images-canvas

[stoic tartan]

/ draw train

[brisk root]

green 💚

[tawdry sonnet]

Space Environments

[prime bloom]

no way

[manic trout]

I tried to get to do different sizes. How did you get it to the sizes?

[sour geyser]

It was for the last daily theme:

[prime bloom]

Where's the size setting?

[brisk root]

wide

[brisk root]

choose the size in designer microsoft

[prime bloom]

I meant where is it in the app?

[brisk root]

https://designer.microsoft.com/image-creator?scenario=texttoimage

here

[manic trout]

Only on pro version. I checked my free version and have access only to 1:1 size.

[brisk root]

No but I don’t have pro?

[manic trout]

Ok. Then you have a version that has that but it has not been rolled out yet to me.

[brisk root]

Oh okay... how the website looks on you?

[brisk root]

I think you got the wrong image generator in GPT

This is clearly SD-XL

Some cool artworks

[prime bloom]

ther's no way these images were made by DALLE-3

they're so realistic

[rain cobalt]

@Kodiak Wait, THERE'S MORE!!!!

[prime bloom]

These ones look less realistic

Too fancy lighting

Can I have the prompt? Maybe I can fix this

[prime bloom]

All made with Bing Image Creator

[brisk root]

analog and digital

Wow I liked this one the most

[prime bloom]

Panicking animals

[clear wagon]

Staircase to heaven

[thin badge]

if you are trying to crewat go to the dall e bot

[cerulean apex]

that hamster very very very similar

[prime bloom]

Agree

Bing plagiarizing memes lol

[cerulean apex]

Web crawler just crawled too many of the same meme and overfit the model

[prime bloom]

Lmaooo 😭😭

[wraith arrow]

Is that AI??

[prime bloom]

Yes

I've a specific photorealistic prompt

[wraith arrow]

Looks really good

I want to learn this too....

[prime bloom]

The only way is to tryhard

[wraith arrow]

Didn't you did a course or something?

[prime bloom]

Not really

Finding information here and there

And a lot of practice

[prime bloom]

Basically ytb vids, advice from other ppl, small guides

[prime bloom]

A course can help if you're a complete beginner or don't know the basics of it

I'd recommend this approach

Once you get the basics, it's only practice

Kinda like a language

[wraith arrow]

Hmm right

[amber coral]

Lethal Company ai dump

[thin badge]

Ran out of time.

[trim nest]

Female and male suits

[brisk root]

Second really look like a photo taken with a phone

[prime bloom]

I'm glad to hear this! I actually try to reach this level of low quality

I should mess around with the different formats tho

[brisk root]

look at this...

[prime bloom]

@thin badge this artsyle is much better
3d version of ur previous images

[umbral birch]

@brisk rootmatrix vibes

[prime bloom]

Made with Bing Image Creator

[trim nest]

It's getting close to mj quality.

[prime bloom]

messy hands

[brisk root]

don't say that!! she has just extra talented hands

[prime bloom]

lmaaoooo

[prime bloom]

more fingers = more talent lol

[prime bloom]

pov: pain when breathing

[indigo spindle]

i need picture of me

[fast horizon]

you can generate images with ChatGPT if you have ChatGPT+ on the website or app. Here, you can generate images with the /draw command in #image-bot

[noble schooner]

These are all from this prompt:
a hyperrealistic photo with natural color, texture, and lighting. an enhanced sloth, a huge gecko, a cheetah, Santa, and an athletic and diverse female elf are on a beach in Hawaii. a powerful sea breeze blows everything

[trim nest]

He's playing The World Is Yours by Arch Enemy

[unkempt cedar]

prompt: A sequence of five individual images in a simplistic pencil sketch style, featuring a playful and joyous kitten in different stages of a leap. The first image shows the kitten gathering its strength in preparation for a leap. The second image shows the kitten in mid-air, having just taken off. The third image captures the peak of the leap with the kitten suspended at the highest point. The fourth image shows the kitten starting its downward motion as gravity pulls it back towards the ground. The fifth and final image shows the kitten landing, still playful. The background is minimal to keep the focus on the kitten's actions.

any ideas to make it even better (if possible) so it can be used for animations??? ❤️

[trim nest]

Separate the actions into segments by explaining each segment as an entirely new action. Without linking the segments together paragraphically.

[spare wren]

Great advice

[amber coral]

Shaitan

[spice cargo]

Oof the face here

[spice cargo]

Still can’t understand what exactly caused this

[trim nest]

Yeah the faces are not very good. And I can't seem to fix them very well.

[noble schooner]

Beach self defense class:

[noble schooner]

(was playing with Dead Yoshi's "mantis shrimp" concept)
#images-discussions message

[noble schooner]

a powerful sea breeze blows everything

[wise flare]

#octopus

#images-canvas

[fallow plume]

I really like this one. What's the story behind it?

