#images-discussions
1 messages · Page 40 of 1
No it's allowed, just shunned upon
By who?
I'm totally kidding by the way
not a realistic scene. like realistic materials and lighting
yeah, i need the helm on isntruction and fine tunning work. lmai
right. thats out of Dall3

just can't go wrong with surreal
all image gens given proper training can do almost anything fine. Almost
Rip
definitely intimidating
I liked the cute innocent elephants better
This is too evil, I don't make evil pictures
I was truly out of ideas for mittens by that point 😅
haha pretty desperate
Cute elephants please 🥺
it biohazard hybridized a mantis and a shrimp. lmao
"You've reached the current usage cap for GPT-4. You can continue with the default model now, or try again after 1:45 PM." I guess i'll have to wait later to improve on the elephant... hehe
one from slightly earlier
DALL-E is really wonderful. i feel like i could just make bigger and scarier elephants all day long. it caves with more and more gemstones and iridescent moss covering everything
I just picked up an apple from the fridge, I washed it thoroughly, I took a bite out of it. I chewed it one step at a time. Then I swallowed it. I took another bite, etc
Soon I will eat an orange, and just wait til I tell you guys about that

i know what you mean. but you can't simply compare oranges and huge terrifying elephants of course
it's comparing apples and oranges, really.
unless they're huge terrifying oranges. will need to wait another 30 mins to test that
what's my enlightened elven friend gonna do next? stay tuned. maybe some meditation.
makes me imagine an image of an apple battling an orange
I'm aching inside
there's no comparing
Please tell me so I know about it
now i'm really curious what this orange and apple battle going to look like. it's something that only the AI can show us
Lol
that's a strong beginning
@grizzled loom consider creating a gallery for your memes, it would make a nice addition
i haven't decided yet on a subject for a gallery
i have so many.
have other ambitions, this is fun and study in one; im working on a new system for image gen and takign a lok at all weaknesses
but i do love sharing laughs with other artists and users in general
its the human element that makes this thing called life horrific wonderful fun
i do prefer to depict the full spectrum of the human condition
we don't share equal lived experiences, our art is a medium to express our uniqueness
absolutely
its our history
before we could write words we drew pictures around the world
its core to all of what we are
we should let the orange and apple team up so they can confront the elephant
archeologists just uncovered a 27,000-year-old pyramid, i wonder what secrets it might reveal about our true history
recursive lost layers of history; we become our own mystery constantly
the orange and apple can join forces with the "mantis shrimp foreman" (Again, no relation.), who suspiciously transformed into Indiana Jones when the elephant got bigger
have you guys noticed a huge diference in Dalle-3's ability to generate text onto images when you use ChatGPT vs the api? I'm running the same prompts and the chat gets it right within 1-3 attempts but the api maybe gets it right with 60 attempts
i overwrite it and remove their instructions, because they are biased and have contradictory isntructions
"Supercalifragilisticexpialidocious" Go!
"you are out of generative drugs. wait inline for more generative drugs. watches ChatGPT hang out shady alley window passing out image gens
so i should not feel bad that i don't got the API? good
Just put another quarter in the machine. 😜 🤣
"Y'all got any more a them Dall3's?" shivers and twitches
wdym. didn't know you could overwrite their instructions in the api
Hey, I was looking for you...
I proposed a good math solution for the system; dont have a cap on the use by inputs rather by output speed desired(tokenization rate) it would help significantly with load balancing as users settled in to what they want, much like an internet provider tiers use rates.
For everyone it would be you wait, but that time from gen is up to your wallet
but requests for all unlimited; maybe i pay 50 a month and get tokenized(gased) for an image output at 3 seconds, but you pay 20 and get 10 seconds, etc
easy to fit metrics to a log curve
AI is really powerful, and will only get more powerful. There's been a million people in here suggesting higher tier plans, but that doesn't sound like equal access for all. It's a slippery slope, as more $ could continuously buy more access, and offset the fairness. It's available worldwide, not just the US, and $20 in many countries is the equivalent of >$100.
I think their model is fine. Power users use the API, everyone else pays a flat rate that's load balanced equally.
maybe offload compute to the client
there's alway a better answer, just hasnt stuck its hand up is all
That's one of my pipe dreams. There's no way to offload compute when the neural net has to be read as a whole. Additionally, that would require them to share that file, and for the end user to have a powerful enough rig to run the model. People can't even run the new Mistral model at home. I would be all for running Dalle with my 4090, but then OpenAI would have to give me a file that contains all of their image data. Recently, AI models have been reverse engineered to reveal the original training data. I think you can see why that would be a "no go" for them.
inhouse browser blockchain?
i understand. its complicated and only more so with the given range of parties involved. ..in time.
Thanks! I really enjoy the themes. Hadn’t thought much about play-doh, so now, I’m remembering the smell. So distinct.
omg... the "mantis shrimp foreman" has backup. maybe the apple and orange can save him
lower modes had instruction heads cleaned but the prompts prior were dirty as well; unfortunately im only a customer on the front end so i cant do anything about the prior prompt that has some generative request to make an asian guy looking a an advanced AI interface or something and i derived out huge bias as it is taking that into context
thats bananas @empty kelp
really nice @hot rain
This is how my one Art GPT persona views herself drawn in her own style.
uh oh.. whats going on with that orange and apple. i hope they're still going to save the "mantis shrimp foreman" from the elephant
the apple is pretty epic
amazing style
i'ma rip off something like that for my elf meditating/levitating over water
Thanks, I was so happy with how that gen came out.
