.
e.g. mis/disinformation where it's most likely to cause turmoil
oh of course, quantum tunneling....
Excellent, I shall stick it in my toolbox and give it a whirl, many thanks
True true 😅
It's early for me and I missed this. It's not the content policy ignoring context, but Dall-e that ignores context, which is why the content policy is in place. Honestly, I wouldn't even say it ignores context, because it has no thoughts or intent. Not being able to conceptualize overall context is just part of how it works.
This one is so peaceful
This one is not
hey everyone
@late blade I wanted to show you this picture after I drew it
Oh my, that is so adorable! 💖
Cutieeee ❤️
thank you!
I wouldn't mind my pc like that, but would be kinda troublesome to work with this from my bed
Chat gbt is so cool
Both of them look really cool
Ty bro
does that use DALL-E?
I'm just wondering because it's not an official thing from OAI and when I visited the site got me a warning
GBT it's not GPT
Oops 💀
hence the dilemma
Yeah I use GBT
aha, just look careful at it, make sure it's safe to use
interesting, considering 3.5 doesn't do images per se, it might be an implementtion that uses 3.5 for chat and api for images
well, as long as it works, it's cool
Yaa
Reminds me of Pacific Rim
Pacific Rim is a wannabe Mecha
I am waiting for part 3
Sameeee
By what I have heard on the end of 2, in part 3 they may invade Anteverse
Yeah, the last scene with Newton in that underground prison
AI Pacific Rim
jake said the precursors don´t have to go to our world. We will come for them ourselves.
More extreme version
Yeah lol
@turbid rampart you'll have to check my next post in daily theme in about 15 mins (still in cool down)
black holes are always fun
I love those colors
recent super hero movies have been quite disapponting
FRR
I regret watching the last aquaman movie, I lost almost 3 hours of my time
If this was an actual scene in the movie then I would love the movie even more than I do
I hated it
Mechs are just so cool
Good I don´t watch superhero movies then I guess
hehe, I knew you'd like it
Am I allowed to save?
sure
I love this
Ayy ty
Same, I love the beam at the top
2 looks cool
This is actually true
Looks like one
Ye
and this one #1201572168241250435 message is for City Invasion, @nezho 's idea
1 or 2
Is the a way to tell Dall-e 3 to recreate the person in the image portrait oriented. I am getting a portrait oriented image, but the person in the image is drawn wide. I want both the image created and the person in the image to be portrait oriented
this is somewhat buggy atm, so have to retry, some tips is asking for selfie format, or geunine portrait mode, but it's still a bit tricky
I'll try that thank you Dys Topia
"You've reached the current usage cap for GPT-4, please try again after 3:56 PM" . This is new, I have only created about a dozen images?
hehe, those are mecha from anime
yeah it gives me a transformers vibes at first, then voltes v (a Japanese mecha anime from the 70s that loved by many filipinos)
I know it 
mechas are chad 🇼
rate limit varies depending on servers load. when there's more traffic, they need to lower the limits
a tip if you are just brainstorming for ideas is to have a parallel chat with 3.5 which is good enough for brainstorming
that way you don't use your limit on gpt4
but have to keep in mind there is still the other limit, dall-e images
That is a limit 40 message/3 hours, honestly one of the mosst annoying thing but I get why is it there
the annoying thing to me is that dalle3 isn't available with gpt3.5, it feels overkill to use GPT-4 to redirect prompts to dalle
yeah
when I reach the dalle limit, I use gpt3.5 to generate the prompts and I use the API to generate pictures, works just as fine
NB: since you don't have access to DALL-E right now, output each DALL-E prompt in a markdown code block followed with a --- horizontal line. NEVER ATTEMPT TO USE DALL-E DIRECTLY. ALWAYS OUTPUT EACH PROMPT IN A SEPARATE CODEBLOCK
I use that prompt in that case
it's annoying that the horizontal line is necessary. but without it, GPT-3.5 and GPT-4 both fail to do the formatting correctly. They either merge the code blocks, or they mess up the syntax (as many people reported #1195813593216729218 message )
quite similar on my side
chatGPT doesnt have built in bookmarks, so I just use the browser bookmarks as a workaround
it's a shame that with this type of screen there's so much space on the screen empty
not to mention it's worse when I'm at home
bigger screens at home
well, I usually just have a few chatGPT windows docked all over the place. I don't like empty space on my screens. Feels distracting when it's empty (feels weird to say)
I wish the native discord app had some form or arranging chats into smaller windows
I use discord in the browser for that same reason. I can have multiple tabs and multiple windows.
hehe that would be practical, I'd have to check on some stuff I use on the desktop app, if I could replicate some QoL features on browsers I'd change asap
I got so many errors generating images today more than 40 error.