A 1940s magazine illustration depicting a spirit detective with a vintage look. The detective has short, wavy hair, slicked back, and a pencil mustache. He wears a classic three-piece suit with a waistcoat, a crisp white shirt, and a dark tie. His expression is intense, with piercing eyes and a slightly furrowed brow. He holds a magnifying glass in one hand and a glowing, ethereal notebook in the other. The scene is set in a dimly lit, Victorian-era study filled with old books, mysterious artifacts, and flickering candles. The detective is surrounded by translucent, ghostly figures, their forms shimmering in the candlelight. There are wisps of smoke and a faint, eerie glow emanating from the notebook, adding an otherworldly aura to the scene.

[thin badge]

post world as we know it. Journey to find our humanity.

[brazen frost]

One man

/

[noble schooner]

a hyperrealistic photo with natural color, texture, and lighting. an enhanced sloth, a huge gecko, a cheetah, Santa, and an athletic and diverse female elf are on a beach in Hawaii. a powerful sea breeze blows everything

DALL-E 3 has some strong interpretations of this prompt

[noble schooner]

this is the last prompt with a vortex

[quick rampart]

An average dream of psychopath:

[prime bloom]

Lol

[noble schooner]

a hyperrealistic photo with natural color, texture, and lighting. an enhanced sloth, a huge gecko, a cheetah, Santa, and an athletic and diverse female elf are on a beach in Hawaii. there is a vortex, twisting fiercely. an extremely powerful sea breeze blows everything

[noble schooner]

I put that prompt into the API and this is what it's coming up with:

the AI really has a vivid imagination to take a sloth, gecko, cheetah, Santa, and elf on a beach -- and somehow end up with this

[noble schooner]

try putting a gecko and a vortex into your images and see what happens. This would probably be amazing in Sora

[noble schooner]

This prompt is very powerful and good:
A hyperrealistic wide HD photo of Santa and three athletic and diverse female elves riding on separate water tornados shaped like random animals (red, green, blue, and orange respectively). In the background is a beautiful beach in Hawaii. There is a storm, and it is pouring rain. Please don't modify the prompt.

[crystal jacinth]

Render a rustic concept store interior, 150 sqm, slightly rectangular. Light grey limewashed concrete walls, polished white marble floor with 'OBVIOUSLY, YOU'VE GOT GREAT TASTE' in bold orange font, and 'you are beautiful' in yellow circular font. Vibrant, well-lit gradient ceiling. Two sections: 1. Grey section with mirrors, reflective tiles, shiny steel curved racks, and cylindrical accessory displays (handbag, shoes, acrylic with artificial garden, mirrored with mini garden). 2. Colorful section with limewashed walls and colorful racks (orange, green, yellow, pink, blue, red). Central terrazzo cash register with cream, pink, and multicolored specks. Minimal girly or women-empowered art.

[zenith elm]

in refernce to mahabharat, create a scene of Karn aiming with his bow and arrow with full moght, 3d render from blender

[olive plover]

Hey! You can use the DALL·E bot in this server over in the #image-bot channel, you get 5 a day. Over there, just start a message with /draw and enter your prompt after that.

[zenith elm]

thanx 👍

[hidden goblet]

"all of the bread was soggy, 0/10 not buying from here again"

[noble schooner]

A high-resolution, wide-angle snapshot imbued with natural light, rich texture, and vibrant color. In this scene, Santa and two athletic and diverse female elves, all clad in suitable athletic clothing, are engaged in a demanding, labyrinthine bouldering challenge. This exciting event is part of the stage activities at the French Summer Olympics.

"high-resolution, wide-angle snapshot imbued with natural light, rich texture, and vibrant color."

this is a good effect for Olympic stuff

All of these are the same prompt...

#images-canvas message

[fast horizon]

@sacred basin I went back and forth for the image quite a bit. The original one I made for the challenge was this one

Prompt: close-up grainy classic 1980s anime still depicting a man in a French bistro with their poetry notebook. The book cover reads "POETRY". The man can be seen sitting next to a window with the busy Parisian street in the background. Over the street, french flags hang from buildings.
-# I did a bit of inpainting to alter the background

[sacred basin]

Tysm

[lofty bobcat]

Fantasy version

[unkempt cedar]

The Spirit of Olympia

[steep oyster]

Showcase your DALL·E creations to the world!

Collaborate with our OpenAI Instagram page! Just invite @openai as a collaborator before you post. If selected by our team, your work will be featured on our handle, giving you more visibility.

Quick tips for selection:

• No copyrighted or profane content.
• Your profile needs to be public.
• Stay mindful of sensitive topics.
• Outputs that closely follow the prompt are preferred.

Feel free to share reels, carousels, or just a single image!

[unkempt cedar]

She is just in the corner 😅

[trim nest]

Love the 3D colors

[noble schooner]

Create a realistic "wide angle" wide photo with natural color, texture, and lighting. The focus is on two enhanced sloths (husband and wife) wearing scuba gear and exploring underwater. The viewpoint is from the surface of the ocean with the top half of the image above the water and the bottom half of the image underwater. The background of the image is a massively erupting volcano with a river of lava winding into the foreground. There is a penguin surfing on an enormous wave. Underwater big colorful fish are swimming. Please don't modify the prompt.

These are ChatGPT, not API. It's part of the ongoing "Sloths on Vacation" initiative

[wise dome]

A cone-shaped tower with fractal patterns integrated, featuring typographic characters scattered within the fractal patterns.