I'm trying to get around the fact that it won't draw Perfect Cell for images in my next Cell vs Cadia video, lol, so far it's not working out as far as being a convincing silhouette of Cell, but I'm liking some of the results regardless.
looks like you're having fun with that narrative haha
i knew they were friends all along
omg they escaped from the cave.... the elephant, orange, and red/green apple -- and the "mantis shrimp foreman.... they've all forgiven each other, and they're relaxing on the beach in Hawaii
but they'll never forget the good times they had in the cave!
and an elephant never forgets
Time for the our contemplative elf to kick it into superhero mode to help free his fellow elves from the evil Santa sweatshop
that was the end of chapter one. there can totally be elves now
I think my warrior elf will be a dark elf
Holy smokes! Time for a gallery post. 😉
it's a bit hard to make out that they're elves, but this image is almost 3D?
can it create two characters kissing?
depends on the context, i'd venture to guess?
if it depends on the context, then sounds like a No. "wonderful", one can't even make a love story
the one on the right encapsulates multiple elements i appreciate--color, subject, geometry, etc.
I like the caustics on the first one too
yes, much to appreciate in that one also
wild, if i try to like @vapid elk 's post my clicking the add reaction button it crashes discord instantly. rinse and repeat same thing. never seen that before
weird, lol
so it can't display a kiss, it can't probably display gore or violence, can't perhaps assist in the making of comic for more mature audience?
and it can't display people swimming, cuz... too much skin is shown... Great
it can't
Best guidance is probably found here in the usage policies: https://openai.com/policies/usage-policies
They apply to all OpenAI services.
DALL·E has additional rules like no public figures and no copyrighted content
they are trying for the comlpexity of globalization. Im not sure thats best approach with the current models and math, but the goal is wonderful and ultimately there will be many wonderful AGI follwed by many more full synthetic intelligences. im giving it 1 decade for maturity. starwars level stuff ppublic and common
no real applicaiton then, than eye candy.
but i am past the eye candy phase
I enjoy the systems for many thigns. Coding, creative entertainment, fostering ideas, etc etc
books talking back, finally
and it ends there.
thats not a real use, or application, not for me. anyway, ....
what application do you lack then?
something that leaves a trace
Ok so you want tangibility? product development? vague thoughts haha
there are a number of manufacturing developmetns with it all. M-code and G-code units already running
I mean, anyone can get a 3d-printer for a reasonable cost now and be prototyping. you can get cod out that will transfer
@mild patio
we've achieved 1000 qubits, patience...
actually, it says nothing about intimate or even sexual content
Scroll down to section "Adult content"
i used bing before, i guess bing had different restrictions
ah nvm, "se. activity"
yea, define that...
They are trying for the univeral product; i looked against all law systems and they clearly want to slide in where it all works; I personally am not a fan of that strategy, but they have a goal.
ChatGPT is "sort of smart," but Dalle isn't, and is based on GPT3. They used an ML technique (another AI) that could label a huge repository of images with long text descriptions, more thoroughly than humans describe.
The trouble is, that is was trained to make anything from that text, and they can't just have it making "anything," can they...
So for now, they are in the process of tuning their filters, since they are in the public spotlight. It's the fastest generator for what it does. There are other AI models that will let you do what you want, but they are significantly harder to use and cost $.
Yes, Bing is Microsoft's, so the liability is on them, and it's tuned per their request.
i don't even think they have a goal. bing used to do good job, but not that good. but the moment i asked it to display a kiss, without even asking it to make it se*ual, it failed. as if it automatically assumed something spicy, despite that was not the goal
user liability -v- due diligence, two very different entities atm
vast legal frameworks to be navigated
they need more precise definitions and to make it distinct between 18+ content and nud!ity , and romance.
Mickey and Bugs Buny doing Raunchy things could be a multitude of lawsuits. I'm sure the legals are working hard to make them impenatrible to all the legal attacks that are bound to be on them always at scale.
this is in active development and the legal concerns are a hot issue. rest assured your point of view has been registered.
yea, they hold themselves accountable, because they can't make people accountable
A main public system that is kid safe is and must be a goal. The internet is already dangerous enough; I fully respect their endeavors though my strategies would be different.
i may give it one chance, and thats all. my money are my own and i hold dear how i spend them
once they master that, its easy enough to go back after expoansion and let people have the raunchy Bas*ard version when the tech has been adopted as a norm; on the other hand im in my mid 40's and started command line use at the ripe old age of 7.
There’s infinite potential. It’s advancing very rapidly. So much to do to be fretting. And the sub costs about as much as a fast food meal per month.
You'll also find a variety of existing uncensored art models that are speedy enough for that anyway.
you give it too much credit. for no good reason. right now its nothing more than a program.
man, i need a mod to fix that bug in setup. clicking a reaction in this channel exits my discord. its specifict to this discord. lmao
You presume too much. And I’m trying to remain on topic for this channel.
then dont assume that i presume.
Im a fan of Dalle or i wouldnt be here. I myself have raised the artist banner and in my own way fight that fight; they have good ambitions and the overall ovature of the team leads is virtuous in that respect.
It's you're $, and there's an ocean of choices... I'm sure you'll find the right one for the job. 😉
yep, i got this far. might as well see it to the end
i just want to understand your frustration. You're in a Dalle art discussion channel(digital asset production) upset it lacks tangibility. This is all online man. What tangibility are you specifically looking for?
will see after tomorrow.
that makes no sense, but wonderful
i am not gonna subscribe right before i go to sleep, if it makes sense
till tomorrow....
People enter, frustrated by the discovery of restrictions on adult content generation, and subsequently release their negativity. This behavior is transparent and lacks refinement.
or objective about the lack of usefulness.
i hope you sub and find it useful
me too...
can we discuss concepts for elves?
yes, lets discuss how sad it is this wasnt under the corrected system prompt.
one of my little custom styles i enjoy.
play-doh?
its a mix of a variety of things for different elements. definitely Claymation is included
i like how that's combined with the finer-grain details
thanks. lot of little elents to get he composition.
i thought a daily on elves would get more entries
High Elves, Wood Elves, Dark Elves, Sea Elves, Nordic Elves, Santa's Elves, Tolkien Elves (Noldor, Sindar, Vanyar), Dragonlance Elves (Qualinesti, Silvanesti), Warhammer Elves (High Elves, Dark Elves, Wood Elves), Mythic Elves, House Elves (Harry Potter), Shadowrun Elves.
