images-discussions

[hearty ether]

Pus lounge 🤣

[vagrant pebble]

Yea I have a prompt for that, one second

[celest pebble]

dalle prompt limit is so so low, I always hit the limit

[dim cradle]

you can always make use of the API with a bit of fairly simple python to make the dall-e-3 calls, charged on a per use basis

[vagrant pebble]

Each aspect of the composition, from the positioning of the subjects to the intricate background elements, is designed to tell a story purely through visual cues. The absence of any direct explanatory or descriptive elements encourages the viewer to interpret the scene based on the visual information alone, making for an immersive and interpretative experience.

works great

Since negative commands don't work, I crafted a very positive command

[dim cradle]

words like "without" will have an effect

also I'm not sure of the effect they will have with the image generation model

Now I feel like making a swanky dining room/bar

[vagrant pebble]

Not really, it's more like the words of elements that should not be in there themself. DALLE scrambles the whole prompt into a big mush and if it sees "without a green elephant" it will create a green elephant and maybe the word "without"

[dim cradle]

they don't work well in LLM's because "attention" is what words carry the most meaning and by definition negatives are usually not important parts of a sentence

Do not think about a pink elephant riding a unicycle

[celest pebble]

@dim cradle I will get Copilot Pro next time, unlimited dalle usage

[dim cradle]

the "not" there is low on the attention ranking

[vagrant pebble]

ChatGPT is actually pretty great with handling negative prompts

DALLE isn't

And ChatGPT creates the prompts for DALLE which is why you need to be careful

[dim cradle]

if you look at human conversation and written text, you will often see things like "I said NOT to break the hanble off!" and you'll see humans use any number of linquistic and written tricks to try and bring attention to the negative

[celest pebble]

that have that 60"s era vibe

[dim cradle]

I'm sure the GPT models are getting training on negatives to address this, but its an issue that both AI and HI's have in common

[celest pebble]

any rumors on dalle4?

[vagrant pebble]

d3 just released a few months ago lol

[celest pebble]

motnhs ago in ai terms is a long time

[vagrant pebble]

Nope, dalle3 was released over a year after dalle2

[hot rain]

I had it make the architecture in an art deco style. Here's my favorite of the bunch so far.

[dim cradle]

i'm cooking something up

[grizzled iris]

Love the new theme of
🎠 carousel - nostalgia, exciting, fun. round and round we go!

Need to do an image asap

[dim cradle]

hyperrealistic art deco?

[grizzled iris]

Especially industrial photos like https://cdn.discordapp.com/attachments/979617962728226826/1199848388305367060/image.png?ex=65c408e3&is=65b193e3&hm=7ce1f0fcb0289c10ae26054a809c9a3604838137bef4a6ed6b344f0d345170a0&

#daily-theme message

They look super cool

[dim cradle]

needs alterations to the color palette... meticulous, rare

[grizzled iris]

Nice great gatsby vibe

Apparently a great gatsby GIF is not allowed here lol 😂

[grizzled iris]

Looooool 😂

[grizzled iris]

Silcon Valley 🤣

[grizzled iris]

Remind me of a place, I have been to many many years ago back when I had money 😂

[plucky hare]

From the DALL·E devs' AMA:

It only took a couple of months to train. Most of our time is spent trying out various ideas (whether in-house or from academic papers) on smaller models in an attempt to figure out what works and what does not. At some point we commit to compiling all of the ideas that "worked" into a big model push, resulting in something like DALL-E 3.
#dall-e-ama-answers - may need archive access in "Channel & Roles" to view.

However, agree that we're far from DALL·E 4! D1 to D2 and D2 to D3 were both ~1.25-1.5 year gaps.

[empty kelp]

this is really impressive to have so many people sitting at tables correctly

[dim cradle]

they're all very sophisticated, they are more cooperative than geckos i'm sure

[empty kelp]

i was getting all sorts of weird anomalies when trying to get lots of people sitting at tables

[dim cradle]

possibly different compositions, where this prompt is just pulling in static photos of people sitting together at dinner tables, we're sometimes giving arbitrary directions to a dynamic subject or subjects, like in those acrobatic images of yours

[dense mesa]

In this microscopic scene, we delve into the intricate world of cellular structures and molecular interactions. At a magnification of approximately 50,000x, we observe a vibrant cellular metropolis. The scene is centered on a large, complex cell, its membrane a mosaic of lipids and proteins, bustling with activity. Embedded within this membrane are receptor proteins, each a hub of communication, glowing with fluorescent markers. Inside the cell, the cytoplasm teems with life, filled with organelles like mitochondria, their double membranes and intricate cristae visible in detail. The mitochondria, powerhouses of the cell, are glowing softly, indicating high metabolic activity. Nearby, the Golgi apparatus appears as a series of stacked, membranous sacs, busy with the sorting and shipping of cellular materials. Ribosomes, tiny but essential, dot the endoplasmic reticulum, engaged in protein synthesis. The scene also includes a vivid display of the cytoskeleton, with microtubules and actin filaments crisscrossing, providing structural support and facilitating movement. In the background, strands of DNA are visible in the nucleus, with RNA polymerase molecules actively transcribing genes. This bustling microscopic scene is a celebration of the complexity and dynamism of cellular life, each component playing a crucial role in the tapestry of life at the microscopic level.

[dense mesa]

This image presents a futuristic vision of microscopic art, exploring the advanced and imaginative possibilities of molecular and cellular design. The scene is set within a highly detailed and technologically advanced cellular environment, where each component is a fusion of biology and technology. The cellular structures are reimagined as bio-engineered constructs, seamlessly integrating organic and synthetic elements. The nucleus of each cell is portrayed as a high-tech control center, with DNA depicted as advanced data strands. Mitochondria are visualized as energy reactors, glowing with an otherworldly light, powering the cell's activities. The endoplasmic reticulum and Golgi apparatus resemble complex machinery, processing and transporting molecular components. This microscopic landscape is alive with activity, with robotic enzymes and nanobots assisting in cellular processes. The art style blends realism with science fiction, presenting a vision of what the future of cellular biology might hold. The color scheme is a blend of natural hues and neon accents, emphasizing the fusion of organic and futuristic elements. This image invites viewers to contemplate the merging of biology and technology, showcasing a speculative yet captivating view of the future of the microscopic world, size 1792x1024.