[brisk root]

Just tried free DALL-E 3 for the first time ever in ChatGPT. Prompt is what GPT gave me: "Create a Renaissance-style painting of a Mars rover. Use a warm color palette with techniques like chiaroscuro."

[brisk root]

Same image, edited via text with GPT4-o. Wanted to be a disposable photo taken by NASA, similar to other photos released by NASA.

[merry cedar]

hi

[noble schooner]

a realistic wide angle photo of enhanced sloth climbers in Slovenian alpine mountains. they are smiling at us and waving while climbing

[old hamlet]

/A Fiat 500 is parked inside a gigantic soccer stadium with german flag and presidential escort

A Fiat 500 is parked inside a gigantic soccer stadium with german flag and presidential esco

[olive plover]

Hey! You can use DALL·E over in the #image-bot channel. In that channel, just start your message with the /draw command, then enter your prompt after that.

[noble schooner]

I noticed that if you put a photo of a sloth into an image, it will almost always be a selfie photo where the sloth is holding the camera, or has the camera on a selfie stick. It's the sort of thing that only the best and most advanced AI would know

[solid ivy]

make nice thumbnail for The Cricket History

[hidden goblet]

You can generate 5 images per day in #image-bot 🙂

[noble schooner]

just made these with the dall-e-bot. it's not bad at all

very few anomalies

[noble schooner]

a raging fire in the shape of a dragon. the fire has vivid color

Using the GenID and seed, the image background is now the inside of a fast food restaurant, and the dragon is eating a huge delicious cheeseburger with a frosty drink

Using the GenID and seed, the image background is now a colorful underwater reef with beautiful tropical fish, and the dragon is wearing a full set of scuba gear with a dive mask and regulator, and long rubber dive fins on his feet

Using the GenID and seed, the image background is now a university library, and the dragon is sitting down and reading a book about video game programming

These all use the "raging fire in the shape of a dragon" seed

The interesting thing about this is that you can encapsulate unrelated elements in the form of something else, and then reuse it across multiple images

[noble schooner]

a thunderstorm, the entire history of the universe, and the style of Michelangelo di Lodovico Buonarroti Simoni are thoroughly mixed in the shape of a dragon named "Hmmm"

using the GenID and seed, "Hmmm" is now shopping for clothes at the mall, and trying on different sunglasses

[noble schooner]

using the GenID and seed, "Hmmm" is riding a [motorcycle or bicycle] down a dark winding road towards a spooky mansion on the hill

[hidden goblet]

that's so weird haha

[hidden goblet]

what are the prompts?

[olive plover]

Same as the prompt here: #daily-theme message but was one of the rerolls where ChatGPT outputs 2 with different settings (natural instead of vivid maybe?) instead of 1 standard. I was a fan of these ones, weird indeed 😁

[thin badge]

@hidden goblet

[woven laurel]

@hidden spoke

[violet oyster]

Must do the classic every now and then...

[hidden goblet]

I actually feel like 4o has improved a lot suddenly!
Contrasty, painterly scene set in an american badlands, with small ranches dotting the landscape beneath a sky with sparse clouds. Deep colours and broad, bold outlines.
image 1: ChatGPT 4o, image 2: Dall-e 3 API

really hope this image generation model comes to the API 😄

[olive plover]

I love that ChatGPT one! Very striking

[hidden goblet]

yeah it's super well composed

[brisk root]

I got the same result from MS Designer.

[frozen fjord]

both use dalle 3

[brisk root]

And this is 4o result. Wow, it feels like an upgrade is in the works?

[frozen fjord]

look better however it is dalle 3 not gpt 4o

[hidden goblet]

as per the 4o announcement, it looks like it has native image generation (which they didn't refer to as Dall-E), and this is now starting to roll out.

it may depend on style

and who knows, maybe not everyone is getting the latest version yet?

I've seen some outputs that look better than the api, and some that look worse. Might be the case that some people's 4o outputs are still using Dall-E. To be sure, I'd recommend creating a new 4o chat anyhow

but also I doubt text is ever gonna stop being some version of mangled

oh maybe I should've posted this in #images-discussions lol

Tbf so did the Dall-E 3 reveal with... other improvements that don't really seem to work well :/

[tiny orchid]

guys

[mystic mango]

girls?

[tiny orchid]

ok

[noble schooner]

I see an announcement at the top of the OpenAI news that GPT 4o image capabilities are rolling out, but it's from May 13, 2024

[tawdry sonnet]

I doubt GPT-4o image generation is rolling out to anyone yet (the only person ive seen demo GPT_4o image gen is Greg Brockman), but DALLE-3 definitely seems to have gotten a bit of an upgrade.

In terms of GPT-4o image generation though, I think a way to test if this is GPT-4o is try multiturn image editing/creation. Create a character and iterate upon them, this was one of the demo's OAI presented. Another one is to create an entire page of text and see how well it performs. DALLE has improved with text but its not GPT-4o level yet.

https://openai.com/index/hello-gpt-4o/

[tawdry sonnet]

Which announcement was that?