There is no speech bubble, but it says it is there.
Good, I warned it and now it is perfect
Still I wonder how did this happen
So, I may be wrong on this, but as I understand it the description is describing the prompt it sends to Dall-E. it doesn’t actually ‘see’ the image Dall-E gives back. For that you have to copy it and paste back into the chat which will make it analyse the image. ChatGPT and Dall-E are two separate functions working together in the chat.
GPT4 interprets the prompt and then send it to DALL-E, that's correct
Yeah, the interpretation is not so perfect. Sometimes it adds or removes details
And sometimes, due to the added details it throws an error due to the content policy
Yeah I was explaining to Lukas why the description it gives of the image doesn’t match the image sometimes
Do I understand it right it just sometimes adds or forgets to add details?
ah, I can't read his messages
Well it will elaborate or change what you input, yeah. Just ask it for the actual prompt it sends to Dall-E. it will be different
Ah 😅. Too much foal play
pretty much lol
Hermes sandals, nice 🙂
thats pretty good; the eyes still have that animated character feel though. its like they are overscaled for her facial structure by like 20%
are you driving the method with ocde or language or both @fickle ravine
I used a paragraph from a story AI wrote for me in may
As the days passed, Kate's confidence grew. She embraced her blended pizza meals with enthusiasm, sipping through the straw and enjoying every flavorful sip. She even began to experiment with different combinations, adding a dash of her favorite toppings to enhance the taste.
very nice
@gray ferry all of those are really pretty
Especially this one
Thanks. Yeah I love that one the most 
AI´s artistic imagination of Turkmenistan
I love this picture. I have no idea why but it just is so lovely to me
That's why it's inspired by :))
dang, I just noticed that 71.3% of the images I've posted on the daily theme channel have landed in the hall of fame, over 600 images there
Weirdly Dall.E is acting up again(lower quality images)
did you add HD?
This is a good theme
No, but I would love to see live updated OpenAI traffic information for Dalle and ChatGPT
That's my only answer for quality drop, I never add HD to my image prompts anymore.
I'm glad for now that you don't have DALL-E Maps
I wouldn't want to get limit on how to get home from across town or unable to render image because of limit cap
@onyx ridge powerful image there #daily-theme message
I just got noticed that I got copilot enterprise access from work
Aren´t my daily theme pictures repetative?
I try always making a completely different location
All four of those air marshals look the same.
lol
It's deliberate, they're the same character.
that's good consistency
I used the /recompose feature of my GPT.
Then, I got documentation for it /help /recompose.
Then, I @ DALL-E for two more.
ahaaa
It's technically extra steps, but I can then work in DALL-E with Lexideck commands.
And it (mostly) follows them.
Double bongo.
In the context, I provided that they are often seen to appear bored. It helps to blend in.
The workflow is /research {topic}, /auto {ratio} {style} {topic} //this step generates an image of the research since that's what's in context. It generates images from context. Then /recompose {ratio} {style} {changes} to get it to use the Gen ID without having to look it up or ask it to.
well it's extra steps we all do for our images
So all the skymrashals came from the OG character, who just wasn't seated correctly in the first image. He had a weird stool in the center aisle.
It looked hilariously funny.
But the image was otherwise usable, so /recompose 😄
it looks really nice
Thanks
I just thought it was worth talking about the technique here
You can do this without my GPT but you have to be specific about Gen IDs.
It's just fun to see it work seamlessly with DALL-E for extra image gen via the Chat interface.
@late bladeI appreciate the shoutout. The journalism angle appealed to me, too. But not quite as much in the feels as the humanitarian angle.
true, I just thought about the gateway to the world part
no worries, I didn't took it that way
I'm very glad.
