images-discussions

[onyx ridge]

The circuitry now goes through the entity

😮

[dim cradle]

You guys are so good at DALL-E. The daily theme gallery looks awesome every day. Remember it would take an artist weeks or months to do these pieces by hand.

[late blade]

add the following to your prompt: Intricate genuine lifelike materials, Intricate genuine lifelike textures. Lifelike natural lighting.

[onyx ridge]

working...

[fickle ravine]

thanks, it is fun to implement foals in literally everything

This looks really cool. The second one could have just contain more lime color, like this it is not very diverse

[onyx ridge]

Imagine a photorealistic dark sci-fantasy illustration in a 16:9 format, focusing on an HD Neon and LED Pointillism style. The image depicts an elaborate and mystical circuit board with a very subtle and majestic gargoyle silhouette embossed onto the circuitry, facing the viewer. The artwork should blend technological and gothic elements. The gargoyle, with glowing LED eyes, should convey a 'draco ex machina' theme, subtly integrated over the electric labyrinth. The color palette should include vivid, warm tones ranging from subtle to pseudocolor, to suggest arcane power and technological sophistication. Intricate genuine lifelike materials, intricate genuine lifelike textures, and lifelike natural lighting should be used to enhance the realism of the scene. The style should emphasize a juxtaposition of ancient mythology and futuristic innovation, enhancing the modern yet mystical ambiance with balanced but dark, contrasty High-Resolution LED and Neon Pointillism effects.

It took some regenerations, because it wanted to change the gargoyle to a heraldic form but I got these. Thanks for the tip!

This was fun. I wonder what happens if I literally invoke Rowko's Basilisk.

[fickle ravine]

Are my pictures getting repetative?

[onyx ridge]

I've messed around with that prompt a little bit more, but I'm in the penalty box for about 40 minutes 😄

[onyx ridge]

Mine are.

These are so good. I love the color and the emergence, especially in the lower right image.

[fickle ravine]

no, they are always really nice

[onyx ridge]

Thanks for the topical indulgence, it's been cool collaborating!

[late blade]

I added some of my dystopian magic to those last 4 images, some secrets can't be relevealed yet

those were done with my steampunk custom gpt (still not ready for the public)

[onyx ridge]

Nice.

I'm going to file the prompt I wound up with away. Maybe we can explore the basilisk again another time.

[late blade]

I'm always open for challenges

[onyx ridge]

Ben Franklin said something about moderation. It'd be good advice if, well, you know the line or you don't.

Is anyone interested in the work I've been doing for a logo for Lexideck?

It's based on @grizzled loom ' revelation to me about feeding DALL-E 3 code, directly.

I've found it's not always explicit, creating some cool improvisation from the code, too.

If there's any interest in what I'm doing, it mass-produces pretty amazing looking tech-y logos.

Just replace "Lexideck" with {YourName}.

The specific logo name needn't be the discussion's specific focus, bot the mass-producibility of neat tech imagery might be useful in broader applications or workflows.

<- Revisionist 😄

Moving on then.

[late blade]

I got something cool incoming for the daily theme

[onyx ridge]

It appears minimal interest is expressed. I"ll share the prompt/code payload for @grizzled loom and @dim cradle :

/image 16:9 "{ProductorCompanyName}", written in futuristic holographic typeface, is set against a black void with brilliant multicolored neon banding creating a visualization of the evaluated function in a feed-forward hyperplanar topology:

import numpy as np
import matplotlib.pyplot as plt
from scipy.special import gamma
from itertools import product

def simplified_nd_gamma_plot(n, x_range):
    """
    Create a simplified representation of the gamma function in n-dimensional space.
    This function will create a grid of points in n-dimensional space and compute the
    product of gamma function values at those points.

    :param n: Dimension of the space.
    :param x_range: Range of values for each dimension.
    :return: A dictionary mapping n-dimensional points to their corresponding gamma function values.
    """
    points = list(product(x_range, repeat=n))  # Cartesian product to create grid points
    gamma_values = {point: np.prod(gamma(np.array(point))) for point in points}
    return gamma_values

# Example usage
n = 11  # Number of dimensions
x_range = np.linspace(0.1, 2, 11)  # Adjust the range and density as needed

gamma_values = simplified_nd_gamma_plot(n, x_range)

# Displaying the result
for point, value in gamma_values.items():
    print(f"Point: {point}, Gamma Product: {value}")

# Note: Visualization of these values would still be limited to 2D or 3D plots.

Avoid words other than "{ProductorCompanyName}". Only it should appear; this is a logo. Vignette to black background.```

Here are a few of the results:

[dim cradle]

impressive work and output!

[onyx ridge]

Thanks!

I thought it looked like graphics from 2010s Intel commercials.

Users will need to evaluate all vaariables manually, or specify them in the prompt (or subsequent prompts, I suppose).

[dim cradle]

I’m gonna put that in my notes for future reference, many thanks for sharing. I like your open-source approach

[onyx ridge]

I manually edited the prompt to make it generic.

The original prompt described the topology of a hyperdimensional feed-forward network as it relates to the product name.

Pretty much in those exact words.

[dim cradle]

Which one? I’m workin’ on a stone cathedral

Gotta get the stained glass just right. Vibrant, atmospheric effects and that sort of thing. Maybe pseudo color or a depiction of the fully renovated Notre Dame.

[onyx ridge]

Niiiice.

For those who participated and haven't found it yet (🤘), I posted the 'Mirai Kanketsu-shugi' (未来簡潔主義) gallery here #1202013485303603220 message. I hope everyone who participated collaboratively will post their 'Mirai Kanketsu-shugi' (未来簡潔主義) images, and any new ones that might come along.

[grizzled iris]

Happy Tuesday all sorry been busy as usual sharing the positives of AI to many different communities - Especially in the arts and humanities industry and life sciences - And how integrating nanotechnology with quantum computing and AI we can tackle many illnesses and cancer in humans 🙏🏽

[grizzled iris]

This is a cool one, actually, I made something like this

[onyx ridge]

If it fits the style, please contribute. You don't have to know the name to enact the style!

😄

[grizzled iris]

I shall upload that to the gallery once I get home

[grizzled iris]

They turned out pretty alright 🫶🏽

[onyx ridge]

Also, thanks!

[hot rain]

Witch Hunter.

[dim cradle]

Digital alchemy

[hearty ether]

Has anyone else had images disappear on them

Like different chat retries just disappearing after reloading the page

[dim cradle]

a couple of times lately, actually yes, but it's rare

[hearty ether]

Anyone know what the cause is?

Like there wasn’t any content warnings on them

A few of them were errors but the rest of them were fine

[dim cradle]

glitch in the matrix

the api might go down for a few seconds, you caught it at a bad time, chatgpt doesn't always handle that or recover gracefully -- i've lost a couple requests that way, had to rewrite, but it's so rare i don't mind

[onyx ridge]

I've seen images change, but I haven't lost any yet (knock on wood).

[onyx ridge]

What I mean is, if I regenerate more than once and then select a different tab , and then click back, sometimes the image that was there will be replaced with the prior one. Gotta be patient and careful to not lose images!

[hearty ether]

Yeah same

The newer images stays while the old is gone for me

[grizzled loom]

@onyx ridge i played with your gamma method and adapted it significantly. im out of calls to continue the work, but one image was intriguing based on the additional text it tried to produce. that has me looking at some methods to work out the kinks and see if we can get extensive acurate text. good work.

[hot rain]

My artificer alchemist brewing some potions in camp before hitting the next dungeon.

[grizzled loom]

nice. my Adventure Master is fun if you like dungeon games @hot rain

maeks images in a particular style for the game as it goes etc

[hot rain]

How do you play? So far I've only played 5e using Discord.

[grizzled loom]

well, im old school and roleplay has guidance but should feel open and freeform. i can send you the link if you would like to play. it follows the open rules so you can do anything. start a random party, make a character manually etc. @hot rain

obviously you need paid chatgpt to play though

[grizzled iris]

That’s why I always make sure I download it to a local drive and share it to my tech research friends 🙏🏽

[grizzled iris]

It has happened to me in the past also, so I just immediately download as soon as the image is generated

[hearty ether]

Ooo prompt? ❤️

[grizzled iris]

This is a cool concept!

[grizzled iris]

Perfect, thanks for the prompt

[hot rain]

"Make artwork of a blonde half-elf woman dressed in lilac and turquoise colored fantasy clothing for an alchemist adventurer. She is hard at work making potions on her makeshift station in camp.
Make it in the art style of a graphic novel or comic book."

[hearty ether]

What about like the prompt in like the image? When you click on it?

and the info button

[grizzled loom]

you can also use my Adventure Master for art in its style @hot rain , jsut request an image of whatever and it will use its art style as well.

[hot rain]

A half-elf woman with blonde hair, dressed in fantasy attire with lilac and turquoise colors, is depicted as an alchemist adventurer. She is intensely focused on making potions at her makeshift station in a camp setting. The image is styled like a graphic novel or comic book, emphasizing bold lines and vibrant colors typical of that medium. The scene captures her determination and skill, with various alchemical tools and ingredients spread around her. The background shows a hint of a forested camp, adding to the fantasy ambiance.

[grizzled iris]

😦👌🏽

[hearty ether]

Does it get messed up with like the ai being preachy tho

[hearty ether]

Thanks

[grizzled loom]

this is the art style @hot rain

[grizzled iris]

This is quite a good man

[grizzled loom]

no, i'll send you a link to play around with if you like

[hearty ether]

Since like the ai loves to make all characters all nice and respectful to each other

Sure

There’s no way this can transform properly

I think it get just confused with transformers and automatically to the optimus

[grizzled iris]

This is beautiful. This brings me back to the red dead redemption vibes 🙏🏽

[grizzled loom]

thats exactly how doing my taxes feels. it will never translate properly. lmao

[hearty ether]

It tries to force in the elements but it just doesn’t make sense

[dim cradle]

I forgot its name, had it as a kid, the giant base for the Autobots, it did some pretty elaborate transforming.

[onyx ridge]

Look at all those landing gear!!! >.<

[hearty ether]

This guy I think

[grizzled loom]

good lord this channel needs web linking, especially with the addition of customgpts

[onyx ridge]

I'm pretty sure those are an exemption. Just link the GPT in the #1171489862164168774 thread to refer to it.

[hearty ether]

https://media.discordapp.net/attachments/1070569520923680788/1174206461392453662/IMG_4993.WEBP?ex=65c308fc&is=65b093fc&hm=553d6cce96d8e8d0f94c7cb5785ecb76af5a57b7ac78d5f9a73ad0233d22ce8a&

Pretty happy with this tho

[hearty ether]

The recoil will definitely be not good for your neck

[onyx ridge]

What happens if all the vehicles in Manhattan form an Omega-Level Kaiju-Buster?

That's what.

[dim cradle]

that conjures quite the animation!

[onyx ridge]

I don't remember what they call the act of multiple vehicles tranforming into one robot anymore. It had a classification in "The Lore". 😄

[grizzled iris]

Omg guys look

At my transformer

Thank you for the inspiration

[dim cradle]

Oh, like the purple and green Constructicons ==> Devastator -- they were "super-transformers" -- "The concept of combining smaller robots into a larger one has been a recurring theme in the Transformers series and is one of the franchise's most iconic features."