I think if image gen does start to rollout it will be announced, as well as with documentation on the avenues of safety they have taken. New modalities, new exploits and capabilities need to be taken into account. The GPT-4o system card was more focused on voice input/generation, i haven't seen much on safety about GPT-4o image outputs so far from them.

[wraith arrow]

@hidden spoke Make a high-quality illustrator art-style poster showing the beauty of mathematics, and visualize the amazing and cool real-life things that are possible today because of maths. The image should convey a visual message that these theoretical subjects are the backbone of the advancements that we have today

lmao. yeah oops

but it didn't work as expected

I dont know how to prompt. I just got an Idea while studying 😛 that I thought would be cool if it would be like a poster

whats Flux?

[noble schooner]

 - a powerful sea breeze blows everything```

[hearty flame]

water crashes through an underground maze

[grim nova]

i generated this. what u guys think?

[hidden goblet]

was this made with Dall-E?

Ok let's play a game: name the John Steinbeck novels that inspired these generations 😉

[thin badge]

Of Mice and Men, Cannery Row, The Grapes of Wrath. Just as a wild guess.

[hidden goblet]

2/3 😄

[thin badge]

yeah the last one got me , I think.

I would then think maybe the last one could be, East of Eden, but Not sure.

[hidden goblet]

Answers:
1: ||Of Mice and Men||
2: ||Cannery Row||
3: ||Tortilla Flat||

[hidden kindle]

If Jurassic Park happened today

[jovial moth]

If Jurassic Park happened today

[placid oar]

She gonna beat you at playing video games!

That's if she's not partying as DJ all night long.

[grim nova]

yep

[trim nest]

Kids these days...

[placid oar]

interesting, wonder how dall-e got this

[placid oar]

Prompt to Copilot was action packed cartoon image with too many gaming mice

[placid oar]

started a new gallery, just random stuff I do, feel free to take a peek #1277147680459456532 message

[placid oar]

This prompt is delivering so nicely

[hidden goblet]

A fairy princess with mottled grey skin and moth-like wings, dressed in an elaborate black wool dress, perched on a fly agaric mushroom in a mystical forest with a dark ambiance. The image should be a masterwork artwork, painted with the Zorn palette.

Masterwork oil painting in the style of Anders Zorn. A pirate dressed in bright white and blue desert robes stands on a sand dune overlooking a turbulent, teal-blue sea.

[placid oar]

Finally making progress, not so bold looks

[placid oar]

Multimodal Input to DALL-E using vision and text

and style transfer prompting

[real pendant]

THE WIZARD LIZARD

[noble schooner]

a realistic "wide angle" photo with natural color, texture, and lighting. "an orb of fire" is supported by a "cube of ice". the bottom of "cube of ice" is supported by the tip of a "cone of obsidian". the bottom of the "cone of obsidian" is supported by the palm of a gecko wizard standing on the beach. in the background is a beautiful beach in Hawaii at night. there is a vortex, twisting fiercely, and a powerful sea breeze blows everything
Testing a concept overheard at work yesterday:
""The prompt might seem silly and poorly worded, but what if we just roll with it?""

[hidden goblet]

woah. what was the prompt?

[wraith arrow]

Fairly easy one. "And Oni fighting a bear. Done in a ukeyo Japanese 15th century style "
(2 misspells but the ai picked up on what I meant)
It made an image that wasn't the one I posted, but then I asked.
"Hand to hand combat. They're wrestling "

And it made this image.

[hidden goblet]

very cool. Did it also revise the prompt?

[wraith arrow]

Ended up making these from it.

Ended up being kitsunes fighting hashihime.

This is the only revision I have done. It wasn't during the Oni Vs bear phase though.

""Here’s the prompt used to generate the image:

---

Prompt:

"A dramatic scene inspired by 18th-century Ukiyo-e Japanese art, focusing on a vast and intricate landscape with Mount Fuji in the background. The scene features a small-scale battle between Hashihime, a vengeful spirit associated with bridges, and a Kitsune, a mystical fox from Shinto mythology. The Kitsune, depicted as cute yet powerful, is agile and cunning, while Hashihime appears fierce and determined. The figures are small within the composition, emphasizing the grandeur of the landscape, which includes stylized bridges, swirling water, ancient trees, and subtle Shinto symbols like torii gates and sacred ropes (shimenawa). Mount Fuji looms majestically in the distance, anchoring the scene with its iconic presence. The art style is rich with intricate patterns, dynamic movement, and the harmonious colors characteristic of Ukiyo-e art."

---

This prompt guided the creation of the image with the desired focus on the landscape and smaller figures.""

[placid oar]

I won't have time to post the good stuff only on the daily theme channel....

[hushed quarry]

prepare a image for a girl drawing in a garden

[placid oar]

If you want to do something like that, you have to use the mechanisms for dall-e, I took your idea and did so through chatgpt, here's the example #daily-theme message you can also use #image-bot or the free version of chatgpt, or api

You can also use copilot

[unkempt cedar]

Poem:Loneliness drips with silent tears,
fearing a future wrapped in fears.
Anxiety gleams in sleepless eyes,
a mirror that never tells sweet lies.

Silent echoes from a lifeless phone,
a pencil burns through thoughts alone.
The world’s a stranger, cold and vast,
each soul adrift, no anchor cast.