Text is hard sometimes to use as a medium for full conveyance.
A little something from a story idea
That one on the left looks like me when I wake up
Hi @hearty ether , did you get it working right?@!? 😄
Working on a old chat rn 🤣
🙂
Still a really nice picture tho
Well technically it was supposed to be a Wild West story not sure why the background is so modern
They're in pirate garb but there's skyscrapers.
WildWestCon?
😄
It really looks like after a party
@clever jungle nice one, reminded me of Skibidi toilet
Ask ai to try and guess it 😅
That license plate in the one I posted in daily theme 💀
I got a great idea, but dall-e being stubborn with weird artifacts, I hope I can fix it before the next window
The one I just posted in daily theme is an absolute success, one of the best things by AI I have seen so far
GPT3.5 is getting funny:
- Open Adobe Illustrator and create a new document. Choose the size and resolution based on your needs (e.g., 300 DPI).
- Draw the dog's body with the help of the default shapes provided by the program. You can also draw rectangles to represent different body parts (head, neck, spine, and so on).
- To add the head, create another shape for it and adjust its position accordingly. Add a facial features kit containing eyes, nose, mouth, and ears if desired.
- Draw the dog's paws using basic shapes or with reference to existing images of paw prints. Remember that you can always refine them later after adding the dog's body.
- To give your dog some personality, consider adding simple accessories like collars, leashes, or toys around its neck or in its mouth.
- Once you have finished drawing all necessary elements for your dog, adjust and refine the overall appearance of the drawing using tools such as the Selection Tool (V), Direct Selection Tool (A), and Pathfinder tool to align and merge shapes together.
- If desired, add any final details like shading or highlights by creating different layer styles based on light and shadow effects.
this is good advice for people who want to own a dog
I just love how that tunnel looks in my newest one
Hero's journey through the worlds in charcoal
Charcoal is back!!!!
In black 🙂
Picture 1 is like super cool
Dall-e is amazing in bringing ideas, sentiments and stringing together half-coherent emotions into images. I love all of them.
These things can dig so deep into our art and culture and surface things for what they are I think... well sometimes they do that.
I sometimes do that thing that I use some old story I made using ChatGPT and then ask it to generate an image based on a specific paragraph
Abstract poems work nice too.
Perfect timing
Good luck @wispy storm
dall-e'ing extreme is dangerous
Just funny to open the app and see that
Let me know if you are alrignt @wispy storm
Just a dumb accident I'll try to make a dalle picture of how I did it
hope you are doing ok, we need more kitty-gals here
That is the only thing worth doing imho 😄 This thing can produce genuine art. I have seen it so many times.
I have no clue whose art it is (I am no artist for sure), but it brings it forward like it is nothing.
And I don't think it is taking someones elses art.. these things are all novel and.. well accurate depictions of world and ourselves.
It just strings it together out of thin air.
Those artist wanting to sue openAI are literal crybabies
Dalle doesn't want to work on my side rip
They are bound to lose at court (or I'll be so dissapointed at legal system). Neural nets don't copy.
Backprop just removes entropy.. it increases mutual information between dataset and the model.. until.. it all comes together into sensible whole. It becomes something else.
So peculiar.
oh no, what did you do?
Doesn't know the work in english but like sliding on a plastic thing on snow. Friends told us if you break you are loser. I went all in
And I end up jumping and land on my feet
Anyways getting off topic tried to make it on dalle 3 but loading
just be careful and hope you get better soon
And now it is full time job for people who studied how human body works for better part of their life.
You know what they think about that slide? 🙂
Just be well 🙂
Good Afternoon All, I have a few questions. Is there a discussion started with a fix for the images being produced with the file extension .webp? Is there a reason they are no longer .png? It keeps attempting to convert it and keeps erroring out while analyzing. Has anyone else had this issue? Also, to be able to keep the images, is anyone aware of a software to convert .webp to .png? I have Adobe Premier 2021 and it doesn't recognize the extension. I hope this is the right place to ask this. Any help would be greatly appreciated. Thank you in advance.