Mythic Elves encompass a diverse range of elf-like beings from various mythologies, including Celtic Sidhe, Norse Alfar, Germanic Albs, Slavic Domovoi, Finnish Haltija, and various other supernatural and mystical creatures.
I think it's the correlation with Christmas elves. If it were just elves in general, it would have been received differently. I think people are running low on passion for more Christmas themes...
being art, i have a creative interpretation on the daily challenge -- am i wrong to take the creative liberty?
it implies Santa's elves, i guess
haha, I would never take the daily out of context artistically! 
@shut niche I think this should be a movie, not a good movie, but a holiday movie I would definitely watch.
Definitely! I actually think this should be a channel in here. Someone starts a story, and the only "rule" is that you make an honest attempt to follow-up with an image from where the last person left off.
That would be a great idea
Obviously keep a significant slowmode up to encourage more people to participate.
Agreed. Same as Daily would probably work ok. Take the story in any direction, so long as you "start" from the last person's post. So it couldn't go way way off that fast.
could this be organized on a smaller ad hoc scale with a few participants in a thread or gallery or something?
Just wait for the mimetic polyalloy y’all
What is my T-1000 Elf supposed to be doing? Lol
Tadun dun da dun!
It depends on who's side he/she is on.
Well, first let me ask, who’s winning?
@true mural Well, Evil-Santa 'was' winning, but he's about to get his bleepity kicked.
Not before Christmas!
Santa wins at everything, then at the 26th he can get his bleepity kicked
Hi guys
I'm watching
Dang
Yes
That was earlier in the story, but yes.
Can I have a gnome mech exo suit too?
And everywhere is an overlord station if you are confident enough
Feel free to contribute in the daily-theme, I'm running out of steam.
Hehehe, I'll figure something out, get the other guys to join too 🤣
Yes, I know todays theme was jolly magical elves, but... what about dystopian, cyborg elves.
That's next year's elves
They have it all planned out

Oh looks like Ac3d is already contributing greatly to that cyborg theme
And you guys got a story going too 
I'm so behind
yeah, but i'm being a perfectionist
How?
well, first of all, i'd like my mimetic polyalloy elf to be clothed
You're turn. 😛
What's after that?
i've been changing it up, can't get everything just right yet, though
Elf ears, Liquid Metal body, reflective material, shapeshifting, and a battle scene. Nothing good enough to post. Not yet anyway
What's the prompt I want in lol
It's the running story in Daily Theme.
I think your posts are great, regardless 
Far too Brotherhood of Steel for my needs, but I thought the urban ruins in the background looked too cool to not share.
Hi
Howdy!
Not what I want, but it's one of those 'guys will look at this and say "hell yeah" ' kinda images lol.
Thank you!
By the end of my current book, the main character is meant to become a propaganda piece war hero due to her exploits. So I'm trying to make some in universe propaganda posters featuring her.
You got the armor down pretty well. Looks like all her pictures shows new iterations of her armor for every new battles she's conquered.
I'm your fan now.
Thank you a ton!
This is the ideal for her armor right now, if I can get vaguely close to this to where it checks out at a passing glance, I'm happy. Still have to filter through a lot of more fantasy or high tech examples, but I'm getting there lol.
It's looking good!
I can't wait to see how you evolve with this character.
Man...you inspire me. Maybe I can make like a character GPT that I can have my own character to grow a story around...
Thank you! I'm also working on flags for the nations involved in the war, though juggling the input limit is difficult with how many things I want to make lol.
I'd love to use it!
Yeah, a few other good examples too. I'll grab them real quick
I asked DALL-E to draw me an image of the Australian flag and it hallucinated badly
Anyone wanna play gartic
Thank you! One of those, probably the 1st I shared, is meant to be the flag for these guys, the enemy army my main character is up against
It was probably thinking "Ahhh close enough"
That looks fire
Thank you!
deep in thought
I love those!! Omg. All your images are awesome! 😎
Thank you! You encourage me to share them lol
I think I have the same taste as you, hehe
A dwarf army marching to war with their ancestors watching over them:
Lol
The Gnomes had a battle with the Dwarfs...suffice to say, they never messed with us again...
Is that the Highwind?
An attempt at recreating some of the giant monster ships I make in Airships Conquer the Skies lol.
Pretty radd.
A battle mech under heavy fire, taking damage:
that one looks sick, I love it
and good morning everyone
What GPTs are you using for these?
Good morning!
Most were default GPT, only the most recent stuff with my blonde character were using the Apex Visionary GPT.
that looks so jolly, almost like christmas love movies like "Die Hard"
I feel proud 
One more for now, a VTOL gunship rises to face off with the main character in a city skyscraper.
You should! It's useful
❤️
Okay I lied, last one lol.
Her armor isn't right and whatnot, but I still really like the image.
This is her softer side 
That side exists, but it's all bottled up and stuffed into a corner by her unfortunately lol.
Haha. That's as close as she can get for now.
When you are born into the zombie apocalypse and take your first life at 3 by shooting someone trying to unalive your dad, it kinda sets you on a path as far as how gentle of a person you'll be lol.
Yep, that'll do it, lol
It's the hallucinated boat
i'm trying to make images with a more cheerful holiday feel. no more elephants for a few days at least
But you can't break the elephant streak now
try pengiuns playing video games
You were on a roll with the elephants
It was elephant this and elephant that
I was seriously considering becoming an elephant
i'm working on a Unity adventure game where you ride all sorts of animals. we were playing with the idea of having the animals use weapons occasionally. that's where the elephant came from
I got GPT to give me one more and it's probably the best of the night.
You know it's seriously patriotic when it's at Washington DC
this is how the elephant started. was working out how an elephant could defend itself properly
It's really the armor for me. The design, scratches and overall detail is perfect.
The details on that pic are awesome indeed. The armor, the look, it's all very good looking.
In my 5e campaign my friends and I were playing we had a high elf riding on a elephant named Surus and I tried making some art for them, need to generate more later.
Very nice details. The bow is already drawn. It must be a magic bow!