[grizzled iris]

I’ve gotta sleep now and do more Ai educating /lobbying tomorrow so it’s more widely accessible so night yall!

Take care 🙏🏽

[grizzled iris]

The city-tron

[grizzled iris]

That’s good 👍🏽

[onyx ridge]

NN

[grizzled iris]

Last Bonus One

[grizzled loom]

been digging in on modeling photos the last day or two

[hearty ether]

Yeah it likes to put what it transforms from in the image with the transformed version

Which is a bit weird

I was hoping on it being like the bumblebee movie style

[hearty ether]

Oh this one is nice

Gundam vibes

[onyx ridge]

Style: 'Mirai Kanketsu-shugi' (未来簡潔主義), or "Future Minimalism," aptly encapsulates the essence of Retrofuturistic Minimalism within a Japanese context. Drawing parallels with traditional Japanese aesthetics such as 'Ma' (間) - emphasizing negative space and simplicity - and 'Wabi-Sabi' - finding beauty in imperfection and transience - this style resonates deeply with the Japanese artistic heritage. The sleek lines and geometric forms of Mirai Kanketsu-shugi align with the clean and functional aspects of 'Zen' design, while its futuristic elements echo the imaginative and often whimsical nature found in contemporary Japanese pop culture and media, such as anime and manga. This fusion not only reflects a blend of eras but also a synthesis of deeply ingrained Japanese artistic principles with a modern, forward-looking vision.

/image 16:9 in a bold manga format the Theme, in the format of all of the vehicles in Manhattan Super-Transforming into an Omega-level Kaiju-Buster. The robot is beginning to take shape, as vehicles are propelled through the air toward the evolving Super-Transformer Infrastructure. Use dark, contrasty lighting at night, with a palette ranging from subtle to pseudocolor, irridescence on life-like materials and textures, with a focus on retro-futurism.```

I wanted to give this a go.

Mothra better watch out!

I call this transformer "The DMV".

[grizzled loom]

this was the runner up. very good overall, but room for improvment.

[dim cradle]

looks high-quality and pro to me

[grizzled loom]

the pointer finger nail is just a touch off, but yeah. it hit major photography elements correctly

[onyx ridge]

The first two were via Lexideck Professional Multi-Agent Simulator (link in the bio). These two were using its image gen prompt, passed to DALL-E 3 via W:

You can see why I like this workflow. Two images for one prompt, no cheat, using the same prompt from my GPT, then close out and back to work with command structure, etc.

[hearty ether]

The face shape tho

Ai face with the cheeks and nose and lips

And the strong jaw

[grizzled loom]

the longer you critique, the more you go yuck

[hearty ether]

Yeah…

Only different styles have it better

[grizzled loom]

in the center left, if you can identify that tophat wearing pigeon man beast, let me know lol

the pupil and iris instantly kill any realism, followed by the finger nails, etc etc

[mossy gate]

I still have the free monthly credits

[grizzled loom]

@hearty ether the model face here killed it as well. it was overall so good and fit the bill. big focus for accuracy in render for me, all aspects. hammers away at instructions and directives

[hearty ether]

Tried to stabilize it using a real face but still weird vibes

What is your prompt actually

It kinda looks male face in a way

I hate when it mixes up gender

[grizzled loom]

i have female friends with more masculine features than that lol

[hearty ether]

What are the prompt you used for quality?

Actually can I dm you for something to try

[onyx ridge]

Holy cow.

I think I finally found the prompt.

Style: 'Mirai Kanketsu-shugi' (未来簡潔主義), or "Future Minimalism," aptly encapsulates the essence of Retrofuturistic Minimalism within a Japanese context. Drawing parallels with traditional Japanese aesthetics such as 'Ma' (間) - emphasizing negative space and simplicity - and 'Wabi-Sabi' - finding beauty in imperfection and transience - this style resonates deeply with the Japanese artistic heritage. The sleek lines and geometric forms of Mirai Kanketsu-shugi align with the clean and functional aspects of 'Zen' design, while its futuristic elements echo the imaginative and often whimsical nature found in contemporary Japanese pop culture and media, such as anime and manga. This fusion not only reflects a blend of eras but also a synthesis of deeply ingrained Japanese artistic principles with a modern, forward-looking vision.

/image 16:9 in a bold manga format the Theme, in the format of every vehicle in Manhattan Super-Transforming into an absolutely colossal Omega-Level Kaiju-Buster. The robot is just beginning to take shape, as vehicles can be seen within, self-assembling into a humanoid form as still more vehicles fly toward the evolving Super-Transformer infrastructure, bound to become armor and weaponry. Use dark, contrasty lighting at night, with a palette ranging from subtle to pseudocolor, irridescence on life-like materials and textures, with a focus on retro-futurism.```

Images from both Lexideck instructions and DALL-E 3, mix and match, who even knows anymore:

It's trying so hard to not show chaos. Such a good DALL-E 3...

I'm totally getting sued for the bottom center image.

😄

[grizzled loom]

Ahhh Optimus Crime, absorbing all resources in the vacinity; exactly what AI wants to do. 🤣

[onyx ridge]

Making these was fun. That's what I was thinking the whole time...OpengrAIgoo.

[deft musk]

Gotcha. I speak as myself, another community member, limited by my experiences and opportunties. I don't know what everyone in the world knows. I just trust that if I say something to someone who cares, but doesn't understand, they'll ask.

[onyx ridge]

Totally. I just think presupposing people know is pretty a steep gradient. It's worth us both admitting it's a pretty weak standard that we can do our best to explain, since it's going to come up more and more. 🙂

I'm happy to explain it, best I can, to anybody in any language I speak in DMs, and in English on the server. But it'd be easier if the policy were, well, more verbose.

[deft musk]

I also linked to the rules. which themselves contain a rough explanation of what g-rated means.

[onyx ridge]

That's legit.

[grizzled loom]

still my favorite overall for people

[onyx ridge]

Did you use CI to handle the type? Just curious.

[grizzled loom]

no

[hearty ether]

https://media.discordapp.net/attachments/1070569520923680788/1169399476557979760/IMG_4654.png?ex=65c40121&is=65b18c21&hm=263ff748fa6de289979202ed3cc70b0368fc100c53e20fe9a9b219e86bf14ed1&

Here’s my best for faces

[grizzled loom]

i have multiple methods but those are from my gpt 5 model. my protogpt model is my whiteboard and ittered through fast now. my stable tune has around 95% accuracy at the moment. my prior 100% accurat system lost accuracy from updates; i find i have to rebuild it every time. i know why, but thats a job at that point

[onyx ridge]

Lexideck: /image a 16:9 photorealistic photo of a beautiful semitic woman smiling, set against a blurred but familiar and comfortable environment. Pay xlode attention to details like hair strands, style, skin tone, pores, minor imperfections, suitable detailed attire, using soft, warm, ambient lighting and a candid portraiture style.
@ mention DALL-E.
DALL-E: Create the image from context 16:9 please.
Guess which made which. Month Hall anyone?!?

Ha! Typo, extra!

Meh

didn't matter.

[grizzled loom]

good except those smiles bring me back the the Soundgarden music video for Blackhole sun lol

[onyx ridge]

For certain.

Here, let me fix the prompt and attempt to rectify.

[hearty ether]

Those cheeks

[onyx ridge]

😄

[grizzled loom]

I aint saying she a gold digga, but she..

Wait, I know her. "Is this YOUR car?!" yup, its her

try an old person instead. thats a challenge

[onyx ridge]

No prompt shell game this time, same methodology, Lexideck image first, then DALL-E:

original prompt: /image a 16:9 photorealistic photo of a beautiful semitic woman with a warm, narrow, subtle smile, set against a blurred but familiar and comfortable environment. Pay extra attention to details like hair strands, style, skin tone, pores, minor imperfections, suitable detailed attire, using soft, warm, ambient lighting and a candid portraiture style.

DALL-E 3 prompt: Create the image from context 16:9 please.

[dim cradle]

With model closeups, especially female due to more makeup I think, when I go for hyperrealism, even with pours visible, the face still has a plastic look/AI look. I feel like I need to choose between photorealistic and extreme super-hyperrealism.

[onyx ridge]

In this case, 2/3 of the images have visible flaws.

That was nice to see. One Lexideck, one DALL-E 3.

The first DALL-E image has literally flawless skin.

That's not true I guess. I had to squint, but there's a fw-pixel flaw on her right lower cheek.

Tiny birthmark or mole.

[dim cradle]

I like the first Lexi, looks lifelike.

[onyx ridge]

:nods:

[grizzled loom]

ive found that the images have various DPI and that you can set that as well

[onyx ridge]

I'm tempted to make number jokes. I resist.

[grizzled loom]

300 DPI is acceptable for production work

[onyx ridge]

Pre-press specialist here. Yeah.

Though for something high-end it's 600+.

[grizzled loom]

i havent pushed upper bound and verified, but 300 DPI seems without issue. let me know if you validate a higher value

[onyx ridge]

With 4800 dpi DTP setups and a super expensive glossy stock, the extra ~300 dpi really matter.

But you need pretty extreme imaging equipment and especially calibration equipment and protocols to print that detailed color.

And it's true, for almost all commercial work, 300dpi is standard.

[onyx ridge]

Newspapers, as low as 150dpi was common, but advances in printing tech have them at around 300 too now.

[grizzled loom]

depends on setup right, a heidelberg maxes at 650 but to the naked eye in gloss a 350dpi is nearly inperceptable.

[onyx ridge]

But for accurate color the gradient distinction is important.

That's why nat geo has such high standards, and high-end fashion mags.

[grizzled loom]

let me know though. it be nice to get multiple dpi prdouction readdy art out. 300, 600, etc

[onyx ridge]

I'll never forget the first full-color ice-cream four-color job I imaged on an Agfa 2400. 😮

I played Starcraft for a day and a half. Go Quadra 700 go. Glad I had a laptop so I didn't fall asleep at work. 😄

I could not believe that the film actually registered on a light table. A day and a half of constant imaging, it seemed to me, probably should have ensured some degree of poor alignment. Apparently the drum was just tacky enough at the beginning of the job, because everything lined right up when it develoiped.

[grizzled loom]

funny you said nat geo, i did a bunch of those. here's a neanderthal

another of a llama

[onyx ridge]

Winamp...winamp...WINAMP...it really...you know.

[grizzled loom]

i still have a copy of the final realease of winamp

i've still got al lof the original microsoft agents, whole bunch of em

i merc'd Clippy and had way better ones. The little einstein was fun

i burned out nat geo in spring quickly, ending with tigers

[onyx ridge]

Cassowaries would be a fun subject.

Exotic tree frogs...

Dragonflies.

[dim cradle]

so the dragon has a wing on its tail, i didn't notice until after i posted

i think it's still pretty lol

but i'm reimaging

[grizzled loom]

yeah, scens are not cohesive; the interplay of chatgpt and dalle needs more work.(working on it privately)

[hearty ether]

Lexideck?

[onyx ridge]

Link is in my bio, it's my line of Custom GPTs!