[grizzled iris]

That’s good 👍🏽

[dense mesa]

not bad for these little baby prompts lol

[dim cradle]

you might like this

[grizzled iris]

Oh my God, you know I love to see this 🫶🏽

@dim cradle I’ve exhausted my limit trying to make an image. Maybe you could help?

[dim cradle]

maybe, depends

what do you need?

[grizzled iris]

Can you see the AR 3-D interface of the heart

[dim cradle]

mmhmm

[grizzled iris]

I need the image of the hybrid nano-bot right next to on the left side or the right side. I don’t really care as long as it’s on the actual interface of the surgery.

Really been struggling with that for the last several days…

[dim cradle]

that sounds like a fun thing to try

gimme a little bit

[grizzled iris]

I would be amazed if you could do it 🙏🏽

[grizzled iris]

Thank you so much for trying

[empty kelp]

if you keep your cellular prompt exactly the same but say it’s an “underwater image” it makes it a sphere

[grizzled iris]

That’s a pretty cool idea thinking about it now 💡

[dense mesa]

you should say it's a cellular star or something, lol

[dense mesa]

This creation is a microscopic interpretation of a futuristic circuit board, where the natural world of microbiology meets the precision of electronics. The scene is set at a molecular scale, where individual atoms and molecules are reimagined as components of an advanced circuit. Atoms of carbon, oxygen, and nitrogen are portrayed as nodes and junctions, interlinked by molecular bonds resembling traces on a circuit board. Enzymes and proteins take the form of intricate microchips and resistors, each contributing to the functionality of this molecular circuit. The backdrop features a network of biochemical pathways, visualized as a complex layout of a circuit board, with vibrant colors highlighting different functional areas. Ribosomes, the cellular machinery for protein synthesis, are depicted as capacitors and inductors, integral to the circuit's operation. This artistic fusion of molecular biology with circuit board design creates a scene that is both scientifically intriguing and visually captivating. The canvas size of 1792x1024 is filled with these microscopic details, adding layers of depth to create a three-dimensional effect, offering a glimpse into a world where biology and technology converge at the most fundamental level, size 1792x1024.

[grizzled iris]

Thank you so much for the prompt mate 🙏🏽

[empty kelp]

yes, these are some juicy verbose prompts, PicturesOnPictures. they’re very good

[grizzled iris]

This one got me thinking for several minutes

[dim cradle]

Juicy lol

[grizzled iris]

I’ve been trying to scientifically figure it out. I still haven’t got my head around it yet

[dense mesa]

Envision a microscopic scene at the nanoscopic scale (1 nm - 100 nm), meticulously crafted to resemble a futuristic, highly advanced circuit board. This world, hidden from the naked eye, is a marvel of nanotechnology, where each component is an intricate assembly of molecular structures, designed with precision to mimic the components of a circuit board. In the center, a colossal structure made of carbon nanotubes stands tall, resembling a central processing unit (CPU). These nanotubes, each a cylindrical arrangement of carbon atoms, are renowned for their exceptional electrical conductivity and mechanical strength, symbolizing the core of computational power.

Surrounding this central CPU-like structure are arrays of molecular transistors, each meticulously arranged to form a complex network. These transistors, composed of precisely engineered molecules, function similarly to their macroscopic counterparts, controlling the flow of electrons at the nanoscale. The scene is interwoven with pathways of gold nanoparticles, acting as conductive tracks, connecting different components of this molecular circuit. These gold nanoparticles, with their exceptional conductivity and resistance to oxidation, are vital in maintaining the integrity of the circuit's pathways.

In the background, a series of quantum dots, each a semiconductor nanocrystal, are arranged to form a display panel. These quantum dots emit light at specific frequencies, creating a vibrant display of colors, much like an LED screen, but at a molecular scale. The foreground is adorned with a field of molecular capacitors, each a small assembly of layered materials, storing and releasing energy as needed, crucial for regulating the circuit's power supply.

This scene is not just a visual spectacle; it represents the fusion of art and science, where the microscopic world of molecules and atoms is transformed into a functional, futuristic circuit board. The entire canvas is filled with intricate details, showcasing the beauty and complexity of nanoscale engineering, inviting viewers to contemplate the wonders of technology at the molecular level.

[grizzled iris]

Molecular technology is something I haven’t tried to use as a prompt that’s quite a good idea also

[empty kelp]

i think DALL-E cells just happen to come dehydrated till you put them in water

[dense mesa]

that prompt is a longboye

[grizzled iris]

Because every time I say, nanotechnology, it is an issue, so I will try molecular

[dim cradle]

Ironic that microscopy takes such a macroprompt.

[dense mesa]

not without some convincing. overhauled my instructions

[grizzled iris]

Maybe DALLE is trying to tell us that DALLE needs some water cooling - or even better a connection to a quantum API

[grizzled iris]

Actually thinking about it, this is a perfect image. I just need the doctors that are next to the patient have an AR headset to pilot the hybrid nano-bot 🙏🏽

[dim cradle]

it's a neat concept, i'm diving into it now

[empty kelp]

it looks like the stethoscope has a big Q-Tip taped to it instead of ear buds. is that the bluetooth transmitter?