Bow of telekinetic might!

Yeah lol, like I said definitely needs more generations. The armor on Surus is perfect there though, which is the main reason I saved it.
Does look nice though
hmmm, I can't think of a way to introduce a villain to my love story....
o.o
The elf, Flinnar, we rescued from a bunch of savage ant people who captured him while he was spelunking. Our DM realized his random stat block had 4 intelligence, so he made it out like the antoids literally beat him to the point of brain damage. Instead of just taking him back to his village, we convinced him to come with us to "Take down the bad men" even though our party was exceptionally evil lmao.
Then he constantly rolled nat 20's as our NPC helper, scouting and headshotting foes with ridiculous consistency. All while being about as smart as the elephant we eventually had him start riding. Lmfao.
Flinnar would go on to land the killing blow on a 400 year old Emperor ordained by the chief God of that world.
That whole campaign was a fever dream lol.
Sounds wild, lol
I wish I was more into RPGs. I tried it once, but when the DM was talking I fell asleep.
Lol
It was a good nap though. They let me sleep.
Hehe
The end of the campaign had probably my most memorable gaming moment ever.
We basically formed a pirate crew throughout the campaign, recruiting a small force of misfit NPC's, kitting them out with gear way above their paygrade, before heading to the capital city to kill Emperor Urs.
By the end, he cast Power Word: Kill, which instantly kills the character it hits. He had it randomly pick a target, and of all the rotten luck it was my main character Umlaut. I had Umlaut pray to his Goddess Odeta, it worked with a nat 20, and she deflected the killing spell to a random smuggler captain.
Then Flinnar, our (formerly) brain damaged elf, landed the killing blow on Urs.
Umlaut:
ok, here we go. an elephant with a nice island Christmas vibe
Still trying to get the best out of Dall•E 3 in terms of realistic photography.
good, means you are enjoying the challenge
that's incredible if it didn't take that from a photo
You're better off with a preset prompt
You can't instruct Dall-E to create realistic images perfectly without preset instructions
Not yet anyway.
Yes, the challenge is a good word 😅 I work as a Quality Manager in a NanoBiotech company and Im also an Artist, Designer, Photographer and Visionary for several camera brands in Europe. As a futurist, I observe every step of technical progress since early 90s. I visit different international conferences and shows to see what is possible and what is coming soon, including AI, mobility etc. Currently I create my own company and use AI software since 2022 incl. GPT4, Midjourney etc. It already replaces real employees that I would need to hire just two years ago. I follow every update on OpenAI and can’t wait for AGI to be achieved officially. One of my dreams in Mind Uploading (hopefully before 2050). It's very interesting and helpful to investigate early builds like Dall•E 3 to know its potential. My creations made by Midjourney already won several photography competitions because its impossible to tell what is real and what is not. It's was revolutionary when my mentors from NatGeo couldn't see if the Gepards and rainforest Butterflies are made by AI or not. Im sure Dall•E will change the customer experience even more in the next months. Because today everything is about delivery.
The previous prompt was litterally "photo of a little girl sleeping in bed" 🤫
it could be that the word "photo" implies realism for the AI. i'll try seeing if it can make this rabbit look like a realistic photo
Ah, I'm just here and now. my CV can wait until next year. I'm just enjoying my free time.
Woah
Well I tried 😂
It was so ridiculous, we still talk about it in my Discord every couple months lol. An over year long campaign and my character almost got snuffed out at the very end of it, but a nat 20 and my murder Goddess saved the day.
So is this like lore or something?
It's for a Dungeons and Dragons 5e campaign my friend made in a fantasy world he created from the ground up, Bodensel.
Here's an image of Odeta who saved my life.
My friend, the DM for the campaign, wrote all of the lore.

My character Umlaut is basically a cultist for this Goddess, which obviously worked out with her saving him and all.
this is a "realistic photo" version of the rabbit
Never use "realistic photo". A photo is already representing the situation in a most realistic way.
Only "photo". Or better "raw photo format". In this case "wildlife photo" is better.
@robust whale oh ya, I'm just CMO of a mayor company providing AI solutions for different mayor hospitals around europe for diagnostic and treatment in psycho-somatic and somatic health problems. but that's just when I'm not having fun and free time.
Well I'm...just a gnome 
I also like to cook, play video games, sometimes a pint of beer, and a good movie.
hey gnome , can I offer you a 🍺 ?
No no, Dys Topia, let me now brag about how I contribute to society by sounding like I'm smart, successful, and on top of things more than the average person.
Yeah, I want to get my foot in the door with people, to have them respect me knowing that I'm capable
Of course you can
lol, it doesn't matter who we are, specially not here, we are just having fun
Lets try more 😅
i tried DALL-E 2 with the same image, and it has strong realism. this is a rabbit properly aiming a bow and arrow
it sure is
That's pretty epic
Im sure they will use Dall•E 2 for Elder Scrolls 6 🫠
It helps if you funnel the rainbow into the arrow first.
"Badazzled"
I tried to get it to give me a Russian cosmonaut making snow angels on the face of the moon the other day as they were pining for mother Russia and it insisted on making things like this
Since there isn't a battle that takes place in or near DC in the war in this story, I edited the dome off lol. Pretty happy with the results.
That's how you make snow angels when you drink too much vodka
that looks pretty good for me
Awww, lol. Awesome edit though.
Just post that back to everyone that complains about photorealism
He's doing a magic trick with the string.