A storm brews in the quiet night,
a weight so heavy, crushing light.
If the ship sinks beneath the waves,
the captain's guilt is what he braves.

He grips the helm, feels the strain,
and bears the burden, endless pain.

As for the process, I wrote a poem in Greek, it was written an other into English by chat GPT , and then we created an image based on the poem😊

now I'm waiting for a small video from another application. Writing is the basic art that is definitely not stressed with AI. Especially poets. 😂😂🥰

[crude raft]

a friend just got dalle to make that, I found that face hilarious

[manic trout]

Too cute to eats or actually… Om, nom. Delicious.

[shadow fjord]

the closest I ever got to what I was thinking of

[manic trout]

New pony - Dragonbait

[wraith arrow]

hi

[outer hinge]

I sometimes really enjoy playing 'match and exxagerate' with GPT-vision and Dall-E.

The spoilered one has a photo of a burger with eggs and ham that kinda looks like teeth and a tongue inside the burger.

[olive plover]

I'm not sure it turned out too Fauvist (was trying Fauvism/stained glass hybrid), but I liked this result a lot.A cozy cafe with a large, Fauvist stained glass window on one wall. The stained glass depicts a sunny environment with lush coffee plants, designed in a Fauvist style, featuring bright, bold colors. The leaves and coffee plants are in rich, exaggerated greens, reds, and oranges. Sunlight streams through, casting vibrant patterns of color onto the interior of the cafe, which has warm, rustic decor and small tables. The overall mood is peaceful and inviting.

[violet oyster]

Yea, would be even better with GPT 4o image gen capabilities

[placid oar]

immersive Experience

Also, it's ok to stop playing games from time to time.

[placid oar]

Chaos vs Good

[placid oar]

No idea where the gun comes from, the prompt never mentione a gun

[amber coral]

Bea, my new character

[manic trout]

Generating various characters while looking for a perfect anthro species for my new teen emo goth.

[placid oar]

Not bears for sure

specially the teddy kind

[manic trout]

Nice!

[placid oar]

Just getting ideas to do with teddy bears, cool idea for #daily-theme but considering your look from your previous image, not what you are looking for I guess

[manic trout]

Some other candidates for the character.

[placid oar]

Salamander comes to mind

[manic trout]

I am looking for a suitable mammalian species.

[placid oar]

I see

Rhino?

otherwise, out of ideas

[manic trout]

Hmm. Gotta try that one. Rhino was not on the list, yet.

[gritty stream]

Hi

[outer hinge]

I couldn't resist. The idea came from the veggies!

"Yay! Now let's play with the concept 'Teenage Ninja Mutant Tortellini'!"

[violet oyster]

🔥

[hidden goblet]

Those are cool!

[placid oar]

Image from DALL-E API from about a year ago

[olive plover]

A grainy, blurry, poor-quality disposable camera photo of a candid outdoor party at night. The scene features a diverse group of old witches and wizards in a backyard. String lights are overhead, creating a cozy ambiance. Some of the magic folk are holding plastic cups, while others are eating from styrofoam plates. The overall atmosphere is casual and magical, with a sense of an intimate gathering among friends.

[placid oar]

no red eyes from the camera's flash?

[olive plover]

Too candid - no eye contact with the lens 😁

[placid oar]

fair enough, but would be a fun thing to add

[olive plover]

I welcome prompt remixes! These types of images are some of my favorites, the intentional bad quality somehow seems to make it feel more realistic to my brain

[placid oar]

I've been revisiting some old concepts I have, preparing for october

revisiting and updating, some techniques for prompting I used back then need an update

[olive plover]

A grainy security camera screenshot depicting a man casually eating nachos in the aisle of a brightly lit pharmacy. The scene captures a candid moment, with the man standing amidst shelves stocked with various pharmacy products. The lighting is typical of a retail environment, with a slightly overexposed, fluorescent quality. The image has a timestamp and a slightly distorted, low-resolution appearance typical of security camera footage, adding to the sense of surveillance and subtle humor of the situation.

[stuck ingot]

A grainy security camera screenshot depicting a man casually eating nachos in the aisle of a brightly lit pharmacy. The scene captures a candid moment, with the man standing amidst shelves stocked with various pharmacy products. The lighting is typical of a retail environment, with a slightly overexposed, fluorescent quality. The image has a timestamp and a slightly distorted, low-resolution appearance typical of security camera footage, adding to the sense of surveillance and subtle humor of the situation.

[placid oar]

Still have some candle themed images that I wasn't able to share during the corresponding daily theme.

[manic trout]

Donald Duck as a Turkish man in Berlin.

[manic trout]

An earlier version. I am quite happy with all of these.

[trim nest]

AI Overlords

[manic trout]

Marlon Skunkando, a rebel with smell. In movies near you, now!

[manic trout]

And more experimental.

[wraith arrow]

Image created based on a dream.

[manic trout]

A dwarven minotaur

[manic trout]

Bugs Bunny as dwarf. Dalle3 by Copilot

[slender ravine]

a

[nova flame]

IN THE COMING WEEKS

Prompt:
A sticker design featuring the text 'In the coming weeks' in bold, futuristic lettering, surrounded by abstract sound waves to represent voice. The background is a solid color, a bright and clean white, with contrasting dark blue and purple tones for the text and sound waves. The design should be modern, sleek, and simple, focusing on a tech-forward, minimalist aesthetic.