[grizzled iris]

Looooooool

[dim cradle]

futuristic tech, but the idea of nano/ar/vr/etc is to help eliminate the destructive part of the process- less recovery time, less risk, i assume

[grizzled iris]

I just noticed that now

[dim cradle]

greater precision, less margin for human error, list goes on

ai-augmented or

[grizzled iris]

Exactly the use of the hybrid / non hybrid nano-bots, either they be DNA based or bacteria based or enzyme based. There are many different types out there now reduce the risk of complications since you don’t need to cut up the patient and lose blood

[dim cradle]

my first prompt had some issues, i have greatly enhanced it with clarity and precision

[dim cradle]

fascinating

[grizzled iris]

Many applications

Mainly to tackle cancer, but also medicine delivery is also a good one to explore

Because actually thinking about it now, I just realised medicine delivery would be perfect

[dim cradle]

hrm, 2nd revision still misfiring on 2 key points

[grizzled iris]

Because you know how we take medicine right now, we just basically destroy the good bacteria also

When we take the medicine and we drink it with water normally

But if a doctor pilots using a AR headset, a hybrid nanotechnology robot and delivers medicine to the area that needs to be specifically treated, that would reduce complications

[dim cradle]

the scene is clear, must be a matter of weights and interpretation. i understand the requirements, though. might take another rewrite or two. but i do like the concept.

[grizzled iris]

Haven’t still figured out how these work… the magnetic ones below

As you can see from the image, it is not very detailed, unlike the other types of nanotechnology doctors 🥼

[empty kelp]

the Q-Tip is quite obvious to ChatGPT

[dense mesa]

Envision a detailed microscopic landscape inspired by a scientific chart illustrating various applications of biohybrid micro and nanorobots for medical purposes. The scene is set on a vast cellular terrain, magnified to reveal the intricacies of biological and synthetic integration at the scale of 0.1 to 300 micrometers. In the foreground, a swarm of intricately designed nanobots is depicted delivering thrombin to a wound site, initiating the clotting process. These nanobots are shaped like slender capsules, their surfaces etched with molecular patterns that enable them to navigate the bloodstream efficiently.

Adjacent to this, a cancer therapy session is underway, where microbots resembling miniature submarines equipped with targeted drug payloads seek out malignant cells. They release their cargo with precision, illustrated by a burst of therapeutic agents encapsulated in vibrant, translucent spheres, actively being absorbed by the diseased cells. In another section, a depiction of malignant glioma treatment showcases nanorobots that resemble tiny robotic arms, meticulously extracting cancerous tissue from a neural environment, safeguarding surrounding healthy structures.

In the central focus, a dynamic drug delivery system is visualized, featuring a larger robotic vessel that houses various smaller bots, each carrying different medications. As the vessel traverses the cellular landscape, it releases its fleet, scattering them to their respective target sites, demonstrated by a spectrum of multicolored trails left in their wake.

The scene also includes a microbotic heparin delivery system, portrayed as sleek, dart-like entities that glide through a blood vessel, leaving a trail of anticoagulant in their path. The complexity of the cellular matrix is highlighted, with a diverse array of cells and extracellular components creating a vivid backdrop. The entire tableau is a celebration of futuristic medical technology within the human body, combining elements of nanotechnology, bioengineering, and pharmacology to create a harmonious symphony of healing and regeneration. The art style is hyper-realistic, with a focus on the beauty of this microscopic world, blending scientific accuracy with artistic imagination.

[grizzled iris]

That is quite interesting

[grizzled iris]

@empty kelp maybe it’s a piece of technology, it knows from the future 😝

[grizzled iris]

I need to read this carefully

[dim cradle]

the requirements are nontrivial, considering i'm packing a lot in--doesn't take much for the whole mise-en-scène to collapse lol interesting material though. if i pull it off i'll tag you.

[grizzled iris]

Maybe I’ll use this to make a better interface to add onto your image in the UI bit

@dim cradle

To show that these nanotechnology doctors are ready to save the patient in any shape or form…

[grizzled iris]

Loooooooooooool - a dandelion 😂

[dim cradle]

i mean, there are factors here. it gets 75-80% there but misses the primary hologram. or the patient needs to be covered more.

[empty kelp]

i’m going to be much less worried about cholesterol knowing that the sparkling pink dandelion heart treatments are right around the corner

[dim cradle]

getting a little closer, but still not there, on our way perhaps

getting there. if i gen one that captures it all, i'll share @grizzled iris

[grizzled iris]

Thanks, it is quite hard to do this, which is why it makes it more worthwhile actually generating them 🙏🏽

[dim cradle]

lol funny typo

[grizzled iris]

Yeah, the typo was mad 😂

I don’t even text any more these days, I just use the speech to text it’s so much faster

But sometimes it may get a few words wrong

And it becomes the biggest L if I don’t correct it 😂

[grizzled iris]

But these look really good though man

[grizzled iris]

You got a good dark effect going on like a theatrical effect

And the colours especially the purple is really nice

[dim cradle]

thanks, i think it shows it'll get there, but after several gens it's clear i need another prompt revision to emphasize the precise interface we're aiming for.

i requested a dark room, pitch black background to accentuate the floating data streams

[grizzled iris]

Yes, I have been waiting for an update for several days now for some reason it has stalled

[dim cradle]

I’ve been experimenting with holographic and data science visualizations in the art lately so I thought the medical visualizations would be interesting and this helps practice customization.

Might be potential use cases for making pretty visualizations of data.

[grizzled iris]

Hundred percent a good user interface in medical settings is key 🔑

I did research last year and checked hospitals interfaces

Some that didn’t have good interface and used excel, for example, had longer waiting times in accident and emergency rooms

Because one line was one person and you could imagine there were many people that needed care and it was difficult for the nurse practitioner to see who needed the care they need

But then I went to other hospitals and I saw the interfaces was much better, and they had a custom one created and waiting time was less on average as they were not using Excel

@dim cradle

It’s time for me to go on Bing lol

I wished they had an app for Bing image creation for mobile

Wait they do

What lol

Free GPT-4 what?