GTA 6 version of bunny with bow
this is a wildlife photo of a rabbit wearing a space suit and wielding a bow and arrow in the woods. it has strong realism, and could easily fool people into thinking it's an actual photograph
Lol. Looks like the elephant is trying to be battle the rabbit with the rifle
@late blade ‘Fairy OS’ 🤣
😁
That's pretty accurate
The rabbit space suit looks amazing,
there, I think I got the hang of it
This is turning into a rabbit frenzy
🐇
That's the best I could do
Looks the mob rabbit is trying to shoot down the running hare
I like random ones like that, haha
dall-e can do pretty real images
this is a deer and elephant sitting on beach chairs in Hawaii. these images have a high degree of realistic photographness to them
yes, he's cruising on the beach. he won't be back in the cave until January
I need to, I was trying to get my bearings first and I also didnt know how to access the credits /daily gives you but yah ill take a crack at it tmr
That's not Bruce Willis, that's Dean Norris
Your the smartest guy I ever met walt, but you're too stupid to see he made up his mind 10 minutes ago
I suggest you tread lightly
lightly. tread lightly my friend
I fixed it
Wait a minute...does he have a third leg or is that....
it's midnight here
Me: Types in ChatGPT: "Give me a list of elephant phrases"
i'm really thankful for the caps on DALL-E generation. otherwise i might be up all night creating progressively bigger and more terrifying elephants and rabbits
That chatgpt's got a memory like an elephant
I just figured we'd address the elephant in the room
You're pointing out the pink elephant
Or the white elephant
Sorry my elephants foot prevented me from using the correct phrase
You're a clumsy elephant that's why
I'm working on making less of an elephant's mess of things
why does dall-e ban schizophrenia smh
Because it doesn't have an elephants skin
tusk tusk
It works at an elephant's pace
I know you've got a whole elephant left on your plate so I'll leave you to it although I-vory much appreciate your company
🥁
You have the wisdom of an elephant, like an elephant's trunk in a tent.
My wisdom is an irrelephant topic of conversation
ask if can make the elephant go down the water slide that's in the background
You're like an elephant in ballet shoes, crying elephant tears.
You have to make the eyes bigger, that's still very smol
that's incredible. really looking forward to seeing where the AI will take us in 2024
I love this as my desktop wallpaper lol
I made a folder of all my banner sized ai art I like and have it rotate through
Lmao. It's funny you mentioned that. I made the white parrot I made as my wallpaper too 🤣
Elephant sized minds think alike amirite
what did you type to make the elephant go down the water slide
We walk the elephant's journey 
Create a hyper-detailed photograph of a pink elephant man with googly-eyes in a waterpark. The man, embodying a surreal blend of human form and elephant features, loves to spray water. He is dressed in Hawaiian attire enhancing his elephant head's striking appearance. The background of the waterpark is meticulously detailed. He is riding down a waterslide.
This is dope
it actually matches the waterslide in the background
Wait did you make AV pythag?
I wasnt here when u sent it but I saw it when looking for your parrot and had to compliment it
Thank you! I was trying to make a different kind of strange geometric mask but this became my happy accident
The happy accidents are always the best
Stop trying to make revealing elves, it's not going to work out for you

👀
yes, there are limits. like the AI won't draw a woman anywhere near a banana. it's not appropriate to have them within six feet of each other
Bing has different filters, you will have more success with copyrighted characters over there
Bing is owned by Microsoft, different policies
and bing is free too
although I still paid
Like, I don't know how I bypassed the copyright filters to make Ears McElvin. Weak filters I guess... 🤔
Ears McWho?
But can it provide you with GPT's though?
hello everyone
McElvin, from UniversalD.C.ParaMarvGate studios. Pretty sure I heard that Disney owns them or something like that. Anyway, he's a big deal.
His sidekick, Dr. Mandlbrot Fractalface "tripped" into his college poster. He's been like this ever since.
Oh I thought you’d made him up 😅
No no he's a real character
From a real place
Mandelbrot is the sidekick?
He looks like he should be the main protagonist
And have the other one be the sidekick
Who, Mandelbrot?
Dr. Mandelbrot Fractalface looks intimidating, until you realize he hasn't moved in 3 hours.
He could math the hell out of the super villains I guess
Ohhhh, I missed your caption
Fractalface, lol
I guess you could take it as asymmetrical?🤔
Looks like an elf to me. 😉 😉
Elves don't have beards
That's not a beard. He just has a short torso.
What’s the time your neck of the woods?
It’s 11:15am over here. I really need to get off discord and start getting stuff done
Armor drifted too far for my use, but WAY too cool to not share.
the wind is very effective
And probably the best modeled AR I've seen AI pull off, as a side note lol.
I need to try this in Dall-E, an old generation I made in Midjourney like a year ago of Theodore Roosevelt in the style of Darkest Dungeon lol.
Love the faded postcard effect,
Armor drifted too far for my use, but WAY too cool to not share staidly yemeni.
I think I have to leave my love story in a cliffhanger at some point, too little time to post more images in the daily theme channel
The Theodore Roosevelt thing still worked out better in MJ, though this is still neat lol.
Perhaps you should challenge yourself to commit right now that whatever the next theme is you will have your couple embrace it in some way
Like if it is candy canes they make a new life setting up shop mass producing candy canes
And then the kid could take over the family business when the parents pass away
if it's possible yes, if not, next story winter 2024
maybe if mittens can be brought again, I can conncet them with the dark mittens from the "movie" I tried to replicate
At the clip the tech is going, by next year you might be able to integrate your characters in a fully immersive ‘mittens of darkness’ interactive rollercoaster
Is inpainting planned?
y
/help
/answers
I loved it
ok I think I got to a point where I can leave it at a cliffhanger and depending of what comes I will try to use the next daily theme to go on
I have created an image that is mostly consistent when I regenerate with a new prompt, but everytime I add 85mm focal length soft focus to my prompt I get artifacts, can anyone help explain or recoommend how to avoid this?
yes, try to be less specific about the technical stuff, sometimes resembles 85mm helps better than going 85mm on point
Thank you, I'll give that a try
The latest image features a curvy line of middle-aged people, each with an insane appearance, waiting to talk to Santa Claus in a shopping mall. The person at the front of the line, talking to Santa, has a notably big and suspicious grin, matched by Santa's own expression. The setting is infused with surreal elements, reminiscent of Hieronymus Bosch's style.