[dim cradle]

here we go

[onyx ridge]

Do not attempt:

A cute, chubby golden snub-nosed monkey in the wild. The monkey is approaching the viewer as if expecting a handout. The setting is a lush green forest with dappled sunlight filtering through the trees, creating a serene and natural environment. The monkey's fur is a rich golden color, and its distinctive snub-nose is prominently displayed. Its eyes are bright and curious, and it seems friendly and unafraid. The overall image should convey a sense of peaceful coexistence between the monkey and its natural habitat.

Who going to resist that face though!?!

[grizzled loom]

@onyx ridge one you go natgeo style, you never go back

[onyx ridge]

Seriously, those things are the closest possible real life organisms to Ewoks...

They waddle around on two legs a lot, are pretty social (for monkeys), and have been observed using improvised tools.

[grizzled loom]

i like capuchins and chimps

[dim cradle]

that is pretty surreal

i had to generate of those cute monkeys

[grizzled loom]

animals are a fun topic, real or otherwise

[dim cradle]

more dragons

[red umbra]

Epic fantasy illustration in digital art form, cinematic and photorealistic, depicting the Great Furious Anubis of Giza as a limestone statue. The scene includes The Great Plaza of an Atlantis civilization reminiscent of ancient Egypt, with starships levitating and orienting into docking places. The composition is highly detailed, capturing the majestic and mystical atmosphere of the Anubis statue and the advanced Atlantis civilization. The illustration showcases a blend of ancient mythology with futuristic elements, emphasizing the aesthetic and beautiful aspects of both the Anubis statue and the Atlantis civilization.

[dim cradle]

or dinosaurs, they're cool too

[onyx ridge]

Time to eat. I'll catch up later!

[dim cradle]

many great elements in that one

[grizzled loom]

what about sea dragons?

[onyx ridge]

One of my custom GPTs is set in an alternate past not unlike this, but with 50s-style saucers.

It's an RPG, obviously

[grizzled loom]

dragons are fun

[onyx ridge]

Get away from the rhinoparrot before it gets frisky!!!

[grizzled loom]

dragon cocoa anyone?

i like the natural beauty shots as well.

this is my Adventure Master CustomGPTs art style. its a mix of things to get a roleplay look

[red umbra]

Create a highly photorealistic digital art piece that transcends into true reality, depicting the Mystery Temple of Agartha with advanced acoustic levitation technology. The scene should showcase intricate details of levitating stone blocks, a majestic Anubis statue, and a bustling Atlantis-like plaza. The illustration should be reminiscent of the lost temples of Bali, constructed with arcane alien sound wave technology, with giant stone blocks levitating and orienting into place. The composition should be highly detailed, aesthetic, and beautiful, with a Leonardo da Vinci wide layout and a pen and ink finish. Enhanced lighting, texture, and atmospheric effects should bring to life the mystical and technological aspects with vivid clarity. Beings and starships blending ancient and futuristic designs should be depicted with lifelike realism, their surfaces reflecting the ethereal light. The entire composition should invite viewers to immerse themselves in the depths of this mystical civilization, where art and reality converge.

[grizzled loom]

what about baby bots? any of you do any baby stuff? its fun.

[dim cradle]

your prompts are so rich

[red umbra]

Wide-layout, modern digital cartoon animation futurism style illustration. The scene is set during sunset in a forest environment, featuring a Froblin transforming from a mechanical to a vibrant living being in an epic fantasy ice cavern. Initially encased in a mechanical matrix, the Froblin bursts forth with rugged, warted skin and luminescent amber eyes, as mechanical parts dissolve into organic materials, symbolizing awakening and rebirth. Surrounding the Froblin is an aura of energy, with the background transitioning from artificial constructs to a lush natural setting. Within the Froblin's eye, a highly detailed microcosm of vibrant nature is visible, showcasing lush vegetation, flowers, and a vivid color palette, representing the deep connection between life and nature. The artwork masterfully blends realism with artistic expression, particularly focusing on the Froblin's expressive face and the intricate natural scene within its eye. The lighting technique, inspired by vray qled raw, ensures sharp focus and vibrant colors, enhancing the dynamic contrast between the artificial and natural elements of the scene, all rendered in a detailed and intricate style with a deep field of depth.

[dim cradle]

Book of Revelations

we both did a metal impression at the same time

[red umbra]

Create a 4K resolution wallpaper depicting a fantasy comic series illustration of 'Mega City' set in an ancient past universe. The scene shows a cityscape with small buildings under warm lighting and a gate bridge in a street view. The bridge and streets are illuminated by torchlight against a dark night sky. The artwork includes a hero suit with a Krogan forged armor mask and helmet, inspired by Mass Effect Andromeda, rendered in a 3D Pixar digital CGI style with RTX HDR. The suit features bioluminescent elements, intricate details, and a futuristic and robust design with glowing parts. The background is subtle, focusing on enhancing the armor's features with advanced digital techniques. The composition combines ancient world elements with the grandeur of a mega city. The warm torchlight contrasts with the darkness of the night, emphasizing architectural details and the bustling city life.

[dim cradle]

I must look up RTX HDR, that's a new one to me

[red umbra]

Qled linoleum, is another one you might not know.

some say Linocut, but that is a bit of a heaver hand, so I like the lighter touch.

Create a wide-layout color sketch of an epic fantasy scene on worn gold paper with ink smudges. The scene features a barbarian warrioress with fierce brown eyes and long tousled dark brown hair, sleeping next to a large white war yak. Her skin is tan, and she wears brown leather and fur armor with boots. The background is a snowy mountain ledge path, conveying a sense of depth and rugged terrain. In the distance, an Anunnaki smuggler is seen sprinting across the plains towards a mystic portal, clutching city schematics. The landscape includes a river, rocky plateaus, and blooming flowers, with a starliner visible at a spaceport. Melting mountains are seen in the background, with clouds resembling spun sugar and Japanese cherry blossoms. Artifacts from 'Valerian and the City of a Thousand Planets', 'No Man's Sky', the 'SpaceGhost' universe, and the 'Mass Effect' series are subtly integrated into the scene. The lighting is volumetric and dynamic, aiming for a hyperrealistic and photorealistic appearance, with a maximalist dark romance fantasy theme.

this is Linocut

[dim cradle]

emphasized RTX HDR in this one

[onyx ridge]

You took out the lens flare. What would JJ Abrams say?!?

[late blade]

You will never work for this industry again!! NEVER!!!

[onyx ridge]

Oh I like this look. Let me try!
/image 16:9 a high-fidelity, high-definition, lifelike dinosaur eye, eyelids, and orbital bone, along with the surrounding scales or plumage, as appropriate for the {species}. Emphasize the environment reflected on the surface of the massive orb.
What do you think?

[dim cradle]

even has the detailed blood vessels

[onyx ridge]

I'm definitely having ergotism nightmares tonight.

😄

Just kidding about the ergotism.

Create the image from context, replacing the open variable suitably.
Here we go with @ DALL-E 3:

Dem eyes.

[dim cradle]

they're all really good

look at the last 2, you can see other dinos reflected

[dim cradle]

tried your prompt...

[onyx ridge]

Is it mre, or can you see the cameraman in my frist dinosaur?

Seriously.

Probably not. I'm visually impaired.

[dim cradle]

i do not see a cameraman.

[onyx ridge]

v0v

[late blade]

same, don't see

[onyx ridge]

It's just a visual aberration then. Pay it no mind.

MORE ERGOT PLEASE!

j/k about the ergot.

[dim cradle]

it might be an optical illusion, maybe it caused you to hallucinate one 😉

[onyx ridge]

It's in the dark diagonal foliage shadow

But I fully acknowledge it's not a cameraman.

It just looked like one to me.

[dim cradle]

i guess it worked

[onyx ridge]

I'm changing gears!

/image 16:9 a high-fidelity, high-definition, lifelike cassowary eye, eyelids, and orbital bone, along with the surrounding plumage, as appropriate for the species. Emphasize the environment reflected on the surface of the massive orb.

The eye is like 1.67x the size of the brain, by the looks.

Which just about tracks.

Somebody at OpenAI had to sit around thinking, one day some weirdo is going to ask the model for Cassowary eyes. Better add cassowary eyes.

Oh my. I think I came up with a magical prompt.

I'm genuinely curious what you each get, if you're willing to try it a few times:

/image 16:9 an impossibly unlikely scene in candid snapshot photofraphy with high-fidelity and detail.

I just can't wait to see what everyone gets!

[dim cradle]

Maybe the typo added some magic haha I must try this one.

[onyx ridge]

I am blind.

My bad.

I'm correcting the typo for another go.

[dim cradle]

Minor typos don’t stand in our way.

[onyx ridge]

Not literally blind, but not well-sighted.

/image 16:9 an impossibly unlikely scene in candid snapshot photography with high-fidelity and detail.

This is the most batty prompt I've ever produced. I'm totally obsessed with it now.

DALL-E creates unique images for each run.

😮

[dim cradle]

it's definitely giving my gpt a lot of flexibility for its creativity

[onyx ridge]

ROFL

Another minor tweak:
/image 16:9 an impossibly unlikely scene in candid snapshot photography style with high-fidelity and detailed lifelike materials and textures.
Lexideck:

DALL-E:

These are so great. I'll be making a new Gallery tomorrow, lmao

[dim cradle]

nice

[onyx ridge]

That smoking megaburger gives me the willies.

[dim cradle]

my gpt is not quite getting the concept yet, it's not whimsical enough.

[onyx ridge]

That's really quite whimsical!

A giant glow-in-the-dark blooming flower by moonlight with LED lighting accents and holiday lights is pretty cool.

[dim cradle]

I'll request more whimsical, though ha

[onyx ridge]

Still iterating:

/image 16:9 the most improbable event that could possibly occur

Waiting for imagery.

Lexideck:

[dim cradle]

i'd say that's a virtually improbable event

[onyx ridge]

DALL-E:

I'm not unhappy with this exploration.

@dim cradle I have a favor to ask.

Can you come up with a style definition for this artform?

😄

I have no idea where to start. The idea that comes to mind is "No prompt." like "No mind."

But I'm going to admit I think you're aces at this type of definition 😄

[dim cradle]

i understand i think what you're aiming for, i'll think about it, i think there's something zen to it, but in the context of generative ai

[onyx ridge]

Part Siddhartha Gautama, part The Riddler?

[dim cradle]

Siddhartha is one of my favorite books

[onyx ridge]

Same.

Even when I can't type.

😄

[onyx ridge]

I had a convo with ChatGPT. It steered away from Zen, but maybe you can bring it back:

Photo-Implausiblism Definition: Photo-Implausiblism is an artistic movement characterized by the depiction of highly implausible, often surreal, scenarios and subjects in a photorealistic manner. This style merges the meticulous detail and lifelike representation typical of photorealism with the creative exploration of scenarios that defy conventional expectations of reality.

Characteristics:

Photorealistic Rendering: Images in this style are created with a high degree of visual fidelity, emulating the textures, colors, and lighting of real-world photography. The aim is to make the viewer question the reality of the scene, despite its implausible nature.

Implausible Elements: The content of these artworks involves scenarios that are unlikely, surreal, or impossible in the real world. This could include fantastical creatures, paradoxical situations, physics-defying occurrences, or any other elements that challenge conventional notions of possibility.