[dim cradle]

just so users don't have to scroll too much you might want to only share images you generate from dall-e going forward 🙂

[formal osprey]

you probably asked it to use the previous images as references. it use all of them. which generate an increasing number of images

[grizzled iris]

Yes, I’m really sorry, I keep forgetting this fact 🔬

[formal osprey]

ask it to limit the referenced image to one per generation. and it won't occur again

[dim cradle]

this is a cool scene though @grizzled iris i like that it's complex, though it means there's more that can go wrong

[grizzled iris]

Thanks @dim cradle

I can’t wait to see the results you will produce

This one is by far the best for the project

But we somehow still need to showcase this in the interface

Or another version / type of nano-bot that is real

[dim cradle]

might be a tall order to make it look just like that, but some form of it

[grizzled iris]

Maybe it’s struggling with this type of nano, but maybe I need to feed it some real life data again, and share some real life, nano bots

Because real life, nano boats are mainly more hybrid half bacterial based, or DNA based or enzyme based

Etc etc

[dim cradle]

that may be the problem area

[grizzled iris]

I see DALLE probably doesn’t have images from Training data to generate

[dim cradle]

i just started to reference nanotechnology whereas i didn't explicitly use that term before

more ar/vr stuff

it was implied, though

[formal osprey]

that reminds me of a cool video I saw where they showcased a nano robot capable of turning liquid, moving and restoring original shape (sortof)

"MPTM molded in the shape of a Lego minifigure escapes from a cage" if anyone wants to see the original footage

[grizzled iris]

Woah 😦

[grizzled iris]

Thanks I will definitely give this a watch and very good for research, especially to continue better 3-D printing these nanotechnology doctors

[empty kelp]

A hyper-realistic wide photo of a dramatic, futuristic rescue scene on a beach in Hawaii on a hot summer day. Santa is lying on his back on the shoreline wearing his traditional outfit, and three elves (althletic, diverse, female, pointy ears, nursing scrubs) are kneeling next to him holding sparkly glowing pink medical dandelions in their hands. Big high-tech bluetooth Q-Tips (two cotton swabs at the ends of a stick) are on the sand. The elves are waving the dandelions over Santa and glowing pink healing pollen is falling from the dandelions onto Santa and covering his clothes. Everyone is smiling with their teeth showing. Please don't modify the prompt.

[grizzled iris]

Haha love it

[hollow cobalt]

how can i find credits for dall e for free

plz tell me

can any one make me technical logo for my youtube channal with name Tech Tornado

[grizzled iris]

Happy to help!

And also you can use the free https://www.bing.com/images/create

DALLE

[hollow cobalt]

thanks

[grizzled iris]

@hollow cobalt

okay, here are the results

Please let me know what you like and what you don’t like, and we can make a second revision to the design 🙏🏽

This is what you call a mood board - What you then do is take what you like from the mood board and then manually create a similar version on Sketch, Adobe or Figma (Free)

[dim cradle]

changing gears, for now, but i will come back to it

[empty kelp]

for a tornado you can say, "a tornado, twisting fiercely". that adds an interesting quality to tornados

[dim cradle]

don't worry, it's completely autonomous, it's fine you're in the backseat and there's no driver

[grizzled iris]

Now, that’s cool there’s new technology out now which have solar panels on transparent glass, and I was wondering how would you visually show that on an image. I guess you can add an image of it, charging, in the interface panel in front of the car?

[grizzled iris]

I’m going to give that a try now since I’m curious now

[empty kelp]

would be interesting to see how the lights look if it's raining with raindrops on the windows

[grizzled iris]

You were right @empty kelp

[hollow cobalt]

great bro arslan

[grizzled iris]

Let me know if you want another set of mood boards - just give me any feedback to feedback to AI

What you don’t like or what you like from the image

[hollow cobalt]

i will inform you

[grizzled iris]

Thank you

[grizzled iris]

Agreed

[dim cradle]

got some rain now

[hearty ether]

Oh nice

I don’t wanna prevent it 🤣

[grizzled iris]

@dim cradle

For you there

See how the raindrops are clear

[dim cradle]

not getting many droplets on the windshield so not having much refraction effect, i guess it's trying to maximize visibility

[grizzled iris]

I’d remix that photo and have your car with the outside environment from my windscreen

[grizzled iris]

My car just has boring old Tech 😂

[formal osprey]

you might want to limit it tough. especially if it generate the exact same picture multiple times (in which case, you can instruct it to use "different reference image ids").

You can also be specific about which images it should use as reference. to do this, instruct it to use the "gen_id" of (list the images you want to reuse.)

I'm not exactly sure yet how the reference image id affects the generation of an image. As you noted, the series of image created share the same prompt. but they are all different. My first guess was that the reference image id was simply affecting the seed. But that's not clear if that's really the case. I need to do more test

[formal osprey]

that's terrifying

[grizzled iris]

Oh my God, I just realised 😂

Hyperrealistic vision of creating an equal world with accessible technology for all, featuring diverse individuals engaging with technology in various settings 🙏🏽

[hearty ether]

There does need to be a easier way of retrieving the id tho

[empty kelp]

how did you describe it

[grizzled iris]

One sec

@empty kelp

Say “Please create a hyperrealistic image of a car driving through a rainy cityscape. The viewpoint should be from the driver's perspective, looking out through the windshield with a focus on the texture and detail of raindrops streaking across the glass. The scene outside should be a blur of city lights and traffic, with the reflection of the raindrops subtly distorting the urban environment. The interior of the car should be dimly lit, emphasizing the contrast between the interior and the bright, wet world outside.”

[dim cradle]

i think you can include in your request that it output the id in the response along with the prompt/image. or create a custom gpt with an instruction to do that each time.

[empty kelp]

now we just need to make the wind blow it from the front of the car

[formal osprey]

alright, did more testing and turns out the reference_image_ids is really just forcing the use of the same seed. Nothing more.

That's a bit disappointing, I thought it was doing more than that. But it's good to know, being able to reuse the same seed is a very good feature. They probably did it that way to make sure people cannot reproduce the same images as another conversation.