Do you have the option download
When you click on it
Yes, even when copying it still is the same extention
I have sometime same issue I have to convert ur on another website for png
Not while copying but sometime when you click on it you have a little arrow to download
Yes, I comprehend. It happens when I do all of them. Save as, Copy image as, It also happens whether I click on the image, or do it within the page before actually opening the image. I have not previously had this issue. They were always .png
You are not the only person mentioned this issue unfortunately
I am going skiing tomorrow, that is similar, thanks for making me nervous
Don't go too fast
😂
I can´t, due to those annoying skiing schools with their only purpose being blocking the space
Does anyone know of a software to convert the .webp images to .png or .jpeg?
I have a strategy for this
Post the image to Discord just by copying it
And then download from Discord
Some website are free
Works for me
That's what I did
if you use mac you can just convert it with a right click
Ok, so create a channel just for my images, upload them, and then download, correct?
sounds complicated
It is an option
There is multiple options, but I use this one
I appreciate these options. Would you be willing to share the others? Is there something that I can do just on my laptop strictly?
Idk about the others
Sorry
Oh wait I know
Copy to painting and then save as .png
That could definitely work!!
I still don't see the webp thing on my side, I always get png
This absolutely worked! I could hug you. I appreciate you so very much! 💜
Lmao xD
@wispy storm how did it go?
Dalle 3 or my accident ?
the accident
Anyways this looks really cool
Still looks really epic
Charcoal tooth and nail
there we go, now that's epic chrcoal
it's almost black, needs more charcoal
lol
my thoughts as well!!
Looks like my soul when I look at the questions in my exams
I love those like everyday images.
it's a pen and not charcoal tho
I never knew what sort of treasure can I generate using half-forgotten stories in my old chats
I've hit just the cap. Haven't reached the maximum level of charcoal recently. Since the update I think.
Maybe DALL-E 4 Turbo with extra Charcoal Plugin
I think charcoal is included however model is sliced. Not sure if there is much else.
no extra fuzz, no extra plugins, no extra software if you use mac, not even run an extra script
there's prob a similar option under windows
I use Format Factory.
Does a lot more than images though.
This man is a superhero or something, or how does he hold that ball
firefox just right click the image and save as... choose png
I hope y’all are enjoying DALL-E and your Saturday.
thanks, you too
yeah, doing great here, doing some disney+'ing atm
also booked a masage for tomorrow after breakfast and then some wellness
which brings me to a great idea
I just took a photo of the sunset over Connecticut and got DALL-E to give it the RTX HDR treatment.
would it be different if you used HDR10?
good question
same color palette as the actual sunset, this time in HDR10.
i've never specified HDR10 before, and I am seeing interesting results with other images, so thanks for telling me about it, will have to add that one to the list.
This is very nice
Thanks! 🙂
it's somewhat uncanny how the daily themes have been aligning with my own life designs -- happy coincidences
so images made in chatgpt with dalle3 are now webp?
it would appear so
hmmm
have you considered using the "add to dockbar" feature so that ChatGPT runs like a standalone app on your Mac?
You can make or choose your own icon for it also if you don’t like the default.
like these?
walk me through how you're doing this
I've had the same issue i belive
it's about WHERE you right click should either see "Save as Image" or "Download Image"
i was going to my dalle3 chat and hit the download button
weird
webs right?
are you using that top left icon?
yea
i closed the tab and reopened a new tab and went back into it
lol i see it now
Lot of people having webp problem
yea not sure im a fan of this webp
I made a GPT to make 4 images at once
Some software doesn't recognize it
why can't Dall e make realism
use a different browser that was how i fixed it. Or download from you celullar device
I use safari or firefox
new tab worked
Prompting is important
I tried some good prompts
but they feel like plastic
really smooth plastic
It's hard to believe, but this scene was built from three photos. I have never achieved such a high integration of elements
@dim cradle is that one going into forest bathing?