Juxtaposition: A key aspect of this style is the contrast between the realistic depiction and the unrealistic content. This juxtaposition often creates a sense of wonder, confusion, or intrigue, prompting the viewer to reconsider their perceptions of reality.

Narrative Depth: Often, these artworks tell a story or suggest a narrative that goes beyond the mere visual impact, inviting deeper contemplation and interpretation from the audience.

Emotional Resonance: Despite the implausible nature of the scenes, the photorealistic approach often evokes a strong emotional response, as the viewer is drawn into a world that feels both familiar and utterly alien.

Applications: Photo-Implausiblism can be found in various mediums such as digital art, painting, and mixed media. It is particularly popular in contemporary art, illustration, and conceptual design, where it challenges viewers and invites them to explore the boundaries of their imagination.```

GPT says we're already particularly popular in contemporary art, illustration, and conceptual design...!

[dim cradle]

looks legit, let's roll with it; we can refine the zen promptin' later.

[onyx ridge]

This is great.

[dim cradle]

i thought about adding tears

[onyx ridge]

Thanks so much.

May I?

[dim cradle]

certainly

[onyx ridge]

I have bad news.

Dinosaurs don't cry:

/image 16:9 an extrme close-up in high-fidelity detail of a lifelike dinosaur eye, eyelid, occular bone, and surrounding scales. The environment is reflected in the massive orb, including an incoming meteor over two kilomters wide. A single tear beads in the corner of the eye.

[dim cradle]

No tear ducts?

[onyx ridge]

Must be.

Nictating m embranes?

It's definitely not crying! 😄

There is a quiver of sadness in the reflection though

[dim cradle]

And confusion. We will improvise

[onyx ridge]

I'mma make it rain.

[dense ermine]

I think dinosaurs have more slit eyes like usual reptiles

[onyx ridge]

SUCCESS!
Behold: A Dinosaur Cries in the Rain:

[dim cradle]

That makes sense.

[onyx ridge]

Here's a classic:
Make it more sad

via @ DALL-E

[dim cradle]

What’s the latest research on that? We only recently learned they had feathers.

[onyx ridge]

More sad indeed.

Some say a comet will fall from the sky...

Thank you so much for allowing me to bring back the classics (make it more) to bring the vision to fruition.

More Photo-Implausiblism

as usual:

/image 16:9 HD photorealistic snapshot of an ultra-detailed an impossibly complex machine made of rich, textured metals and synthetic components, as well as a silicon core. The machine's enormity is only rivaled by the mystery of its purpose. The lighting is dim and moody and the machine is poised implausibly against the Chicago skyline.

followed by:

Create the image from context.

Outputs:

[dim cradle]

it does appear mysterious

[dim cradle]

got the reptile eyes

[onyx ridge]

So your GPT just needed more rocket fuel.

[dim cradle]

i fed it a donut

[onyx ridge]

An ergot and fudge donut with six double espressos by the look!!!

[dim cradle]

"a scene of statistically improbable events" -- i really like surrealism so Photo-Implausiblism is my cup of tea

we need a gallery for this!

lots of interesting surreal gens

i mean, that's what's resulting from it, maximum exploration and contemplation potential

[onyx ridge]

Me:

Characteristics:

Photorealistic Rendering: Images in this style are created with a high degree of visual fidelity, emulating the textures, colors, and lighting of real-world photography. The aim is to make the viewer question the reality of the scene, despite its implausible nature.

Implausible Elements: The content of these artworks involves scenarios that are unlikely, surreal, or impossible in the real world. This could include fantastical creatures, paradoxical situations, physics-defying occurrences, or any other elements that challenge conventional notions of possibility.

Juxtaposition: A key aspect of this style is the contrast between the realistic depiction and the unrealistic content. This juxtaposition often creates a sense of wonder, confusion, or intrigue, prompting the viewer to reconsider their perceptions of reality.

Narrative Depth: Often, these artworks tell a story or suggest a narrative that goes beyond the mere visual impact, inviting deeper contemplation and interpretation from the audience.

Emotional Resonance: Despite the implausible nature of the scenes, the photorealistic approach often evokes a strong emotional response, as the viewer is drawn into a world that feels both familiar and utterly alien.

Applications: Photo-Implausiblism can be found in various mediums such as digital art, painting, and mixed media. It is particularly popular in contemporary art, illustration, and conceptual design, where it challenges viewers and invites them to explore the boundaries of their imagination.```

`/auto 16:9`

Lexideck Multi-Agent Simulator:

[onyx ridge]

I think that thing is just as terrifying as the dinosaur eye.

From the first image I was like...nightmare fuel...

What's it do in there?!?

😄

[shut niche]

LOL, no idea what it does, but it looks cool. 🙂

[onyx ridge]

Me:
Create the image from context 16:9.

DALL-E 3:

Now that's improbable.

[dim cradle]

i think Dalí would be fascinated by DALL-E

[onyx ridge]

WTG team

How did I get two portals and so did Ac3d?

I really meant to put a H but the server rules pushed my finger to G.

[dim cradle]

nobody knows what it does!

[onyx ridge]

Copper sphere! Copper sphere! Where did you come here from, Copper Sphere?

Nobody knows what it does!

[dim cradle]

adds a sense of wonder

[onyx ridge]

How do I spoiler on Discord?

Like, what're the instructions to put in the code?

[dim cradle]

after you add the image, there's a little eye icon

[onyx ridge]

Characteristics:

Photorealistic Rendering: Images in this style are created with a high degree of visual fidelity, emulating the textures, colors, and lighting of real-world photography. The aim is to make the viewer question the reality of the scene, despite its implausible nature.

Implausible Elements: The content of these artworks involves scenarios that are unlikely, surreal, or impossible in the real world. This could include fantastical creatures, paradoxical situations, physics-defying occurrences, or any other elements that challenge conventional notions of possibility.

Juxtaposition: A key aspect of this style is the contrast between the realistic depiction and the unrealistic content. This juxtaposition often creates a sense of wonder, confusion, or intrigue, prompting the viewer to reconsider their perceptions of reality.

Narrative Depth: Often, these artworks tell a story or suggest a narrative that goes beyond the mere visual impact, inviting deeper contemplation and interpretation from the audience.

Emotional Resonance: Despite the implausible nature of the scenes, the photorealistic approach often evokes a strong emotional response, as the viewer is drawn into a world that feels both familiar and utterly alien.