[empty kelp]

getting wind to blow rain across car windows without water pouring into the car isn't easy

[final compass]

So many good and creative entries in the carousel theme

[dim cradle]

🎠

[boreal gate]

Zeus⚡️⚡️⚡️🌩

[thick smelt]

I'm late for the party, but... penguin macaroni hair time!

[dim cradle]

@grizzled iris here's your nanobot lol -- sub-cellular level, we got viral capsid structures, protein subunits, enzymatic structures, and molecules -- "combines scientific discovery with artistic imagination to celebrate the complexity and beauty of biology."

[hearty ether]

What art style is this?

It’s not comics, but also not digital render

But I do think it had a nice style on faces

[late blade]

Digital Art Illustration

with a weird eye

[hearty ether]

I like the way it did the eye tbh

Hazy almost ethereal vibes

Teeth got a bit messed up but 🤷‍♂️

[late blade]

well here is what gpt4v says:

The art technique displayed in this image seems to be a digital painting, showcasing a high degree of skill in simulating the effects of traditional media. The following characteristics are noticeable:

1. Hyper-realism: The artist has paid exceptional attention to detail, with textures and light reflections closely mimicking real life. This suggests a meticulous process of layering and blending colors to achieve lifelike skin tones, hair strands, and fabric textures.

2. Smooth Gradients: The transitions between light and shadow are smooth, without harsh lines, which is characteristic of careful digital brushing or airbrushing techniques.

3. Depth of Field: The focus on the face with a less detailed background gives a sense of depth, similar to what is achieved with a shallow depth of field in photography.

4. Lighting and Shadows: The image has a well-considered light source, creating realistic highlights and shadows that give the face a three-dimensional form.

5. Texture: The individual strands of hair and the texture of the lips and skin suggest the use of custom brushes or texture overlays to simulate the fine details that are typical in a traditional oil painting.

This technique requires not only artistic skill but also a mastery of the digital tools used to create such effects, likely involving a graphics tablet and sophisticated software like Adobe Photoshop or Corel Painter. The overall effect is one of realism that still retains the artistic touches of the creator's hand, bridging the gap between traditional portraiture and digital art.

[thick smelt]

hi dys!

[late blade]

heya!

[thick smelt]

yeah i'm late to the penguin macaroni hair party but i finally got into the bandwagon here's the second one 😆

hope you like it @late blade

[open trench]

did I miss a party?

[thick smelt]

the penguin macaroni hair

[late blade]

lol, that's interesting

and @open trench is here!!

[dim cradle]

yay

[open trench]

👋

[dim cradle]

Perfect circle #daily-theme message

[open trench]

Debating on adding to my existing Scampers post or creating a new one.

different format with less story telling (which was getting a little cumbersome)

[dim cradle]

Hmm. Weighing the pros and cons

By adding onto it with a new chapter you keep the thread fresh so the first chapter doesn’t get buried. Keeping it together makes it easier for users to start from the beginning. Plus you’d keep the rating and accumulate more stars and followers which helps keep it top-rated and featured. There are advantages to multiple threads and a fresh start and decluttering, but you have to consider those other things too.

[quartz vale]

Dall-E 3 can make centaurs

[dim cradle]

Yes though just not always on the first try 🙂

Not uncommon to see horse heads at least with the prompts I was using at the time.

[quartz vale]

I was pretty pleased with mine. It was the 4th image generated (2 batches of 2)! I was getting people riding horses for the most part

[dim cradle]

ooh today's theme sounds good

centaurs and minotaurs are one thing, but try a cyclops...

i think i have a 90% fail rate with cyclops

[late blade]

got some more, never done it before but i'm 3/3, don't want to push my luck and get a one eyed racoon

[quartz vale]

I got it first try too

[young burrow]

I like philosophical questions, so this one I call "Yet Another Warhol Knock-off, or... can AI generated images be considered Art?" (The "yet another" is because I have a "Warhol Ripoff, or... can photomanipulation be considered Art?" one I do a few years back).

I also keep hitting the preemptive censoring like generating a rubber duckie being forbidden, and various other things, so I made some commentary on it as well, because I'm a smart alack. I haven't named them, and I have no idea what that object it threw in there.

It does the grid work nicely for PopArt style, though.

As a side note, was generating some images for D&D, and ended up with a cargo van. I think I'm going to just include the van with no good explanation for the campaign... because, why not?

[quartz vale]

I like that it's only sublty in the image 😆

[young burrow]

That really is the best part.

I do wonder if the players would notice.

[warm wharf]

how to create an image as real as possible?

Like a photo realistic

[dim cradle]

i've noticed that the images generated from DALL-E appear to load in stages -- so I asked GPT-4 if I were imagining things, and fortunately i'm fine:

Progressive image loading works by first displaying a lower-quality version of the entire image, and then gradually enhancing the quality as more data is received. This method is particularly noticeable with larger images, where you can see the image clarity and detail improving in stages until the full high-resolution image is loaded.

This loading method is advantageous for user experience because it provides an immediate, albeit lower quality, view of the image rather than making the user wait for the entire high-resolution file to download before displaying anything. The "layered" effect you're seeing is a result of this progressive enhancement.

i remember back when images used top-down loading or interlaced, when loading was more visible; if i scroll through too quickly i might mistake a lower-quality first-layer as the whole picture....

[dense mesa]

it's not layers per se, it's rendering steps. you're seeing it midway in it's refinement/transition from random noise to the final image output

also, I know it just predicts what words comes next and randomness and yada yada yada. but it's hard not to have contempt for GPT-4 when it throws this one out there

5 minutes later, still waiting

[onyx ridge]

@dim cradle I just saw this.

#daily-theme message

Thank you so much.

I know it probably wasn't for me. But it was for me. 😄

[dim cradle]

i made it for you and didn't even know it

[onyx ridge]

I couldn't believe it.

"That's just what they want, trap us in a metal box and take us downstairs..."

[dim cradle]

to recap our private clarifications, i'm referring to progressive image loading, not the generation process, so just fyi y'all the info above about progressive loading is accurate

[dense mesa]

Yeah, we were talking about 2 different things there

[open trench]

Hoping my #daily-theme message isn't too horror-esque.