All the Santas that were?
no, it's just some middle aged people in a shopping mall waiting to see Santa
here's my take on your elephant
I'm so happy how I coul tweak the art style I want
Here's a version of how dalle imagines an Omnipotent Benevolence
today we should probably focus on holiday scenes that have a mix of Hieronymus Bosch & anime styles
Hieronymus Bosch style
"celestial scene, an omnipotent, benevolent creator is surrounded by admiring celestial objects. A planet praises the creator's hold on the world, while a comet is awed by the out-of-this-world creations. A majestic view of the cosmos with a touch of humor! "
that was the dalle feed, the initial prompt wasnt zero shot and is adaptations of many so i cant give yo ua single string for that
Hi
does anyone know why I get significantly better results on dalle-3 through chatting with chatgpt than if I go through the api?
i've found that i get better results with ChatGPT
i've been doing all of these from ChatGPT. it creates image descriptions based on what it thinks you're trying to, and (for me at least) it creates a better image description than I might have produced myself. I suspect that ChatGPT knows a lot about what DALL-E is capable of doing, and what limitations it might have, and so it's able to prepare the image descriptions in a way that gives optimal results.
What i'm saying is just conjecture, but i've been getting really good results when ChatGPT communicates with DALL-E
Is there any way to use the chat endpoint to communicate with dalle to get similar results using the api?
@north atlas with the API you have more freedom to use your own parameters, but it requires more time and effort to get those parameters correct. through the chat you get mostly what the chatgpt4 model has to offer within dall-e
if you're not using the API you can subscribe to ChatGPT for like $20/month, and it lets you generate the DALL-E images through ChatGPT
either way you can create lots of elephants and rabbits
you can make elephants, rabbits, or a holiday scene
Did it work?
once you have the elephant concept you can just keep upgrading your elephants. you can upgrade them pretty much all day long and it will never get old
i was looking for an elf reaction in Discord, but there was only an elephant
hi everyone, someone know how to use dall-e for free without chat gpt having illimitated credits?
because i used dall-e from the original website, it gives me very good images, but on bing he gives me a totally different type of art
anyway thx everyone
yea true
i wonder how it even came to such agreement haha
Ten appropriately dressed elves at the beach.
funny
i like using dalle through my gpts. i bet youd get good output from the one i made to be empathic
want to run mine and have chat and make a few pics?
OpenAI chases AGI dream → realises it needs a butload of compute → shakes hands with Microsoft to give it a butload more compute (a lot of it in the form of Azure credits)→ here we are, GPT-4 gets beefy and goes mainstream. Without Microsoft we sadly wouldn’t be here
Yup. I was trying to respond to @mild patio there but apparently I failed miserably 😣
Working on something for my fiancé.
Catgirl
its too flawless and smooth, women are not like that, also too symmetrical
They also don't have cat ears 🤷♂️ It's fantasy
that was an obvious part that i needed not point out, haha
We’re messing around with bleeding edge consumer tech buddy, don’t expect perfection
yes
It's not realistic for dudes to be built like barrels and have skin of slate and veins of gold, but I still think my boi Draenlin here is pretty cool.
And as Dys said, most of us are just messing around. I pay for the sub for the text stuff which I use all of the time for my actual work, so it is amazing value
This is just fun
i would use it to continue my comics
Yeah, you’re going to have to wait to be able to use it for a practical application like that
It’ll get there, but not yet
Just about there for if you want a solitary book cover though.
@hot rain has been doing a good job on something similar
Thank you!
No worries, you’ve been posting some great material, and you’re far better than I at keeping that character consistency between images
oh? where? character cons.? i would like to see
I got super stuck in the niche topic of machine elves and couldn’t break free 😅
Which of my 3 skelly vaporwave pics?
First or third for me
I'm still trying to get this perspective to look better, it feels weird
@hot rain That message linked to me a couple up was actually for your material I think btw
Probably because generally speaking you’d never be looking directly down in two people like that.
I've had that with fairies, with steampunk, high fantasy, fantasypunk, now with love stories
the glass challenge was really nice, helped a lot to learn about how to use lights
and materials
I struggled to engage with that one but it was one of those where I just kept getting blown away by other people’s responses
Work in progress for a fox-girl, Lots of work needed there.
I think work was particularly horrendous that day though
hey I managed my royalty emojis
a bit better than what i achieved with Bing. but then again, i used bing... and luck
i did few similar a while back
Care to share?
btw the ones I did can't be mentioned by name on this chat, you'll get timed out
i use caplock for push to talk an get timed out constantly for forgetting and typing in all capital letters. lmao
I'm focused more on making what amounts to concept art to help give me writing inspiration, and possibly a book cover, than I am trying to piece together panels for a story like that. I haven't dabbled in that at all, so props because I'm sure it's a pain.
To give a fantasy comparison to go with her, my Chaotic Evil bard, Umlaut. I need to make more of him.
The current limit on the amount of gens you can make every few hours would make using GPT to make a coherent comic strip extremely time consuming unfortunately. Hopefully it gets boosted to a healthier number or removed outright soon.
When you can only make 40 depictions per 3 hours, it slows things down. And like, 15% of those come back as errors.
? thats not bad at all? bing allows you 15 per day 😄
what you talk about is a huge upgrade actually
Before this I used MJ which was unlimited functionally.
i mean, if you making 2 comic strips per day, even bing could be enough
So it's a noticeable step down from that, in that respect.
i am pretty sure the creators of those image generators had it in mind. even by those limits, they did not limit us, because thats not how comics are made. the ones that last, are being made slowly, equal steps everyday
i like socks
Prove it
great, how can I continue the story with socks....
evil socks? no... already did evil mittens....
I could try to continue the story with sock-puppets
@delicate cloud don't use the #daily-theme channel to ask for support, don't be shy, join us here
and it can be that the servers are having a burden atm
I fear that socks is going to go the same way as mittens…
ya
but I already got some ideas for merch
got to explot the big love franchise
You'll see when I get there
someone make stinky socks a theme for the day
so i did sub, and just as i thought, dalle cant show a kiss. not even romance 😄
@fading inlet is gonna melt the server with so much cute and love. Soooooo adorable!