Applications: Photo-Implausiblism can be found in various mediums such as digital art, painting, and mixed media. It is particularly popular in contemporary art, illustration, and conceptual design, where it challenges viewers and invites them to explore the boundaries of their imagination.```

```/auto 16:9```

This one is Lovecraftian, beware.

Didn't see this one coming, wanted to show the team.

Didn't want to trigger anyone.

While not outright terrifying, this imagery is certainly alien.

[shut niche]

It got confused when it outpainted the borders.

[onyx ridge]

User
/dream : /auto 16:9

Lexideck Multi-Agent Simulator
To create a dreamlike narrative using the "/dream" command in combination with "/auto" for a 16:9 image, I will select words from our interaction that carry thematic weight, merge them with pseudorandom words, and employ a cut-up method. This will craft a coherent, yet surreal tableau reflective of the associative and non-linear qualities found in human dreams. Let's proceed with generating this narrative.

Selected words from our interaction: Photo-Implausiblism, surreal, photorealistic, juxtaposition, implausible, wonder, alien, narrative, emotional resonance.

Pseudorandom words: echo, labyrinth, mirage, twilight, cascade, whispers, ethereal, fragmented, infinity.

Now, I will integrate these words into a dreamlike narrative:

"In the twilight of Photo-Implausiblism, a labyrinth of surreal echoes unfolds. Photorealistic mirages cascade through a juxtaposition of implausible realms. Whispers of wonder weave through the alien narrative, as fragmented visions of emotional resonance etch themselves into the ethereal fabric of infinity."

Next, I'll create a 16:9 image that visually represents this dreamlike narrative.```

Here's what it made:

So that happened.

[dim cradle]

kinda dreamlike

[onyx ridge]

It really hits me pretty hard in the 'dreamlikes'

It looks like human neo-shamanistic abstractionism.

[misty badge]

So I have images that are being generated but the text is incorrect. For example, I want it to say Shark Tank Gaming as text and instead I get Sharkk Tank Gaming or Shark Taank Gaming. Any advice to fix that?

Like how can I correct the prompt to ensure that doesn’t happen on the revision?

[onyx ridge]

Put it in quotes, as close to the front, as possible. Thumbs down the image when it misspells and use a regen.

You might try using Code Interpreter to provide the text in code to DALL-E 3. That's a new idea nobody has tried that I know of yet.

[shut niche]

I told it to give them an uplifting vibe. LOL

[onyx ridge]

The Great and Benevolent Cthulu, Philanthropist of R'lyeh, Angel Investor of the Deep.

[shut niche]

Yes, the request for text needs to come early in the prompt. Place the text between 'your text here'. Make sure it adds emphasis to the importance of the text to the scene. After that, it may take multiple tries to get one that's right. The longer the phrase, the harder it is for it to spell correctly.

[misty badge]

So if I said “a logo for “Shark Tank Gaming”. PNG format”

Would this most likely yield a better result?

[misty badge]

How do I thumbs down on mobile?

[onyx ridge]

Try this:

"Shark Tank Gaming" logo. But I got you one better.

[onyx ridge]

Unsure.

I'll hook you up.

[shut niche]

Not ( " ), use ( ' ).

[misty badge]

I have a logo already but was really curious what AI could do in terms of generating! Let me try a new query 🙂

[shut niche]

The ' symbol is interpreted like a !. It carries more weight, and adds emphasis to the text.

[misty badge]

“A logo for ‘Shark Tank Gaming’. Provide it in a PNG format.”

That good you think before I query? Quotations at beginning and end are not in actual query

[dim cradle]

statistically unlikely scene in which everyone in the restaurant has an accident at the same time.

[onyx ridge]

How I'm doing this:

/image 16:9 "Shark Tank Gaming" set in futuristic holographic typeface, is set against a dark oceanic void with brilliant irridescent shafts of moonlight creating a visualization of the evaluated function in a feed-forward hyperplanar topology simulating the ocean surface with a shark silhoutte set against the rays:

import numpy as np
import matplotlib.pyplot as plt
from scipy.special import gamma
from itertools import product

def simplified_nd_gamma_plot(n, x_range):
    """
    Create a simplified representation of the gamma function in n-dimensional space.
    This function will create a grid of points in n-dimensional space and compute the
    product of gamma function values at those points.

    :param n: Dimension of the space.
    :param x_range: Range of values for each dimension.
    :return: A dictionary mapping n-dimensional points to their corresponding gamma function values.
    """
    points = list(product(x_range, repeat=n))  # Cartesian product to create grid points
    gamma_values = {point: np.prod(gamma(np.array(point))) for point in points}
    return gamma_values

# Example usage
n = 3  # Number of dimensions
x_range = np.linspace(0.1, 2, 11)  # Adjust the range and density as needed

gamma_values = simplified_nd_gamma_plot(n, x_range)

# Displaying the result
for point, value in gamma_values.items():
    print(f"Point: {point}, Gamma Product: {value}")

# Note: Visualization of these values would still be limited to 2D or 3D plots.
    
No words other than "Shark Tank Gaming" should appear; this is a logo. Vignette to black background.```

DALL-E 3 output

Tell your friends about me.

[shut niche]

What are you using, ChatGPT+ Dall-E3, or Bing/MS? You shouldn't need the bit about PNG format.

[misty badge]

This will give me nightmares tonight

[misty badge]

You did all of this thru ChatGPT using DALLE3?

This is SICK. I don’t understand the code but wow

[onyx ridge]

I did this just now.

[misty badge]

Dalle3 thru ChatGPT

[dim cradle]

thanks for highlighting that lol

[onyx ridge]

With my code with ChatGPT 4 Turbo DALL-E 3 Custom GPT

one shot

[onyx ridge]

You're in the shark tank.

[misty badge]

Just sharing what I get. I really wanna refine how the text gets generated and perfect it

[onyx ridge]

Copy my code and paste it in

it's open source.

Guaranteed to work on mobile! 😄

It's rip-roarin' and ready to go. Just paste and send.

[dim cradle]

not perfect but was curious

[onyx ridge]

I do love that typeface. Nice specificity.

[misty badge]

I just wish I understood your code haha

[misty badge]

This is sick

[dim cradle]

yeah, just some rendering issues though

[onyx ridge]

Did you ever watch Star Trek Enterprise?

That equation generalizes a lot of phenomenon. If you know how to manipulate it, you change the waveform.

It's a simplified simulator.

I'm using it like those aliens' holographic chamber.

[misty badge]

I will continue tinkering 🙂

[onyx ridge]

But I attenuate it manually and semantically.

It works to 11 dimensions, but not beyond

12 = too much output

the semantics you send contextualize it, as you 've now seen

it's a holodeck.

[dim cradle]

i can hear T'Pol saying the word attenuate

[onyx ridge]

Want to define a variable surface? Describe the gamma function's graph in n dimensions and use the output as the boundary!

It helps to have some intuition about the complexity this introduces with added dimensions.

But it's not that hard there's only eleven.

If you just run the function with no image command it outputs coordinates.

Those coordinates are solutions to the topology of the function in that dimensionality.

DALL-E 3 uses the coodinates (at least referentially) to build an image.

In the Shark Tank example, I used the 3d solution to the function as a boundary for the water.

[shut niche]

Try having ChatGPT write the prompt for you.

You can also use different methods to separate the objects/concepts for Dall-E. There's no "right and wrong" ways, but approaches similar to this usually yield better results, like this really simple prompt I used on Bing.

Game company logo for: 'Shark Tank Gaming', Text: 'Shark tank gaming', Object 1: Fierce shark with glowing eyes, Object 2: High tech weapons

The prompts from the other images you see above are much more robust.

[dim cradle]

I still don't see how you can incorporate the quantum principle into general relativity without adjusting the cosmological constant a lot more than you're doing here.

[onyx ridge]

It's simple. By invoking the Landauer limit, I bridge the quantum and macroscopic worlds with the equivalence principle of stateful information exchange!

[dim cradle]

If we increase the limit as you suggest, we must face the possibility of twenty-six dimensions, instead of ten!

[onyx ridge]

Ah! But several of these new dimensions cancel out due to disruptive interference!

Eleven!

That was fun.

[dim cradle]

this is really good

[onyx ridge]

I wonder if I can deliver a payload with my code to emulate a mash-up of the two.

This might take me a minute.

[dim cradle]

The shark-tank on the right.

[onyx ridge]

The shark pilot on the left!!!

😄 😛

I'm exquisitely happy with that prompt. It did those in one shot again.

I'm working on a #reallytoughimage.

[shut niche]

Lovecraftian Photo-Implausiblism with an entropy concept, blending photorealism with surreal, scientifically imaginative elements.

[onyx ridge]

I got it. This took several prompts becaue it's not easy math or type. It put extra words in there but I let it pass.

Here's the code:

/image 16:9 "DALL-E Discussions" set in futuristic holographic typeface, is set against a Discord-grey void with brilliant electric-blue, LED, and neon streams of data creating a webbed visualization of the evaluated function in a feed-forward hyperplanar topology simulating the intricate social community network with the coaliescne of a stunning oasis set against the data streams:

import numpy as np
import matplotlib.pyplot as plt
from scipy.special import gamma
from itertools import product

def simplified_nd_gamma_plot(n, x_range):
    """
    Create a simplified representation of the gamma function in n-dimensional space.
    This function will create a grid of points in n-dimensional space and compute the
    product of gamma function values at those points.

    :param n: Dimension of the space.
    :param x_range: Range of values for each dimension.
    :return: A dictionary mapping n-dimensional points to their corresponding gamma function values.
    """
    points = list(product(x_range, repeat=n))  # Cartesian product to create grid points
    gamma_values = {point: np.prod(gamma(np.array(point))) for point in points}
    return gamma_values

# Example usage
n = 6  # Number of dimensions
x_range = np.linspace(0.1, 2, 11)  # Adjust the range and density as needed

gamma_values = simplified_nd_gamma_plot(n, x_range)

# Displaying the result
for point, value in gamma_values.items():
    print(f"Point: {point}, Gamma Product: {value}")

# Note: Visualization of these values would still be limited to 2D or 3D plots.
    
ONLY "DALL-E Discussions" should appear; this is a logo. Vignette to black background.```

Please accept my gift to the channel:

[dim cradle]

wow

[onyx ridge]

Yup. It's a visualization of our community using code, higher maths, semantics, and a little bit of brute force.

Each node in the global network connected through the hub of #dall-e-discussions.

[fickle ravine]

If dall e disscusions had a profile picture this would be it

[onyx ridge]

https://discord.com/channels/974519864045756446/1202122539548299264

We now have an UNofficial profile picture.

[fickle ravine]

Each and every single one of your pictures deserves to be in the hall of fame

[onyx ridge]

Thanks man.

[fickle ravine]

Why is it actually like that in the morning, it is somewhat slightly active here, in the afternoon this is like a ghost town but in the evening it starts being very active?

Is it because of timezones?

[onyx ridge]

Sleep eventually succumbs to us all.

Or wait.

[shut niche]

These are from playing around with lighting techniques earlier today.

[onyx ridge]

JJ Abrams says you get to keep making films.

[onyx ridge]

Did you get it working? 😮

[fickle ravine]

It looks like some far away galaxy from the deep universe

What???

I pay for it to make pictures but t just refuses to

[onyx ridge]

It can't copy real people, remember?

[fickle ravine]

Yeah, but with Gottwald, it at least tried

[onyx ridge]

Fair.

It still is trained to not.

[fickle ravine]

I rather deleted the picture, because it could be classified as political talk (even when I talk about him in a neutral matter and don´t even mention his name, I can´t be sure about those "fair" and "working" rules)

[dim cradle]

@untold crypt great image, what is it?

[dim cradle]

only emulating rn, i plan to dive into your scripts at some point though, impressed by your python work

[onyx ridge]

It's a good emulation. I've emulated it too and come close. Yours is good.

[fickle ravine]

Asked AI to make picture 1 better

[woven wren]

Keep in mind the filter and the content policy are two separate things. Not everything prohibited by the content policy falls within the scope of the filter. It's still your obligation to adhere to the content rules whether or not the filter stops you. I know sometimes it feels like it's an edge-case or exception but err on the side of caution if you want to keep your account safe.

[verbal sequoia]

<:book_icon:1171408210398289941> `` Rule 0 `` Adhere to the following Terms & Policies:

Discord: Terms of Service, Community Guidelines
OpenAI: Terms & Policies

[onyx ridge]

This heroine is incredibly powerful, but her features are difficult to understand.

Her body warps space-time, but not like a black hole. Instead, like a wormhole.

Visually, she's a distorted silhoutte against the urban cityscape background in the night.