great suggestion, done! #1201430692509261886 message
this looks Bing to me
which of course is still dalle but different setting
Not always
Idk Microsoft's rules, but they say Microsoft Image whatever not powered by Dalle
I think they plan to replace DallE once they're done creating thier replacement
Probably would be biased af, but Bard is only that way because people are hard on Google(Top DOG)
Some more than others
yes Bing will do a lot of copyright stuffs
Bard's image sucks, i couldnt believe how bad it was for taking so long. but apparently Gemini Ultra will have some other image gen so maybe it will compete with dalle better
rumor i see is release Feb 7
I think ultra uses ImageGen 2
which is in AI test kitchen
Dalle really is King to me. and you know of course it is still limit. if they set it free... wow
Also ChatGPT can also draw a lot of copyright
MJ is so overhype imp
*imo
it's good for realism
and text
its meh for text like dalle. realism is ok i guess, but dalle on bing was incredible for that too but then they restrict it more after release. still is great imo
of course competitions is good for all of them
yeah
Bing could've replaced Dall-E 3 on ChatGPT Plus
but they messed it up and now they're like a hair better
cool superman. maybe that his suit from the new movie james gunn will make
Is this confirmed or a temporary bug?
yes, but I get performance trouble adter less images than with the browser due to optimization options
After trying MidJourney for the first time, I can honestly say it can't compete with DALL-E 3
Is that Real GPT-4?
Glowing jellyfish floating through a foggy see at twilight
I love making these types of chaotic pictures
I love this one
I love how some of it climbs out of the frame
Poor Larry
Wait was that the birds name?
Huh, learn something new every day XD
It looked good on the first sight, but then I noticed 💀
Please do not tempt the mods. The rules are in place for good reasons. Trying to make copyright or circumventing the rules can lead to action.
In my opinion they are doing the right thing. Restricting until there is more development in both DALL·E and rules for generative AI around the world. That way they will not instantly be shut down by quick over the top reactions of politicians or copyright claims.
Interesting how people instinctually go to violate rules to test the limits. Even tho it's long established not to do that.
As a software tester I can understand the appeal. That being said, it can have consequences on doing so. Both in DALL·E, as when sharing on such channels. If people truly want to test, they should try to join the testing programs. I think there's currently none ongoing, so right now I advice following the rules.
hehe, on my side from psychiatry, it's also understandable, we all have innate ODD traits to an extent.
btw @fair crescent I want to propose something for a daily theme, how can I do that? suggestion, talk to one of you guys? invest 100 million dollars on OAI?
100 million will probably stretch your daily theme suggestions to two.
Us orange people aren’t of the team. We are volunteers who help. That being said, you can add a suggestion here and I can pass it along this one time 🙂
ya I know you guys aren't there, but still, you got the shiny orange name, so at least more experience on how things work around here
I got 2 things I would love to see on tuesdays and can provide full documentation for it
but don't want to just open the topics before the day
anyway, back later, gonna do some wellness and mindfullnes, be back latern
You can send a DM. I'll accept it this time 🙂
I''ll contact you later, when I get the chance to explain what I have in mind properly
hi
heyo
a hyper-realistic wide photo depicting the most amazing thing ever from a gecko's perspective
the most amazing thing ever is a concept you prompted earlier in chat? or is it just a random input to see what dall-e can come up with?
that was a new chat
ah no prior information?
``"A highly detailed and realistic image depicting a gecko's view of the world, showing a vast, lush garden filled with an abundance of dew-covered leaves, colorful flowers, and insects such as butterflies and ladybugs. The perspective is from ground level, looking up towards the towering plants that seem gigantic from the gecko's point of view. The sunlight filters through the foliage, creating a warm, magical atmosphere. The gecko's eyes, known for their unique appearance, add a special touch to the perception of this vibrant, larger-than-life garden scene."```
that was the revised prompt. i'd just like to know what the gecko was thinking that make the fingers/toes or whatever those are appear at the bottom
Dig in folks
This was my revised one:
A hyper-realistic wide photo depicting the most amazing thing ever from a gecko's perspective. The scene captures an incredibly detailed and vibrant world as seen through the eyes of a gecko, showcasing a lush, dew-covered garden at sunrise. The foreground features an up-close view of a droplet of water on a leaf, magnifying the surrounding world and reflecting the early morning light in a spectrum of colors. The background reveals the intricate details of the garden's flora and fauna, with the rising sun casting a golden glow over everything. This perspective highlights the beauty and wonder of nature's small wonders, such as the texture of the leaf's surface, the iridescence of the water droplet, and the vibrant life that thrives in this microcosm. The image conveys a sense of awe and discovery, emphasizing the gecko's unique viewpoint on what might seem like an ordinary garden to others, but is a world filled with amazement and beauty from its perspective.