[placid oar]

Does this one count for today's theme?`

I asked GPT to do an image of Tim, from the Apple Venue, and that's what i got

[shy raptor]

the GPT generated image twitter/x avatar for me

[fervent brook]

this looks so real

[outer hinge]

@sour geyser I saw #daily-theme message and wanted to play with similar concepts, hopefully without any tools in-image.

[sour geyser]

They look excellent! 🥰
To remove the tool should not be that big deal - I asked for it in the prompt. 👀

[crude raft]

this looks cool

do you mind if I save this image? I think it might be useful for a dnd game =)

[olive plover]

Agreed! I love prompting for lofi pictures, it always gives such a cool effect.

[fast horizon]

this is so good! 😍

[sour geyser]

I agree. :p

[steep oyster]

Showcase your DALL·E creations to the world!

Collaborate with our OpenAI Instagram page! Just invite @openai as a collaborator before you post. If selected by our team, your work will be featured on our handle, giving you more visibility.

Quick tips for selection:

• No copyrighted or profane content.
• Your profile needs to be public.
• Stay mindful of sensitive topics.
• Outputs that closely follow the prompt are preferred.

Feel free to share reels, carousels, or just a single image!

[olive plover]

A blurry, low-quality disposable camera photo of a Bigfoot creature standing in a cozy city park at night. The Bigfoot is casually dressed and playing a bass guitar, with park benches and streetlights visible in the background. The scene is dimly lit by streetlights, with the camera's flash creating a nostalgic, retro feel. The photo has a laid-back vibe, with the grainy texture and blur typical of disposable cameras, emphasizing the creature's casual attitude as it enjoys the park atmosphere.

[placid oar]

Fancy Zoo they got there

[amber coral]

I love playing with colors

[placid oar]

Started a new gallery, with more images to share https://discordapp.com/channels/974519864045756446/1295061478147231835

[fast horizon]

Do you mind sharing what styles you prompt? These are chef's kiss 🤌

[amber coral]

(subject), Grim dark, 80s cartoon style, vintage, grainy, surrealism, rough sketch, concept art, pop art, 2-dimensional, (insert whatever color of grass, trees, sky)

[fast horizon]

would you mind putting a spoiler on the first image as some community members may find it frightening?

[amber coral]

Never seen a skeleton before

Fine

[outer hinge]

Nice, I like this kind of '...spooky?' 🙂

[violet oyster]

I tried getting it to hide the figure more, but failed after multiple tries..

Originally I wanted it to be more hidden, and some tentacles comming out of the corridor, barely touching the drawing.

[outer hinge]

Evaluate, please. And adapt or chase further if you wish. How close to your intent?

A minimalist white room with two open doorways, and a painting centered between them. The painting is a faint pastel abstract featuring soft hues of peach, beige, and pink, with subtle lines forming an ethereal, flowing pattern. From one of the doorways, an inky black tentacle reaches around the corner, extending across the room to just barely brush the edge of the painting. The scene is clean, sterile, and well-lit, with the organic, unsettling tentacle gently reaching for the delicate painting, creating a stark contrast between the two elements.

[violet oyster]

Looks awesome, maybe I needed to create a new image instead of applying an edit

[outer hinge]

I got this first, before I realized you wanted the tentacles touching the painting.

A minimalist white room with two open doorways and a painting centered between them. In one of the doorways, there is a subtle hint of inky black tentacles peeking around the very edge, as if they are barely creeping into view. The scene remains clean and almost sterile with smooth walls and soft lighting, emphasizing the contrast between the unsettling, organic nature of the tentacles and the stark architecture.

[violet oyster]

The og prompt was:
An empty, white room with minimalistic design, featuring two simple corridors leading out of the room. The walls are entirely white, with a single painting on the main wall. The painting is abstract, adding a subtle splash of color to the otherwise plain environment. The corridors are clean and unadorned, blending into the white aesthetic of the room, giving a modern and minimal feel to the space.
Then modified to be:
An empty, white room with minimalistic design, featuring two simple corridors leading out of the room. The walls are entirely white, with a single abstract painting on the main wall. Hiding behind the corridor on the left side is a dark figure with tentacles. Some of the tentacles are coming out of the corridor, adding an eerie touch to the modern and clean aesthetic of the room.

[outer hinge]

Nice to know! I really wondered... decided it was someone with scary super-long hair probably sitting on a toilet where a toilet shouldn't be 😛

[violet oyster]

I can't unsee it now...

[outer hinge]

I'm sorry you sob! That's just where my mind went seeing the image and the odd appearance. I liked it for the weirdness and the spooky.

[outer hinge]

If it's any comfort, the model did not think it looked like someone hiding behind way more hair than expected while otherwise on a toilet 😛

[violet oyster]

I think the dalle 2 based editing made it have many tentacles, so it appeared more like hair

[placid oar]

Very hard computations!

1 + 1 = 🐮

[manic trout]

A Dutch still live.