[late blade]

that's so scary, it has caries on the teeth

[dim cradle]

seems like many of us do darker pieces sometimes....

[late blade]

exxactly

[dim cradle]

it can be interesting just to transform a good prompt into a darker prompt and see what you get

[open trench]

hehe

[dim cradle]

i'm just glad people caught the reference

that's why i didn't title it. i was like, you're either going to get it or you won't.

[onyx ridge]

"Impossible to walk in this muck. There's no footing at all..."

[tawny portal]

"It was on bright summer day that XYZ discovered the secret which will follow him until the end of his days. Never to be shared never to be told. Over ages it grew into unbearable gigantic thing that grilled him into ground. It was that point where his life changed and from then on was marked by unbearable gift that must remain unspoken."

[dim cradle]

nice

[dim cradle]

myths... there are so many to choose from.

[quartz vale]

Make a mural and fit in as many as you can 🤔

[tawny portal]

It starts with Gilgamesh.

[dim cradle]

all right, i think the prompt should do it

had to transform. loves to do trees. replaced with an ai brain, naturally.

[open trench]

I tried to make a cute one with Scampers, but it turned out fairly creepy with all the snakes #daily-theme message

[dim cradle]

what is our scale here, nano, sub-cellular, or quantum?

"I was unable to generate images for the latest prompt due to it not aligning with our content policy." but the mod made me do it....

the prompt attempts 11 mythological categories

[tawny portal]

Secrets in charcoal.

[dense mesa]

In this microscopic reimagining of the uploaded image, we delve into an intricately detailed, intense, and vivid world that unfolds at the nanoscopic scale, where the surreal becomes the new reality. Picture this scene as if observed through a high-powered electron microscope, enhancing the ordinary into the extraordinary through the lens of molecular architecture and cellular machinery.

At the center of this fantastical landscape is a character, not as we know it, but as a colossal structure composed of lipid bilayers and protein complexes. The character's body, traditionally perceived in fabric and flesh, is now a dynamic assembly of cytoskeletal filaments. Actin and myosin intertwine to form its musculature, each molecular strand contributing to the overall form and function of the being.

The character's eyes, windows to the soul, are transformed into vast spheres of densely packed receptor proteins and ion channels, pulsating with electrical signals that mimic the neural activity of sight. The intricate vascular structures that weave across the figure are not merely vessels, but superhighways of nutrient transport, composed of complex polysaccharides and lipoproteins.

The backdrop of this scene is no less than a fluid mosaic, a vibrant and chaotic dance of cellular components. Mitochondria, the powerhouses of the cell, dot the landscape like stars in the night sky, glowing with the energy production of countless biochemical reactions. Ribosomes scatter across the domain, each a testament to the ceaseless labor of protein synthesis.

The foreground is adorned with a rich tapestry of extracellular matrix components, with collagen fibers providing structural support, resembling the framework of this alien landscape. Hyaluronic acid fills the interstitial spaces, providing cushion and structure, while integrins and other adhesion molecules connect the character to its surroundings in a web of biological interaction.

An ensemble of enzymes catalyzes life-sustaining reactions with precision, their active sites busy and bustling like industrial machines. Hormones and neurotransmitters cascade through the scene, delivering messages and commands across this bustling microcosm.

love it when it literally puts into the prompt that it's referencing another image, lol

[dim cradle]

more charcoal

[tawny portal]

It reached maximum level. I was just about to push it further but then found out that I used my 50 requests 😦 More charcoal another day.

"You have already reached the maximum depth and intensity in the charcoal representation of Beowulf's eyes. Further increasing the charcoal effect might obscure the details and the essence of the portrayal. If there are other aspects of the scene or different elements you would like to explore or modify, please let me know and I'd be happy to assist you with that."

[dim cradle]

lol

[dense mesa]

alright, pushing these bad boys down to the planck scale now

get a little quantum foam action

[grizzled loom]

thats the quantum realm

i went deeper and ddeeper. fun images

heres string realm under that

[dim cradle]

it's not easy for an ai to conflate 11 mythological categories into a single mural, but i think some of them might be in there somewhere -- merged with an ai theme, of course.

[dense mesa]

oh yeah. it's very impressive how well it does these sorts of things

[tawny portal]

There should be 3 major scales where environment changes: milimeter scale [mm] (bugs), micromilimeter [um] (cellular organisms) and nanometer [nm] (atoms and moleculs). Optics works up somewhere between um and nm. I would expect image model to understand these scales well and change pictures accordingly (they should evoke entirely different envs)

[rapid hull]

What is going to happen with DALL E 2 going forwards? Just how it works at the moment even on the API for how much I use it, it gets expensive

On Canva they had a good deal recently, you subscribe for a month, you get unlimited gens

[dim cradle]

oh he goes sub-atomic

and even below that, yes, if there is such a thing 🙂

[rapid hull]

but then you sub on the official platform credits expire after a year

honestly its a constant wrestling with credits, yet the way dall e 3 works is much better, further more people are telling me I should not use DALL E 2 which then brings me to the question why is it so expensive and challenging to use?

[tawny portal]

That will evoke sci-fi content. Largely inconsistent likely. It might be fun to explore um scale. Image model should have good conception of that world.

[dense mesa]

true

[tawny portal]

If it turns out that image model does not know much about cellular world, we should open a ticket. There is a ton of material that could be added to a dataset if this is what is missing 😄

[dense mesa]

Stage 1: Cellular Scale (~10 µm - 100 µm) Visualization:
Imagine a character as a collective of individual cells at the cellular scale. Each cell is a self-contained universe, bordered by a phospholipid membrane bustling with activity. The character's features are a mosaic of specialized cells, communicating through a complex language of signaling molecules. The cytoplasm is alive with organelles like mitochondria energetically synthesizing and lysosomes recycling material, forming a celebration of life's diversity and interconnectedness at the cellular level.