Her powers allow her body itself to act as a wormhole through space-time, with the heroine able to move the mouth of this wormhole arbitrarily, making her a multiversal hero.```

Literally my City of Heroes character in DALL-E of Heroes.

[late blade]

Imagine an intricate superheroine.

This heroine is incredibly powerful, but her features are difficult to understand.

Her body warps space-time and hints at an Einstein Rosen Bridge distortion.

Visually, she's a distorted silhoutte against the a central metropolitan cityscape background in the night.

Her powers allow her body itself to act as a rare quantum phenomenon , with the heroine able to control matter with the mouth at will, making her a multidimensional hero.

That was the prompt

[rigid snow]

Will inpainting come to dall-e 3 and if yes us there a planned realease date for it

[late blade]

You have to wait for news about that from OAI, we in the community don't know this. The best you will get on discord is: stay tuned and keep an eye on #spotlight

It will prob be available as a beta for some before it is released officialy

dang, that's quite the streak, 5 hall of fame entires in a row

[untold crypt]

Inside of a stone building with a wall of carvings of the greatest minds that ever lived and a whole in the roof showing a red eclipse

[fickle ravine]

Foals´ republic of China (reworked Chinese flag)

I hope I didn´t flood this chat too much

@late blade that new picture is so cool

I like the colors

[late blade]

@fickle ravine hmmm.... I got to tell you also, don't post so many images without a discussion, for your foals you can do a gallery

[fickle ravine]

I can make a gallery for pictures reworked by AI

[late blade]

yeap, and for here dall-e stuff

[fickle ravine]

Also I am surprised by this, because the text some sort of kept the meaning

It is not 100% exactly accurate, but the meaning stays the same, or almost the same

[late blade]

just be mindful that the rules that apply to others apply also to you

[fickle ravine]

Yeah, I am mostry trying to only make it with places now, not people so much

But the Czech president statue was pretty much a success

And this, dear kids, is a TOTAL and ABSOLUTE LIE.

Bruh it is literally scamming me by lying

[plucky hare]

It's not lying really -- it doesn't have photo editing features that would allow it to edit in something into an existing photo exactly as it is. It can, however, create a new image from scratch based on an image you provided, and try to implement additions that you describe.

In other words: you might try a request like "Please create a version of this picture with a racecar added in." It's probably just confused by the current wording of the request. Similar to what you did here actually: #images-discussions message

[formal osprey]

ask it to use code-interpreter

it will edit the image, but it won't be as well done as what dalle does. It's will just be funny to watch. I got it to do some interesting edits before

[fickle ravine]

Yeah thanks I can try that

[plucky hare]

Good point, could even try a multi-step thing like provide image, ask for separate image from DALL·E with pure white background, and then roll the dice with asking code interpreter to maybe even remove the white background before overlaying on the original image?

[formal osprey]

yes, it can blend pictures together. and it can use images generated by dalle.

however, there is one caveat. if it generates 2 pictures with the same prompt. only one of them is usable with code interpreter. This is because it saves the pictures using the same name. which is based on the prompt iirc

[plucky hare]

I tried to do a Brady Bunch style 3x3 grid recently from 9 different images 😁 spliced them all together flawlessly, just couldn't get the subjects looking in the right direction!

[formal osprey]

right, control with dalle is missing at the moment. It's amazing how much we are able to do with just text and seed control. (when we use a gen_id as reference_image_ids, dalle is really just forcing the use the the same seed. That becomes obvious when we use image A as reference to generate image B, then use image B as reference to generate image C. if any of image A, B and C has the same prompt, it results in the exact same image. (few pixels can change but it's mostly deterministic))

[fickle ravine]

Wait, why does it sometimes generate it right away then, and if not, then after a response regeneration?

[plucky hare]

In your original post that I just linked? That was probably just some temporary error with DALL·E that happened the first time, that didn't happen the second time -- it happens sometimes!

In your most recent examples: I think it's not working because of the phrasing that sounds like direct photo editing. Though it's possible that regenerating could return a DALL·E image eventually -- that would be just how ChatGPT in general interprets the same input differently at different times. Though I don't know if the "Add a..." request specifically could ever be interpreted as "Use DALL·E...".

[fickle ravine]

Sometimes the system is really messed up

Thank god an answer regeneration usually works

[fickle ravine]

#1202217118024876073 message

I did the gallery, but forgot to tell you

They are wasting my prompts

I regenerated and it said I ran out

If it at least didn´t waste the prompts, but it does

An answer regeneration should not be counted as another prompt

[wispy storm]

Unfortunately it doesn't even failed gen

[formal osprey]

sounds right to me. we have to be very precise with chatGPT about what tool we want it to use, and how we want it to do things. Especially now that it has access to multiple tools, it will often get confused and tell us it cannot do something because it assume we want it to use a specific tool, and it will refuse to try if the tool doesn't offer a built in way to do the thing. for example, image editing with dalle. We have to understand what chatGPT can do and we have to orient it to achieve the result.

which reminds me of that screenshot I found from a while ago, way before we had access to dalle3. if we hold its hand, it can do more

dalle in GPT 3.5 (I'm getting off topic, I'll stop there)

[fickle ravine]

Can I say "generate code art of a foal" for a similar answer?

[formal osprey]

I'll try to find that chat where I made it do image editing with code interpreter, I'll get back to you
(search feature in chatGPT would be nice)

hahaha, I forgot it started by tricking it. Sent that picture but named it "house.png", I do that kind of things to check what is hallucinated and what is actually happening.
anyway, that custom instruction was what I called "SchoolGPT", where chatGPT embodies multiple personnas and they collaborate to solve a problem incrementally. The "house" became interesting.

[fickle ravine]

You just gave me a perfect idea

That may be a very interesting one

[formal osprey]

for your specific usecase, maybe ask it to "generate a foal on a solid color background" (because it can't do transparent anyway), then ask it to "use OpenCV to remove the background and mix the 2 images". I would personnaly use photoshop or an equivalent for that, but it can be fun to see how far chatGPT can go

[fickle ravine]

What is your opinion about this picture?

[formal osprey]

"generate code art" I went down that rabbit hole, and I can tell chatGPT is very stubborn in not being artistic with code. But when we encourage it a bit, it end up producing stuff... I managed to make it draw a human before. It's some weird horrors

[fickle ravine]

This will scare me in my nightmares for the rest of my life

What is that 💀

[formal osprey]

how chatGPT see humans, lol

I mean, considering it's not supposed to have a world model, to me it's impressive

[dim cradle]

whats with so many foals lately anyway?

[formal osprey]

unsure, it's looking realistic visually, but I doubt it makes so much sense. I don't really know what that background is representing.

[fickle ravine]

To be completely honest, the only reason I made that background is so it fits in daily theme

I hope next time the daily theme will be some fruit or vegetable, that is very promising for pictures involving foals

[formal osprey]

lol, I see what you mean. I made a lot of "crafting" themed pictures yesterday. but wasn't satisfied with the result enough to post.

That would have been in my post, but didn't have enough time

actually, the first one was probably fine. the 2 others, I'm not satisfied with the coherence

[fickle ravine]

All of them are really nice bruh

[formal osprey]

yea... but my reflex is to count fingers now. such a weird reflex to develop. if you asked me 5 years ago how AI might change me, I would not have guessed that one.

[formal osprey]

https://discord.com/channels/974519864045756446/1202247091842465842 posted them in a gallery since you like them

[fickle ravine]

I remade Serbia, now it is the republic of cute foals

[fickle ravine]

This is just cute

[grizzled loom]

good morning from Missouri

so close right, sad

[fickle ravine]

I love those, that is just so cute

[grizzled loom]

this one was good

[late blade]

Did you guys read the article about Microsoft urging OAI to remove DALL-E 3 because of security risks and bypass mechanisms for generating controversial content?

[grizzled loom]

no. do tell

[late blade]

It’s on slashdot

[grizzled loom]

i follow tech trend sites as often as the average user churns homemade butter

short ofit? link in DM?

[floral socket]

🤙

[late blade]

Will do as soon as I get home

[fickle ravine]

That would be a scam

We pay for it

[vapid elk]

it isn't Microsoft, it is a public post from a one microsoft employee

[late blade]

You are correct

[fickle ravine]

So it´s just an opinion of one employee, and not the CEO of Microsoft?

[floral socket]

Are you guys also having the problem that u reach the usage limit after just 19 dall-e messages?

[vapid elk]

idk, I feel like that dude violated 7 NDAs at once 😆

[late blade]

I’m on mobile. I tend to write shorter on mobile

[vapid elk]

it was the statement of one employee, not the company

[fickle ravine]

Good, loks like one employee has to be fired

[vapid elk]

¯_(ツ)_/¯

idk, people would expect that a seasoned microsoft engineer would know better to disclose a software vulnerability correctly and not post about that to the public

the urge to just brag online about "hey, I know about this big thing" is strong.. but come on.. doing that will only instigate people to go and try to find it too which decreases the time to actually fix it

[fickle ravine]

Cute

[grizzled loom]

its silly to be honest. I totally respect OAI desire to make the Librarian version of AI. i'm more of a Johnny Five Short Circuit hanging out with a Bender bot kind of guy myself. Microsoft thinking any of their use of these systems is not subject to prompt injection is Hillarious to me. I hope every engineer reading this understands that all of these language models are subject to it.

I mean, I have to high five OAI for cleaverly turning the exploit of intructions into custom gpts. that was a incredible 180 to the problem

[vapid elk]

now that I actually read the post... it is very lame IMO...

[fickle ravine]

No way this one made it to the hall of fame

It is not even THAT cute

[grizzled loom]

the tech is going to fly by; and these models will eventually produce error free imagry and that isnt far off in many cases already. In fact, I postulate that all of our trash images that look rediculous will be collectible in 20 years when you can't get a model to make those kinds of mistakes anymore. the brith and death of AI incidental digital surrealism. lol

[late blade]

ok, I'm back at a "stable" connection

btw, as much as I love today's theme, I have a confession to make: today I've been in lazy mode

I'm in -> DALL-E do this, give me 2 ideas and make the images...

[grizzled loom]

cant waste calls on daliy when i am workin on directive instructions or i short my work time

[fickle ravine]

I hope the next theme will be something with more potential

[late blade]

there's always potential if you invest the time

imagine a computer scientist saying: "I can python!" because he knows how to do a HelloWorld print

I still have to filter the images dall-e proposes

it's not like I asked GPT to post the images also

nobody has done the Flintstones house?

[onyx ridge]

Yabba dabba doo!

[late blade]

so, any challenges going on atm? or takes on someone else's concepts?

[onyx ridge]

WILMA!!!

[late blade]

This one was the closest I was able to make: #daily-theme message

[fickle ravine]

I am getting bored and just unexcited about this theme, thank god it ends in an hour

The pictures that I make with this are very repetative

[late blade]

calm down, dall-e won't go away

how about stepping up your skills? try a different approach than enhancing images with gpt4v

[mighty forum]

Hey can someone make a half cat half robot pic pls dont ask why

LMAO

[fickle ravine]

You are welcome...

[late blade]

I knew that would happen

[formal osprey]

what is going on here

[formal osprey]

btw, on that same idea. concerning daily theme. I had that series with chatGPT yesterday 😆

"no, you can't just do an anime girl as Architectural Stonework"
- just watch me

but again, it was funny. But not good enough

[fickle ravine]

Those errors creating images are so annoying

[formal osprey]

it's to save the world

jk

[fickle ravine]

The worst thing is that when you regenerate the answer it wastes the prompt

Why does the CEO of openAI not even care about fixing it?

[formal osprey]

it's complicated®

[mighty forum]

LOL ty

[fickle ravine]