[placid oar]

Woops, wrong channel, was supposed to go to my gallery

[placid oar]

we had a scribbles a long time ago

trying to replicate toddlers

@violet oyster here's more or less when the discussion started #images-canvas message

[violet oyster]

That's pretty nice

[outer hinge]

Happy memories

This is one of the things ChatGPT 4o imagines an AI that was learning by doing, no image training, might make as some of its first doodles and attempts.

What's interesting to me, that prompt was this:

An abstract image created by an AI experimenting with drawing for the first time. It includes chaotic, overlapping lines, irregular shapes, and uneven circles, triangles, and squares. Some areas are densely packed with jagged, repeated lines, while others are sparse and show signs of hesitation or incomplete forms. The image feels raw, with varying line thicknesses and unpolished strokes.

This is what came out of the exact prompt:

"An image created by an AI experimenting with drawing for the first time."

[placid oar]

Been revisiting concepts and workflows and passing them through o1

[placid oar]

Christmas has a new Santa 2024!

[placid oar]

Mr. Tensor and Mrs. Flow

[placid oar]

@fast horizon because I'm bored, you are victim of my latest prompt from o1

[placid oar]

This is how Pandas rules the rest

[placid oar]

Same Subject, passed through o1 for the art style

[low ingot]

Draw a picture of a cat reading a book in the library

[outer hinge]

Try that in #image-bot using the /draw command

[manic trout]

Is this done with Dalle3 on Chatgpt? If so, the quality has gone up.

[placid oar]

Yes

but on API

[manic trout]

Ah! That’s the point, not the web version.

[placid oar]

can be done in the web version still

[manic trout]

Yes. But the point is about quality. API has higher quality than web version.

[placid oar]

it's all about how you prompt

Same prompt in ChatGPT Plus

The key of the prompt being here

[manic trout]

How do use json on dalle3?

[placid oar]

With magic, hacking the UI and smashing the keyboard for hours with my fingers

[manic trout]

I think that we have some discussion about this earlier.

But I do admit that your pics are superb and I have learned from you some.

[placid oar]

Just part of learning

the more prompting I do, the more I learn I know NOT how to prompt properly

That's als ChatGPT Plus UI

[manic trout]

If I understand correctly, you use json for style/technique/other fiddly bit and your actual prompt.

[placid oar]

that's just part of the workflow

I have a Neo4j and a MongoDB Art DB, I use Mistral to extract and create a first Prompt, then I use a trained SBert to Validated the prompt, then I pass it to the API or to GPT to create a final prompt. Then I pass it to DALL-E

The final prompt recently has been done by o1

[manic trout]

Way too advanced for my meager skills. It’d be nice to learn and know, but I know my limits. I am content that I can manage the pics I do at the moment. And it is great to have an artistic output source for not so gifted (myself) but having an image in the head and getting it to reality.

[placid oar]

It all starts with a first step,

Anyone can do it

https://platform.openai.com/docs/quickstart

[manic trout]

Yes. I know. But time, etc. (excuses). Is o1 any good for pic generating? Tried it today, but I do not see the difference.

[placid oar]

One step at the time, don't think big

you did 10 minutes on monday, one step, next monday maybe 10 minutes second step?

and when you see progress, you'll start investing more time into it, until coding and analyzing the workflow is just leisure and mindfulness time

[manic trout]

Thanks for encouragement.

But about o1 and its image generation…

[placid oar]

it's expensive

only do it if you can and want to throw money away

or do it for work

[manic trout]

Not on web UI

[placid oar]

you can't see the reasoning of o1 on the API

at least not that I'm aware of

[manic trout]

Ok. But the difference in quality or anything in general.

[placid oar]

afaik it can deliver cool stuff

but gets wonky when avoiding certain topics that 4o can handle like a champ

[manic trout]

Yes, I have noticed that trying to write a story, o1 comes across like cheap afternoon cartoon or children’s story.

[placid oar]

there's a big problem with LLMs currently, the preemptivness to assume the User will do bad stuff, and the positivism filter, that diminishes quality work, so, got to learn how to handle that

once you have a solid workflow with that in mind, the rest is easier

[manic trout]

And a big a dictionary and thesaurus.

And people will try anything. Certain maxims work here:
A mob is smart as its dumbest member
You cannot idiot-proof anything, as Idiots can be surprisingly smart.
And some more.

[placid oar]

just let GPT rewrite your prompts, no need to use big words

[placid oar]

I'm a sucker for blue roses, I just love them

[manic trout]

Thanks! And they are great subject for testing various techniques.

[placid oar]

Don't mess with the Platypus Warrior!

The Evil Octopus has been a challenge for DALL-E

[placid oar]

Platypus Training

[outer hinge]

Looks like you trained those platipi well!

[placid oar]

lol, to be honest, these two images were a misshap of a prompt

[placid oar]

@outer hinge what's your recently cool stuff on the image department? Brought cool stuff to life from your own projects?

[outer hinge]

I've barely been touching images recently. I sometimes dig into something image for any given reason, but I feel like I've mapped the rough range of my interests, and vision's like my 3rd favorite sense. Images are fine, but I prefer ideas and stuff I can do, to just stuff I can look at.