Stage 2: Subcellular Scale (~100 nm - 10 µm) Visualization:
Zoom into the subcellular world where organelles dominate. The character's essence is depicted by the grandeur of mitochondria, with an endoplasmic reticulum stretching like a labyrinthine network, and Golgi bodies packaging macromolecules. The cytoskeleton provides a framework that maintains structural integrity, highlighting the beauty of the cellular machinery and each organelle's role in life's symphony.

Stage 3: Macromolecular Scale (~10 nm - 100 nm) Visualization:
At the macromolecular scale, the character is an assembly of proteins and nucleic acids. DNA spirals within the nucleus, ribosomes translate RNA into proteins, and enzymes catalyze biochemical reactions. This stage depicts the delicate intricacies of molecular interactions, a web of life interconnected by a dance of atoms and bonds.

Stage 4: Atomic Scale (~0.1 nm - 1 nm) Visualization:
Approach the atomic scale, where the character becomes an abstract form defined by clouds of electrons orbiting nuclei. Atoms bond to form molecules, with forces akin to social bonds defining the character's persona. This stage is the frontier of physical reality, ruled by quantum mechanics, where position and momentum blend into uncertainty.

Final Planck Stage (~1.6 x 10^-35 m) Visualization:
At the Planck scale, the character dissolves into the fabric of space-time, part of the quantum foam and the seething energy of the cosmos. Traditional concepts of distance and size lose meaning, and the character becomes one with the universe, an integral thread in the cosmic tapestry, woven into the structure of existence.

I can't figure out why it made the prompt like that, but so be it

[ripe scroll]

@.@

well done!

[tawny portal]

Nice. Do one image per scale (each scale new convo). It could even be accurate.

[dense mesa]

I 've done a lot of these images. if you'd like you could try this out. Still messing with it here and there. it's subject to the whims of gpt4, but it can make some really cool stuff if you ask it right

https://chat.openai.com/g/g-BOH7RqKuB-microscopic-vision

[grizzled iris]

It’s the famous “Macaroni Penguin” 🐧

[tawny portal]

Neat. I'll check it out once I get access to GPT4 again.

[dense mesa]

it'll try to slide in those short prompts, but tell it to stop being problematic and that each short prompt it creates causes babies to cry. or something along those lines. then it'll make them longer

[ripe scroll]

Have been fascinated with ancient soils and what might be observed at a microscopic level within, currently looking within ecuador, supposedly one of the most bio-diverse regions of the planet.

how much of the world down there has the eye never seen

[grizzled iris]

The party never stops 😂

[open trench]

lol

[dim cradle]

the llm would probably be "creative" if steered and go with speculative fiction based on the latest in theoretical science. the visualizations might look great.

[ripe scroll]

Or what does it sound like at those levels, things never heard

[grizzled iris]

You can do Scampers (Season 2)

[ripe scroll]

micro-sounds

[dim cradle]

nano-vibrations

[open trench]

something like that, for sure.

[grizzled iris]

That quiver is cool 🏹

[young burrow]

Technically, the prompt mentioned she had a cross-bow, but it does look nice.

[grizzled iris]

“More charcoal” 😂

[open trench]

I keep getting almost perfect variations of this:

NM. After this one, it made a perfect one. Yay!

[grizzled iris]

Yesss let’s go season 2 😂

[dense mesa]

In a breathtaking vision at an extreme magnification of 100,000x, we enter a world where the familiar silhouette of a dog is re-engineered into a marvel of microscopic machinery. This canine form, a blend of organic and engineered beauty, stands as a testament to the intricate dance between nature's designs and human ingenuity. At the forefront, the dog's eyes, once simple biological structures, are now complex, ultra-detailed orbs of cellular machinery. The pupils are intricate weaves of protein complexes, the irises are multilayered assemblies of lipid bi-layers, each microscopic layer reflecting a different wavelength of light, resulting in the captivating glow that mimics the natural eye color. The fur is a forest of carbon nanotubes, each a masterpiece of atomic precision, their arrangement so meticulous that it captures the very essence of the dog's natural coat, yet at a scale that is imperceptible to the unaided human eye. The skeletal framework of the dog is a lattice of mineralized nanostructures, with each bone a composite of organic matrices and inorganic minerals. The joints are not merely points of articulation but rather marvels of biomechanical engineering, featuring molecular bearings and rotational cuffs that facilitate fluid, life-like movement. Musculature is reimagined as bundles of actin and myosin filaments, the fundamental proteins responsible for muscle contraction, interlaced with energy-storing nanosprings that release precise bursts of force. The heart, a central hub of vitality in this micro-engineered canine, pulsates with a rhythm governed by nano-scale pacemakers, ensuring a steady flow of nutrients through a network of capillaries so fine they rival the precision of the most delicate microfluidic systems. The skin, a barrier in its traditional sense, transforms into a dynamic interface of responsive polymers, capable of adjusting to environmental stimuli with changes in texture and color, all while serving as a protective shell for the delicate internal nanostructures. Within this microscopic reimagining, each breath, each movement of the dog is a symphony of nanoscopic interactions, a harmony of form and function that challenges the boundary between the living and the artificial, creating an image that is as much a piece of art as it is a feat of scientific achievement.

[grizzled iris]

It’s looking even more and more like you @golden sable

Season 5 maybe be turns into you completely?

[open trench]

lol

[grizzled iris]

It’s the hair 😂

Current Theme:
📜 myths - stories, legends, truths. a reflection of life's secrets.