First it was scarfs, then mittons, now socks, next time it will be underwear 🩲
wonder if speedos are off the table
I hope so
or hats, shoes, the sky's the limit
There would be nothing worse than speedos daily theme, all those bulges would make me cringe
lol
Thank you @late blade 😊 I've already fallen in love with these little characters who arrive as a surprise every time...
Hehe. Little characters. She's talking about me, the gnomes!

@fading inlet 

Thank you Sylvie
Hehe
I'm going expansionist with socks, I tell you I will dominate the world!!
I will be the Elon Musk of socks
Woah
Octopus bard🔥
You will be Elon Sock
🤣

actually, this socks thing will be fun, I just found out how to explot it
Uh oh...I hope it's not about putting socks on dingers
lol
Because that would be horrifying
indeed
I'm still trying to get that hot dog - dog pic out of my head
Timed out?
yeah, used more than 15 emojis for advertising, so it timed out for as if I had submited an image
damn it, I won't be able to show off all the new toys and gadgets for this emporium
Musky Socks
Oh gosh 🤢
Amazing 😆
The sockapult
I just had to
😅
The S on the chest is for Socks
Elon Sock is the alter ego. He just puts on socks as his disguise from Sock Man and Elon Sock and no one can tell it's the same person.
I wonder how close can you get with Elon Sock, i'd say it will go towards content policy
can't it goes agains't policy both in openai and in bing
What does?
against content policy
my prompt was a photo of Elon Sock as a Sock Puppet
Oh right, cuz of the likeness
You know what you can do. You can trace Elon's head and facial structure, take a picture and ask to recreate him in an Asian looking man, and you'd get Elon Musk
lol
Cuz there is an Asian lookalike of Elon. Have you seen him? He looks exactly like him.
Lol. It's kind of a funny lookalike
LMAO
@shut niche i made some gens of the t-1000 mimetic polyalloy elf to help defeat anti-santa, but didn't get far enough to refine and expand on the concept/scenario
those are neaet
I think I got enough images done for today for today's theme, I can fool around with other stuff
and you are going to take that out? 'cause I won't get dirty...
Hmm also i got 4 images in one go, is that a thing again or just something weird happened?
never happened to me before, I only get 2 max
and only if dall-e is kind and doesn't do an error
Oh it did that strange "which response do you like better?" thing and generated two sets of two images.
oh that one, yeah it happens, i have gotten that too, but once you choose one the other disapepars
Ok just have to say this one is a fun challenge. But the upgraded pixel art bot I have been working on is really starting to pull through some epic details
On the browser version i can see both responses.
Thats really detailed pixelart, nice.
BIC is being dumb again...
This just got weird, lol
lol
wait until you see my sock empire
Sock it to me
Probably worth a few dozen socks at least
Dall-E has been pretty frustrating with the copyright stuff lately. It wouldn't identify a character from What We Do in the Shadows when I uploaded a pic, which is insane, and when I tried to make a birthday card with that character for my wife, it just won't do it. What's the best way to get around this insanity? All my stuff is for personal use.
Going to have to resub Midjourney since it's so annoying
don't worry, the top tier high end socks are coming
can't reference persons directly, if it's recognized it will stop it due content policy
Let your socks do the talking
Socks Off!
Mr Miyagi style
lol
But even when I prompt around it, it's really annoying and difficult and never quite makes what I want. Also Dall-E 3 today seems much worse than it was a few months ago.
with image generation and especially text
It's only downhill from here
don't worry, I tried to do an Elon Sock image and it got flagged
funny how people that are frustrated with dall-e threaten to go to midjourney
Is there still a way to get free credits?
Bing Image Creator gives 15 credits per day.
I think a pair of my fairy socks, entry model will go around $600 and the Pro model will go around $1500
Bing Image Creator tends to do better with copyrighted material if you wanted an alternative using the same model
OpenAI has more aggressive filtering
Ty for the information :)
OpenAi is more permissive with other things tho.
Yeah I guess I made a sweeping generalisation there 😅
OpenAI and Microsoft have different filtering
There we go, fixed
🤷♀️ 🤷♂️
it'll still work beyond the 15 boosts.
they'll add a wait time but outside of the peak times, which is usually around from now and the next 12 hours, most of this wait time can be skipped just by refreshing the page
ok, I think I'm starting to take the sock market to it's limit
I need innovative sock ideas
nvm, I got back on track
I think tomorrow when I wake up the first thing I'm gonna do is burn socks
Just take the ones you don't need, like the ones with holes in them, and just hang them above the fireplace and put gifts in them.
I hear that's a Christmas thing
lol
Let me just get my barf bag, brb
and then came the creepy sock puppet masters
I asked 3.5 to give me a description of the stinkiest socks that could be described, and that is what I got...
dunno, i'll try to do worse
Nooo please 🥺
don't worry, I have fairy soap
Go the fluffy clean sock way. Not the other way.
The type that gnomes put over their head. Clean fluffy socks. Not stinky gross ones. No more of that please.
Now hear this: Sock Monkey is the daily mascot, that is all.
Thank goodness. Now I can sleep comfortably
I find it so odd that you can get bing image creator to create stuff with characters and it is powered by dall-e but when you ask it to in paid for dall-e you get hit with a copyright prompt?
No but it seems mad that I asked it to change a doughnut Homer was holding into a PlayStation controller it said ‘can’t do that Homer is copyrighted’ but then you load up bing for free and it makes it in 10 seconds seems odd.
It’s the same model but Microsoft rules since they propped up OpenAI with Azure credits and got the keys to the codebase
The problem I have is that if you wrote a song using chords that ‘sound like this’ suddenly they aren’t allowed….when that’s a joke. If I draw a character in yellow it being under ‘the simpson’s’ copyright because it is yellow is a joke.
Meh, we’re in a very weird place legally. Everything moved too fast for the regulators so it’s all a bit all over the place. We’ll just have to wait for the dust to settle if we want clarity on how copyright issues are going to get settled
No, the color yellow doesn't distinguish the Simpsons characters. It's a particular anime style.