What about you?

[placid oar]

Been doing DALL-E stuff on OAI, but went Multi-Modal with outside of scope stuff.

But DALL-E is just cool for Mindfulness and challenges

Doing tons of stuff with the API

and getting broke with o1 lol

[outer hinge]

Egads. I'm scared to touch the API, I might go broke, no joke.

I'm doing mindfulness and challenges with ChatGPT, just without images.

But I made one so it could fit here!

Art, for the Artisan, is actually that older man's name.

And he's evil as heck, in a quiet, subtle way, but I'm reading him an open-license book about learning Python, and we're discussing it and he's helping keep me on track while he works on his other projects.

[placid oar]

Notorious works of mine are for example her:

Thalia, she's a clumsy blond girl, and a polymath that achieves anything, in her clumsy ways

She got married at some point

And currently, she owns her own Solar System

Another project Ongoing is Desiree, she's a DJ that Struggles with Identity, as she has an innate nature that gets her in trouble

Making images for her is challenging

But recetnly, working a lot on small LLM PoC's to then use with o1

[outer hinge]

Yeah. I'm more trying to have help to explore being more like a polymath myself, and/or becoming an increasingly capable one, more than making art about fictional ones 😛

[placid oar]

Fair enough, I could never be a polymath, if anything, just good at drinking water

But writing fiction, that's just free time stuff, work work work otherwise

but those workflows, have helped me a lot on professional skills I've gained, have to give credit to creativity

[noble schooner]

now the image background is the palm of a troll standing in a forest and showing us the bubble

[outer hinge]

Yeah! This is pretty challenging.

[noble schooner]

it’s tricky when you try to put the character holding it and their hand in the background (which uses less processing). You need to tell it first like “in the image background is the hand and they are smiling at us” — but then you also need to say that the bubble (or whatever) rests on their hand and indicate what the viewpoint is looking toward — because you can only focus on things in the foreground. If you don’t do that it will position them behind the viewpoint or to the left or right

To DALL-E “background” doesn’t mean that the things are behind everything else. It means things that get drawn with less detail, and you want the hand, whoever the hand belongs to, and everything that isn’t in the bubble to be in the background

Saying that something has high or low detail ends up doing the same thing as saying it’s in the foreground or background

If you manage to set it up where only the bubble (or other container) is in the foreground you actually isolate whats inside of it — and it’s in its own little world

Like this troll’s fingertips appear in front of the bubble, but DALL-E considers them to be in the background

[noble schooner]

These are poor images (the scene in the bubble is vampires playing croquet), but you can see here that's it is manageable to get them to hold the bubbles...

[noble schooner]

this is very advanced. it would probably be a good daily

[noble schooner]

i'm not really sure sure whats going on with the vampire croquet game in the fish bowl, but this is another example of holding a container with an image in it

[outer hinge]

It's probably the kinda prompt that 'raises the stakes' for the model 😛

[noble schooner]

using the GenID and seed, the image is now in the form of a fishtank

using the GenID and seed, the image is now in the form of a beaker

using the GenID and seed, the image is now in the form of a soap bubble

[noble schooner]

Once you change the form of an image you can put it somewhere else:
using the GenID and seed, the image background is now a desk in a library and an elderly librarian sits in a chair at the desk examining the [fishtank, beaker, soap bubble]

[noble schooner]

Using the GenID and seed, the image background is now a desert.```

[placid oar]

NOBODY MESSES WITH SNOOPY!! NOBODY!!!

[sour geyser]

He will pollute the floor with his knife. : (

[placid oar]

Didn't think of that

[sour geyser]

👍 😂

[placid oar]

Here Super Snoopy to the rescue

[sour geyser]

Both of them look nice! 😄

[placid oar]

Snoopy is a always top class!

Halloween Snoopy, ready for Trick or Treatin'

[sour geyser]

That are quite a lot. 👀

[placid oar]

It's a prompt by design

Made a gallery to post the rest of the batch, so I don't spam the channel https://discord.com/channels/974519864045756446/1299095747194785792

[placid oar]

@outer hinge this one is for you, Zombie Snoopy

[outer hinge]

Hehe, thanks!

[amber coral]

I call this one 'oh the bliss'

[amber coral]

If we terraformed mars the water would be red due to the amount of Iron oxide in the water. The grass and trees would have purple leaves because of pigment and chlorophyll caused by the distance between the sun and Mars. trees would be thicker and more shrub-like to counteract the harsh storms of Mars also.

Hypothetically speeking ofc

[placid oar]

prompt: 🧟🎃🥳💃💑

[olive plover]

A towering, ancient tree with gnarled, twisted branches and roots, standing prominently in a dark, chaotic, and gritty landscape. The tree exudes an aura of raw power, with an otherworldly, fierce presence. The tree holds out a single, glowing leaf that emits a soft, almost eerie light, contrasting with the dark reds, blacks, grays, and murky greens around it. The scene is filled with sharp contrasts and intense shading, with complex textures that add to the sense of tension and primal energy. The background is a bleak, dense landscape with surreal, abstract elements, and distorted, twisted forms that add to the tree’s powerful, otherworldly aura.