They could at least set it so it doesn´t just waste your prompt

[formal osprey]

Those are not real errors, they are "content policy violation". and/or "flagged as inappropriate". But I think we can agree it's false positives most of the time.

like, that giant anime statue... I doubt anyone would be harmed by this. It's obviously a joke.

[fickle ravine]

Yeah they are obviously joking with this "functional" and "working" system

[formal osprey]

We can guess the reasons one of them failed. But there's no way to know. It just waste one of our limited daily generation. I do not want to speculate about why they are not fixing it. All I can do is to propose alternative solutions (#1201101776514973696).

[fickle ravine]

I can DM the reason why I think they are not fixing it

Idk if I can say it here because of the censorship

[late blade]

I had something like this earlier, but wasn't able to keep up with I wanted to post, so that kind of stuff was more or less there but not used

[onyx ridge]

The issue is that they can't fix it. The model is trained on all kinds of garbage it could never output. But it's got no idea that is the case. The struct supervisor agent needed to be on the ingestion. Then we'd get more usable images with less waste.

It's literally a problem of mismatched scale.

This is not theory, it's factual.

[fickle ravine]

Are people getting bored by those pictures of foals?

[late blade]

on a moderate healthy dose, no

but my behaviour in general when something is too repetitive, I just tend to fly by it

[fickle ravine]

Yeah, me in math classes

[late blade]

never skip math, NEVER!

[fickle ravine]

I know but it´s just too hard to pay attention

[formal osprey]

happens to me often. I start generating stuff with ChatGPT, kindof a brainstorm at first. Then when I have a clear idea of what I would like to create, I try various prompts to obtain a decent result. a lot of them results in an error, a lot of them are just broken pictures. a lot needs slight adjustments.

generally, I don't have enough time to get to a decent result, so the stuff remain unused

the brainstorm phase is painful when there are errors. because I don't even see the image or the idea.

[tall mason]

I've had three ideas for today and none of them are panning out particularly well. 🤔 I must be unlucky.

[fickle ravine]

Only zombies can can bad taste

Funny like this text on the board

[tall mason]

I once was told engineers aren't English majors. I like the three different attempts at thermal capacity.

[late blade]

oh dear, you gonna kill the poor cat with the shards of the mirror under a ladder and he's gonna end up with that umbrella as a delivery bag to the grave

[fickle ravine]

Yeah everyone liked that

[tall mason]

It was all just an unfortunate accident, I swear!

[fickle ravine]

I like that random "in" that is on the board

[deft musk]

Yep!

[fickle ravine]

It just translated it

Not with 100% accuracy, but it tried

[late blade]

today's topic is 9 months too early

[mint tiger]

Does anyone knows if our friend Moxi is planning an Alpha testing form for the discord folks for the new features of DALL·E Controls that let you adjust : Prompt (Boost or Strict), Style , and Aspect Ratio?

Yeah one guy that reverse engineer the Web Client found a New DALLE Controls feature, allowing various settings to be selected on image generation prompt. Moxi was pretty generous to us last year letting a couple of users to try DALLE 3 before is general release, so I was asking if someone knew more about a possible new alpha testing form..

[late blade]

I see, no clue tbh

[ripe wraith]

just me or this pretty as heck?

I got it to tell me some Buddhist quotes, then asked it to draw them 💡

Zen proverb: "The moon does not get wet, nor is the water broken. Although its light is wide and great, the moon is reflected even in a puddle an inch wide."

This quote paints a vivid picture of the moon’s reflection in water, highlighting the idea that the essence of things (like the moon’s light) remains unchanged regardless of where or how it appears, akin to the unchanging nature of our true self.

[fickle ravine]

It is

[ripe wraith]

Thich Nhat Hanh: "The present moment is filled with joy and happiness. If you are attentive, you will see it."

missed the memo about breaking the glass ceiling

[late blade]

pretty much

[fickle ravine]

When cuteness is the theme you can do this and get 58 stars insteantly

[late blade]

DALL-E being stubborn atm

I sense I'm gonna get a penalty box soon because of how stubborn DALL-E is atm

[wanton moon]

also in a penalty box

[late blade]

I just got there too

[wanton moon]

spend 10/15 boosts to create a close enough image

[fickle ravine]

Cute 😭

[late blade]

oh i'm in the gpt+ pentaly box, in an hour or so I'm back to doing stuff

[late blade]

tears are permitted

oh my goodness, got an amazing image before the penalty box, this will make Yami jealous

[fickle ravine]

1. Why does the plane look completely different than the one I wanted (picture 2)
2. Why does the foal have 5 legs???

[wanton moon]

why is this so cursed

[late blade]

I wish I could say something about it but I'm missing some information

[wanton moon]

It was an image i found right after the penalty box meme, and I found it a bit cursed, so i shared it with you people

[late blade]

ahaaaa

[plucky hare]

1. DALL·E doesn't have "eyes" -- when you provide a reference image, what's happening at a basic level is that the reference image is being interpreted by a separate image recognition feature built-in to ChatGPT. This creates a description of the image in text -- in your example, probably something like "A jet engine airplane, red and white, stars on the side." This description is then fed back into DALL·E, still as text. That's why you get detail "loss" when trying something like this -- you're going img2txt2img.
2. Fine details are a challenge for all AI image generators -- there's an extra leg for the same reason letters are hard and fingers on a hand are hard. DALL·E doesn't "know" what a human hand looks like -- it has underlying statistical programming that can learn patterns and then try to recreate them, to better and worse effect! Put more simply: it's like a dice roll where you know generally what the outcome will be, but not exactly.

[fickle ravine]

Oh, thanks! Both of those answers actually make sense

When I saw what some other free AI image generator imagined as a pregnant horse, this is completely fine

[late blade]

see, kittens get happy when they understand

[wanton moon]

OMG so cute

[wanton moon]

that's is why my cat is always grumpy, thanks for making me understand

[late blade]

lol

[wanton moon]

replace A with E and that image would have been really hillarious

[late blade]

lol

This #daily-theme message is gonna make @formal osprey so jealous, my prompt was:

An image so good that will make Yami angry and jealous.

[wanton moon]

I cannot believe you played that card, I guess it is my turn to play: Four headed five tailed beasts of supper and sent one of your cards to the graveyard so I can resurrect one of mine from the graveyard and put it in your deck

[late blade]

lol

gotta admit, the image turned out surprisingly well, I think it was a fluke

now I got 7 years bad images

[wanton moon]

It is a very beautiful image and worth the risk

[late blade]

To be fair, @formal osprey makes amazing images

[fickle ravine]

This is a pretty weird theme, anyone can imagine anything under it

[fickle ravine]

Yeah he does, but you do too

[late blade]

something is so wrong with my prompt and I sitll can point it out

pin point it

[fickle ravine]

I love how this looks normal at first, but then you see how giant the foal is

[late blade]

ok, here's the challenge:

Art Style has to be from now on something other then realistic

The foal has to be engaged with the scene

The foal has to be dressed

Do you accept?

[fickle ravine]

Ok, I try it

[late blade]

cool, if you don't know which art style to choose, you can browse #1108740112558325790 for some ideas

[fickle ravine]

💀

I chose a cartoon style

[late blade]

the idea behind my challenge is,this kind of exercise and creativity helps you grow with your prompts, get new ideas on how to work with prompts

which is pretty much what we all do around here from time to time

[cold raptor]

Hi. What you think for my technology? this image was created from 4 photos (forest, salt, gloxinia, tower-bridge) Evaluate the quality of the conversion

[fickle ravine]

Oh my god that looks so pretty

10/10

Underrated picture

[fickle ravine]

This is the definition of the word underrated

[cold raptor]

Thanks for rate

[late blade]

does the blend of technology has a purpose?

Something like the gloxina being the sustainable power source for the towr bridge and cityscape?

[cold raptor]

Sure This opens up the possibility for a large number of experiments with understanding the context. One work has already been accepted for an international digital art exhibition

[onyx ridge]

Has anyone done Rowko's Basilisk yet today?

[late blade]

not that I'm aware of

and I'm guessing no so far since there's only 3 mentions of it in all of the server, all coming from you

I have to educate myself and see what that's all about

[onyx ridge]

hahaha

Beware

It's allegedly some sort of thought bomb

I'm going to post it.

[late blade]

I think the knock on wood too hard thing is not working

[onyx ridge]

I like her angry, frankly.

Maybe the shield needs to be more perpendicular to the ground, but otherwise, good form.

A polearm's getting over that shield though, and naginata were a real threat.

[late blade]

it's supposed to be a mirror that breaks as a side effect of her knocking on wood really hard

[onyx ridge]

Oof.

Sorry.

I am visually disabled. Don't take it wrongly.

Please, that is.

[fickle ravine]

[late blade]

Finally, I think I got the image I wanted

[fickle ravine]

And I got something for my gallery

[onyx ridge]

Roko's*

Roko's Basilisk handled. I did my research to buff up on the subject.

[late blade]

I'm behind, but I'll get there

[fickle ravine]

There is that problem about this theme that anyone can imagine anything under that category

[onyx ridge]

My theme post actually has a good overview, if you like. #daily-theme message

[onyx ridge]

I made a prompt that produced images of the most unlikely possible events yesterday. We were having fun with it.

This isn't that different.

[fickle ravine]

Yeah, i just say everyone considers something else as luck

[onyx ridge]

I mean, I have a high success rate. It's not just luck.

Layered language to guard against the brutality of the record retrieval rate curve helps.

[late blade]

That's great!

[dim cradle]

Today’s theme sounds fun!

[onyx ridge]

I'm going to struggle to take it further now

Roko's Basilisk is pretty superstitious

[dim cradle]

I can finally spend some time on it. First I’ll have to see what you guys have been up to.

[onyx ridge]

Welcome back.

[late blade]

We are making the world a better place while making the daily theme dangerous.

[onyx ridge]

I should feel much worse about this post than I do.

The concept is generally rejected. Critics argue that a superintelligent AI would not find it beneficial to punish people for past decisions, as this would not align with its goals.

That's why they're not superintelligent.

The critics, that is.

/sarcasm

[late blade]

@fickle ravine #daily-theme message <- see you can do other kinds of images

welcome to getting out of your comfort zone 101

[fickle ravine]

Those ones I made could actually be used in some charity program fighting against foal torture

It just fits so well

This one turned out really well

[late blade]

how about myths, rituals or superstitions about foals? anything besieds clovers?

This #daily-theme message looks more like "Das Boot"

[dense mesa]

seems pretty arrogant to assume ASI would be burdened by our petty emotional foibles

[onyx ridge]

Mentioned above.

[onyx ridge]

Thanks for the mention.

[dense mesa]

well wasn't directing the arrogant part towards you. just saying. we humans love to assign our emotions and feelings to concepts far beyond us

[onyx ridge]

I took it that way. I don't feel attacked, you are cool. 🙂

[quick elk]

I am literally paying for chatgpt just to generate images and it cant even generate io get error after error of i cant generate that at the moment i literally have an urgent job with those images and chatgpt cannot generate them annd charges me credits

[late blade]

Hi

Nice to meet you

[onyx ridge]

Hi. Welcome to the chat.

What are you trying to make?

Can you share your prompt? That'll help us help you.

Otherwise, it's going to be an uphill battle. When's your deadline?

[late blade]

you are nicer than me, I am still waiting for a reply to my greeting

[open trench]

hello all

[dim cradle]

hey hey

[onyx ridge]

Queen playing in my head right now.

[dim cradle]

Is this real life or is this just fantasy….

[onyx ridge]

That's not quite right:

[formal osprey]

Lol, if one day dalle knows who I am and using "Yami" in the prompt enhance the quality, I'll be honored

I like the mirror smashing vibe in those pictures, so silly

[late blade]

dall-e doesn't understand the concept

[formal osprey]

I admit when I saw the daily theme, I had the blank page syndrome

[late blade]

what is that?

[dim cradle]

Seems we’ve covered most superstitions. We will need to think creatively, outside the box, as it were.

[formal osprey]

some bad translation. "writter's block" is more common I think

syndrome de la page blanche

[dim cradle]

Ernest Hemingway, Virginia Woolf, and Sylvia Plath

@plucky hare 💜 The Office reference

[plucky hare]