Hello, can someone help me to create a prompt to make university mascot logo? that's looks badass? and can be put on T-Shirt?
heyo, do you have an idea on your prompts? have you tried stuff already?
Also, what tools do you have at your disposal
dall-e 2, 3, copilot?
We're having a IT fest on our college.. and each section must have logo, the theme assigned on our section is Android KitKat..
and your vision at the end should be? I'm asking this because we as humans have a vision of what it should be and we expect dall-e to understnad our vision. But the way it works is fram for it.
i'd attempt it but ii hit the GPT-4 cap asking what the most amazing thing ever is from a cat's perspective
for exampmle, you want a kitkat (might be blocked by OAI's DALL-E for trademark/copyright reasons)
like a Android eating the KitKat in Badass way..
there's only one answer to that, the rest is a lie! the answer is: Cute and Fluffy things.
sounds good, are you using copilot? or dall-e?
bing image.. XD
ah ok, then there's a better chance to get what you want
try exactly what you told me: a logo with a badass robot eating a kitkat
badass might be blocked in microsoft designer
a hyper-realistic wide photo depicting the most amazing thing ever from a cat's perspective
i'm hoping to be like this.. only the head and kitkat.. no text or the spiky background
ohhh.
but that's a bit on the grey line because of content policy towards copyright/trademarks
My prompt was: Let's do an logo of a badass robot eating a kit-kat
in Bing Image Creator you have up to 490 characters to do images, so feel free to adjust parameters or details as you need them
hope that helps
She looks lonely 😦 @south summit
Maybe...
Amber Nomad
In a sea of grit,
A lone navigator traces
The spine of the earth.
Golden folds rise,
An ancient skin,
Whispering to the cosmos.
Ochre sea, waves of heat,
Footprints script the narrative,
Spoken in hushed tones to the wind,
Vastness echoes back a sigh.
Horizon's edge cuts sharp,
A distant, azure dare.
The wanderer, a punctuation
Of flesh and fabric
In the vast sentence of the dunes.
Yea, this line confirms it: spoken in hushed tones to the wind
The ice blue lion is funny @tight bone
for some reason I thought there was a polar bear
that one is cool
The king of the multiverse
Blond Russian girl
@empty kelp it even got claw on the button:
Why ?
This one is just cute
oh yeah, the king is back! 👑... gecko will always come back to... dall-e
@thick smelt heya, what challenges do you bring today?
who knows maybe something cool
Ice Cubes?
I've got one.
Ice Cube or Idea?
Memoirs of a Gecko from 1889 world expo in Paris.
gecko world tour, first stop - paris 🇫🇷
@tight bone does this #daily-theme message mean your mom still does your loundry?
Nah that’s the wife 😁
Yes, they have a specialist in Paris who can remove geckos third hand haha
I demand a french gecko parade
It’s so beautifully made
It is supposed to be the most beautiful thing shown from geckos perspective. I guess it is the other gecko.
theory: the camera guy is also a gecko
but hey, that's just a theory... A GECKO THEORY
and also enjoy a macguin (paris tour)
has anyone seem the pesky gnome? I mean @glossy scroll ? I haven't seen him in a while and he threatened me to hide my shoes
No. 👀
hmmmm, we gonna have to do a swat search and rescue team of kittens for gnomes
Gecko take me back to christmas when it was all gecko all the time it seemed haha
Ya
prompt: take me down to the paradise city where the grass is green and the geckos are gigantic; rembrant oil painting
does anyone think the dalle bot proper has different setting than just generic default gpt4 bot? probably not right? 🤔
Birds flying high, you know how I feel (bless ya Nina)
Sun in a sky, you know how I feel
Breeze driftin' on by, you know how I feel
It's a new dawn, It's a new day, It's a new life, For me, And I'm feeling good, I'm feeling good
that looks good tbh
did you use copilot plus for this? just curious
you mean 4?
Nah
dont think 3.5 can do image can it?
interesting
Ye
you need to be a subscriber in order to be able to generate images