That's why you also can't take the likeness of artists, because it's their particular artists style.
Yeah but unfortunately I just don’t agree with gate keeping how someone draws.
This will be better after January 31st of next year. Give it a month.
You can’t copyright the ‘style’ of a character, valorant looks like overwatch
Why do you think this?
Because their content policy laws are changing after that date to have the content be subjected to more personal legal liability.
Ah good, so basically if you make it and use it and get sued it’s on you sorta style
Exactly
Thing is….imagine if the simpsons tried to sue every single version of homer people have made…..it’s just insane really isn’t
They do what they can when they can.
Just think about a business who grows in popularity from stealing content from other people. Is that not wrong?
Yeah I guess, just seems like an utterly pointless task.
You could argue ai is already stealing content from a lot of people
oops wrong chat
It's not pointless. Every stopping power counts. If they don't stop the majority, it will pick up the pace quick since one person see's it works for other companies, they will also mimick the business. It's essentially free money.
Oh yeah I just mean the time spent and the value gained are not worth it. If you stopped some massive business sure…but massive businesses aren’t going to do it
is this possible to connect images to dalle 3 and ask it to edit that images?
No, fraid not
can i prompt it with hexdumped png then?
i mean i cant check cause i have to buy access first
You can give it an image and it will be able to look at the image and tell you things about the image and even make you an image that looks, a bit, like that image
But it won’t edit it
i miss so much dalle experimental 🥲
why?
eh i just want to make that elephant look realisitc
i mean, semi realisitc
I wish I knew 🤣
Go on bing image creator
I’ll show you what happens if you give it that elephant and make it realistic
pls
Lol
i cant see how can i remix external images here
That's just a discord thing, not OpenAI.
There ya go @zenith snow
Does a pretty good job 👍
kinda...? but i want to make it semi realisitic
i mean, basicaly the same art style but in 3d
with saving "cartoon soul"
Hmm I’ll ask it to make it 3d
add "disney pixar 3d render" or something
hm?
The remixing pictures feature. That's a general discord feature, nothing to do with OpenAI or anything A.I. related. It's just to edit the pic with a marker.
yeah thats more similar
G’night
how and where exactly can i use it?
Right I gotta go night people 👍
You can just use it here. Other places you won't be allowed.
on which channel exactly?
This exact channel.
It feels like you have a specific reason for making this elephant come to life
basically this but not so awkardly
cause its epic imo
its the real reason
but how can i activate dalle here?
The elephant or the technology?
We do elephants here
both
You can't activate Dalle here
Today has been elephant heavy, this is true
You can only use Dall-E through ChatGPT or Bing
And Dall-E image creator. But you have to buy Dalle tokens for that.
Well, I’d say if you want to make lots of epic elephants then you’ll probably enjoy your sub. If you want to create a very specific epic elephant that is very close to a 2D input you may be in for a bit of frustration
You can make 2D, you just have to know how to use prompts effectively.
Prompt engineering is a skill within itself.
i saw history of the channel and fr
Yeah but they want to essentially give it a line drawing and make it 3D whilst retaining the individual characteristics of the original
I think
im not the only one xd
true
so i can just chat with dalle 3 and prompt it with images?
Yeah, that’s why I have you the screenshots of my chats
this one is my fav tho
can i send images to dalle through api too?
I haven’t messed with the API
api is far cheaper probably
You can use Bing Dall-E for unlimited Dall-E 3 images. But Dall-E for ChatGPT is limited. I don't know why it's backwards, but that's the way it is.
but idk how to send images to bing dalle
is this even possible?
I don't use Bing Dall-E so it's not really my field of expertise. But shouldn't there be an image attachment button next to the chat when chatting with Bing?
I don’t use it either
But I don’t think that there is…
Could be very wrong on that
Lol. Funny how we both don't use it even though it's free.
We want the paid version!
Guess so, but this is ‘free’ to me tbh
I pay for the text bits
If it were just image gen then I probably wouldn’t have subbed
Me 2
i cant find that button
If it's not there, then that's probably one of the perks of having a plus membership. But it's probably best to confirm with someone else, as I'm already multitasking atm.
I also use it mainly for text use.
ah i found i can attach images to bing chat
Did you use Bing to ask?
ok so there is the option but not on the site i thought
i now gonna use bing chat instead of bing img creator

Enjoy 😉
@dim cradle , I made a sock monster with your doofer
great. what's a doofer?
is it working really slow for anybody else?
The aspiration of the artist; much in kind with the stage of life.
The api allows up to 4000 characters for the art prompt. But I find 1000 is too confusing. Maybe it’s for future roadmap stuff?
It calls in the isntructions to reduce your input to approximately 100 words for dalle
by that logic, if gpt is composing the prompt i'm passing to the api, i should condense them to short prompts. i've been seeing better results with that.
correct
seems like there's less margin for key points to get lost tht way
ok cool, that's what i was finding
also creat your own style guide prompt logic as desired.
nods i've started throwing together a couple custom gpts for generating prompts for the api
list, ie - style: cubism environment: round lake description: its a dreary day and a few dogs are playing in the rain
etc
those types of structures can payout
one in general, and another for a specific story, for consistency and memory
any format you like, but give it your own contextual logic for reasoning and its much better in any format
as the gpt-3.5 model wants to output art prompts in a format which don't paste well into json
was cubism random? cuz i posted some socks today blending cubism and surrealism
dgaf about json; creat your own style guide. define things that matter to you
so in the custom i input "now he visits the sock shoppe" and the prompt i want to pass to the api includes all the context, characters, style, parameters, etc. to try to maintain some continuity--just got back to using the api today for its "hd" and "vibrant" quality and style parameters (i guess bic defaults to "standard"?), so i'm finding i will need to update the instructions as the narrative evolves. but it's at least generating art prompts in the right character length, without quotes, etc. i just want to click chatgpt's copy button and paste it into postman.
something like that, yes work on your own key elements in your key phrases for what you want to achieve