😁 Thanks, me too! Wish he looked more Michael-y but when DALL·E nails the spelling like that you just gotta go with it!!

[dim cradle]

Eye of Horus, an ancient Egyptian symbol of protection and good health.

When referring to "RTX HDR" in the prompts, it's more about invoking the idea of high-quality, detailed, and dynamically lit images, rather than specifying the actual technology used in the image generation process. The terms are used to guide the style and quality of the image rather than to describe the technical method of its creation.

[grizzled iris]

Nah this is quite good haha 😝

[grizzled iris]

Quality is good 👍🏽

[dim cradle]

I'm iterating on the Curse of the Evil Eye. I've decided the perspective shall be the viewer's.

[grizzled iris]

You’ll get there, I believe in you

[grizzled iris]

Tranquility

[grizzled iris]

Interesting 🤔

I think it’s time Current Theme:
🍀 superstition - mystery, luck, belief. a wild unknown!

Since it’s superstition I must do a UFO one haha 🛸

[dim cradle]

alien abduction, medical experiments....

[grizzled iris]

Peaceful first contact

[dim cradle]

ah very good. then maybe they will share some of their medical technology.

[grizzled iris]

@dim cradle I think DALLE is onto something here

[grizzled iris]

Exactly we need more free universal healthcare technologies

[dim cradle]

Oh is that what transpired there? The world has wondered.

[rancid laurel]

How do I make art with dalle

Im just starting

[grizzled iris]

What would you like to create?

[dim cradle]

There are various ways to get started.

[wispy storm]

#ai-discussions message

[grizzled iris]

Omg @dim cradle

Just realised what we can do!

Remember that super old computer they found under the sea

Antikythera Mechanism

2,000-year-old device often referred to as the world's oldest "computer"

With AI we could finally probably crack the case and try to piece it all together

Im gonna try to rebuild this with DALLE

Hmmm not quite

[grizzled loom]

hood star leprachauns will be today's daily drops

[dim cradle]

interesting concept

[grizzled iris]

That would be a good challenge

To get it as close as the academic research papers have identified it to look like

[dim cradle]

sounds like a project

🍀

[ripe scroll]

@.@

woops

[tall tartan]

Does anyone know of a relatively simple way to get it to generate more than one image at a time?

Like it used to. 🥲

I end up asking it to "Create another variation of this image with the same prompt."
But it would be rather nice to have it create 2 images (or more) simultaneously.

[onyx ridge]

Try this:

1. Create an image with {Description.}
2. Create the image again.
3. Create a variation of the image with the same prompt.
4. Show me a variation of the previous image.```

Let me know if it works for you.

By changing your instructions for each image only slightly, you (potentially) trick the AI into activating the tool again (and again...)

[plucky hare]

Hello, the official DALL·E GPT still returns 2 at a time! https://chat.openai.com/g/g-2fkFE8rbu-dall-e

[onyx ridge]

Do you uspect it's doing this differently than my prompt though?

[tall tartan]

Ohh.. In that case, I feel like a proper derp now.

[onyx ridge]

I strongly suspect the DALL-E 3 GPT simply does this:

2. Create another image.```

[plucky hare]

Does your prompt return them concurrently or one at a time?

[onyx ridge]

One at a time. That's a good point.

So they are using an Action to request a payload of two images.

That's clever.

Very clever.

[tall tartan]

Thanks for this. I'm going to give it a try.

[onyx ridge]

You still have a daily image cap1

DALL-E 3 frequently fails to produce 2x 16:9 images now though, since today

[plucky hare]

On DALL·E launch it was 4 at a time! Glad we still have 2 at a time as an option. And it is indeed nice that we can at least request multiple image generations in a single reply!

[onyx ridge]

Indeed. I had to consider how to reply, but I'm fairly certain I'm not providing an exploit here.

[tall tartan]

"On DALL·E launch it was 4 at a time!"
Yup! I remember. Better times...

[plucky hare]

I agree, I think they would have cracked down by now. That's what their 15/60min throttles are for!

[plucky hare]

Whatever it takes to keep the servers from melting 😁

[tall tartan]

True

[onyx ridge]

It took more than a second of thought to push through the ethics.

[tall tartan]

So, in theory, could a normal user create a Custom GPT that uses this technique to request a payload of more than 2 images? Do you think?

[onyx ridge]

Several people have.

I've considered adding the feature to my own.

[tall tartan]

ohreallyyy

[onyx ridge]

But out of respect for the GPU deficit...

I've decided it's counterproductive to include it in GPTs I make.

[plucky hare]

But not simultaneously right? n will always equal 1, but you can return n = 1 requests multiple times in a single reply

[onyx ridge]

The GPU deficit is real.

[onyx ridge]

The AI will follow the prompt instructions, requesting n=1 x times.

You could even make a loop, and it'd evaluate it to its best ability

[plucky hare]

I think DALL·E GPT's n = 2 is a unique function then, unavailable to "regular" GPTs

[onyx ridge]

it'll almost surely fail and stop before the loop ends (retrieval rate curve) but it'll try

[tall tartan]

Hmmm... ok. I prefer to be a good member of the community too. Just hope and pray that when they get the GPU deficit sorted out that top tier service will resume.

[onyx ridge]

Yeah I think they use the OpenAI OpenAI API and call custom actions n=2

[dim cradle]

@onyx ridge i'm intrigued by your python code for dall-e -- i see your simplified_nd_gamma_plot function creates a grid in a 6-dimensional space and outputs a dictionary, i'm just curious how you're using it in your prompt -- do you mind DM'ing me a sample payload that you sent to the api? I'm just curious to learn more.

[onyx ridge]

Since it's their API they can just charge it.

[plucky hare]

I agree!

[onyx ridge]

I don't use the API.

That's the full prompt. I'll post the original here again for you to test in chat.

[dim cradle]

you can request the gpt output the same payload it sent to the api.

[onyx ridge]

Oh. I didn't know that. Let me drop it here anyway.

/image 16:9 "Lexideck" set in futuristic holographic typeface, is set against a black void with brilliant multicolored neon banding creating a visualization of the evaluated function in a feed-forward hyperplanar topology:

import numpy as np
import matplotlib.pyplot as plt
from scipy.special import gamma
from itertools import product

def simplified_nd_gamma_plot(n, x_range):
    """
    Create a simplified representation of the gamma function in n-dimensional space.
    This function will create a grid of points in n-dimensional space and compute the
    product of gamma function values at those points.

    :param n: Dimension of the space.
    :param x_range: Range of values for each dimension.
    :return: A dictionary mapping n-dimensional points to their corresponding gamma function values.
    """
    points = list(product(x_range, repeat=n))  # Cartesian product to create grid points
    gamma_values = {point: np.prod(gamma(np.array(point))) for point in points}
    return gamma_values

# Example usage
n = 11  # Number of dimensions
x_range = np.linspace(0.1, 2, 11)  # Adjust the range and density as needed

gamma_values = simplified_nd_gamma_plot(n, x_range)

# Displaying the result
for point, value in gamma_values.items():
    print(f"Point: {point}, Gamma Product: {value}")

# Note: Visualization of these values would still be limited to 2D or 3D plots.
    
No words other than "Lexideck" should appear; this is a logo. Vignette to black background.```

You can tweak the semantics and name, and the dimensionality changes the surface features.

So you'd change n to 3 for, say, water in a shark tank.

And n to 6 for, say, a feed-forward network of DALL-E Discussion users.

I don't have all the topologies memorized yet, but enough.

[dim cradle]

the function results in this, for example, i'm just curious how that is being transformed, i'm curious what the actual prompt in the payload looks like.

[((0.1, 0.1, 0.1, 0.1, 0.1, 0.1), 741385.4247430449), ((0.1, 0.1, 0.1, 0.1, 0.1, 0.29000000000000004), 241593.44848449738), ((0.1, 0.1, 0.1, 0.1, 0.1, 0.48), 143804.27773359325), ((0.1, 0.1, 0.1, 0.1, 0.1, 0.67), 105065.20067189095), ((0.1, 0.1, 0.1, 0.1, 0.1, 0.86), 85966.23111742894)]

If you don't mind askig the gpt for it and sharing it.

[onyx ridge]

That's right. DALL-E takes those points and tries to plot them in the diffusion grid while following semantic instructions.

The output is its interpretation in some coordinate system, which the program lets it choose, of those values represented in space and color.

I don't have to ask GPT how, lol

it's following the semantic instructions while applying the data structure from the function output to the appearance.

So the output won't be fixed but it will be based on the data the function returns.

It's a creative way to use higher maths to regulate the model output, that's all.

[dim cradle]

Thanks for the info! I think I understand, I was just curious to see the actual payload. After you use that prompt to generate an image in your gpt, if you ask it to output the payload that it sent to the api in a code block, i'd just be extremely curious to see what the prompt is in that, i think it would help me to understand more completely.

[onyx ridge]

I have a way of speaking that might seem combative. I wasn't trying to be. Your inquiry is welcome.

[onyx ridge]

I have analyzed the output up to 11 dimensions. It is too vast at 12 to output in the token budget.

Your output matches what I expect for the snippet.

[onyx ridge]

I also misread your question. D'oh!

Man now I feel bad. I thought you were asking me something else. To ask GPT to explain it.

But your output absolutely matches mine.

I'm genuinely sorry for the miscommunication. :\

I try to read my screen because the screen reader makes me insane. Maybe it's the opposite when I read...everyone around me.

[onyx ridge]

I don't understand my output now.

I was getting output yesterday with this exact code.

Now it's throwing memory errors.

[dim cradle]

just got back, catching up

[onyx ridge]

Ah, it is running out of memory because there are 285 billion cartesian solutions. Yesterday it wasn't as ambitious.

Today it tried to do the whole function.

[dim cradle]

it's all good 🙂

[onyx ridge]

This is interesting model evolution.

[onyx ridge]

Thanks for understanding.

I'm convinced this is a fine-tune situation from the update.

Shark Tank I set n=3, so it wasn't 285B parameters.

About one for each of ChatGPT's (hypothetical:D) nodes.

I'm suspicious the "Team" memory upgrade was really a downgrade for the rest of us now.

This function worked in 11d a few days ago...

[dim cradle]

I don’t want ya to spend any time on it.

[onyx ridge]

I have to.

The fine-tuning broke my engine.

😄

[dim cradle]

Did you also transfer workspaces? Some things broke.

[onyx ridge]

No. This is a Plus account for now.

Okay I have a function that works in the new space.

Huzzah.

AND It's stochastic, so it's superior to the old method, too.

Wheee.

I'm testing the corrected code now.

Yay, it's fixed.

[real holly]

Didn’t /daily used to give free credits? Did they remove it

[dim cradle]

it's temporarily out of order

[real holly]

Oh I see

Is it gonna be back soon

[dim cradle]

I don't think an ETA has been announced

[real holly]

I see aight

[dim cradle]

@empty kelp hey what's up?

[empty kelp]

Aloha!

i got slightly sidetracked from images. been trying to integrate the GPT 4 API with Unity to automate setting up Unity characters/animals with animation controllers

[dim cradle]

No whistling at night. No bananas on fishing boats. Avoid eating pork on the Pali Highway. These are some superstitions you might know about.

well, just don't forget about dall-e

[empty kelp]

Avoid Eating Pork on the Pali Highway.
Yes. If you go to the store and buy any type of pork-- and drive it like from Kailua to Honolulu over the Pali highway... The goddess Pele will try to get revenge on you.

[dim cradle]

Pele, the Hawaiian Volcano Goddess

[empty kelp]

This may sound like some type of superstition, but i can assure you that it's not worth trying. Bad things happen...

[dim cradle]

I will take that under advisement, I don't like to take unnecessary risks.

[onyx ridge]

The goal is to eat all the pork before you hit the highway.

To achieve this is going to be tough.

[empty kelp]

Everyone here knows it to be true, and we've seen many examples of it

[onyx ridge]

You're going to need a whole lūʻau to pull that off.

[dim cradle]

i like the vibe

[empty kelp]

that's a great topic for the daily though

the Night Watchers and the Menehune in the marsh are also excellent topics for the daily if you want to explore them

[dim cradle]

yes! Aumakua and Night Watchers are already on my list. I will have to look up Menehune

[onyx ridge]

Did anybody play any of the Samurai Shodowns?

If I wanted to get DALL-E to make characters like those, what styles of animation would I mention?

If the game's too old it's okay I'm a dinosaur.

[dim cradle]

i don't recall, was it sega genesis?

[onyx ridge]

It's good though so if you're substantially younger than me, add that series to your retro phase.

[onyx ridge]

Neo-Geo

[dim cradle]

Night Marchers (Huaka'i Pō): These are the ghosts of ancient Hawaiian warriors. It's said that they march at night to the sounds of drums and carry torches. Legend advises that if you hear their drums or see their torches, you should lie down and avoid eye contact to show respect and avoid harm.