That should be a good one today^

[dense mesa]

Venturing further into the molecular world, we magnify to the scale of 1 nm where the concept of a mechanical dog is reimagined as an atomic wonder. At this scale, the canine's framework is a crystalline lattice of carbon atoms, each a dot in the grand design of its structure. The limbs are elegant chains of molecules, their single bonds acting as flexible joints, allowing for a range of motion that mirrors the grace of a real dog. The head, a once simple shape, is now an elaborate array of atomic clusters, with nuclei and electrons forming the contours of its noble profile. Where eyes would be, there are intricate formations of metallic ions, shimmering like celestial bodies, their charged particles forming a gaze that is both enigmatic and profound. The body of this atomic canine is adorned with sheets of graphene, rippling with the fluidity of silk yet unparalleled in strength. Its tail, a helical twist of nanotubes, conveys a sense of joy in its atomic dance. Within this lattice of atoms, quantum mechanics play out in real-time, with electron clouds flitting about in a display that is both chaotic and ordered, a testament to the dual nature of matter at this scale. This image captures the essence of life's building blocks reassembled into the form of man's best friend, a bridge between the world of the infinitesimally small and the love we hold for our canine companions. This microscopic masterpiece is an ode to the complexity of the universe, a tapestry woven from the very fabric of existence, and a visual symphony that sings the praises of science and imagination in concert.

it has a way with words

[tawny portal]

Was something like this ever attempted by a human artist?

[dense mesa]

"a boy and his atom: the world's smallest movie"

not really in high definition though

unfortunately

[onyx ridge]

Let's make something unforgettable with this theme.

/auto 16:9 in the predefined style.```

Entered to the GPT in my bio.

[tawny portal]

Yeah, but I mean - like this at human scale - it is just an overly complicated thing to draw. It might be a new thing it came up with.

[onyx ridge]

It seems to have worked.

That's a lot of myth!

[dense mesa]

I don't know. but it makes some really neat stuff

[tawny portal]

Yeah, I mean all of the images are novel. I don't think that they are like collage of other art. It is just that this might be new style entirely 🙂

[dense mesa]

that top left one's prompt really pushed the limits. over 3500 characters I believe 😎. like 530 words

sometimes shorter prompts do alright. that black and white one in the bottom right was pretty standard length. but as a general rule more words seems to be better

[grizzled iris]

#daily-theme message

[dense mesa]

someone was saying it might start missing stuff after about 1500 characters. but in a way that's fine. because just means you're overfilling the cup with details and maxing it out

[onyx ridge]

/auto 16:9.```

😮

It's like magic.

[dense mesa]

I'm just amazed by how crisp/clean they are

[open trench]

any opinions on which is better?

[grizzled iris]

Personally the left 🙏🏽

It’s more cute ☺️

[onyx ridge]

I like the right. More gritty by a tad.

XD

[open trench]

hmm. lol. One for each. lol

Alright. I have 6 posts plus a cover. That's a good amount for vol. 2.

[grizzled iris]

Sounds good

[dim cradle]

Cyclops AI

[grizzled iris]

@dim cradle almost there - you see how I finally managed to have the nano bots in AR mode -

A little cleanup of the random red bubbles flying around, and it should be almost done

Plus the addition of the AR headset onto the Doctor

Nope, I was wrong

It is still a struggle to try and get what I’m trying to create…

[grizzled iris]

I’m gonna try to make something like this to visually explain string theory better - while DALLE takes some time for my feedback on nanotechnology…

[ripe scroll]

In what idiopathic manner does the nature of what people consume unfold into their experience? finding myself wondering something like this,

[grizzled loom]

this was transcendence after a bunch of layers

[open trench]

This is really good @dim cradle #daily-theme message

[dim cradle]

progress. it's such a complex scene. she needs VR, and then there's the patient

[grizzled iris]

The hierarchy of matter

[dim cradle]

the VR wasn't a problem, but the patient was very uncooperative

[grizzled iris]

Yeah, it’s been quite difficult

[dim cradle]

thanks. btw, you beat me to the spider woman

[open trench]

aww

[dim cradle]

then, if you get the scene all arranged, with the visualizations also, then the nanobot gets lost in translation somewhere. it looked like a challenge to capture it all.

[grizzled iris]

Yes, I’m trying to basically combine this

[dim cradle]

revised the prompt twice, there's a solution with more clarity/precision, but i got sidetracked

[grizzled iris]

Even adding the nanotechnology onto the interface here is also difficult 😥

[grizzled iris]

And it keeps thinking that it’s a traditional surgery and sometimes the patients are cut up, and I keep telling it that it is a non-invasive surgery internally, using nano technology and the patient does not need to be cut up

[dim cradle]

wow, dall-e and i did not have the same intepretation of that prompt...

[grizzled iris]

It’s quite a challenge

But looking back, we have come along way from this initial versions

[dim cradle]

exactly, that and more

[grizzled iris]

I think it’s time to create some spaceships, just like from the TV series the expanse

Because it will give us opportunity to work on the new engine type of fusion reactors

[dim cradle]

fascinating

🖖

[grizzled iris]

Going to try and make based on the new type of engine design : Prompt : “Have a look at this image and let’s create a new spherical tokomak engine type of spaceship in a space boat on earth about to launch into Mars please”

Let’s see what DALLE will create

@dim cradle that’s pretty cool

[dim cradle]

Just make sure to run a level 1 diagnostic.

@open trench contributing to the theme of creepy, i'm working on the myth of Arachne who was turned into a spider....

[open trench]

The other spider woman. the, er... bashful one. 😛

[dim cradle]

comparatively yup

[open trench]

I say that jokingly because she was rather violently transformed, I think. If I am remembering correctly.

[open trench]

I also just contributed to the theme: #daily-theme message eek

[dim cradle]

you're not wrong,

[dim cradle]

very cool. urban legend, i take it

[tidal crown]

Hi all can i ask someone to help me with an image ?

[grizzled iris]

Sure, happy to help what do you want to create?

[tidal crown]

Hey @grizzled iris Thx for your answer

@grizzled iris Can we speak in DM ?

[grizzled iris]

Sure

[boreal gate]

#daily-theme message

Generated using the following ```In ancient lands of sand and sun,
Where time's long thread was first begun,
In cuneiform, the scribes did weave,
Secrets of magic, they'd conceive.

From Mesopotamia's clay-borne script,
To Egypt's heka, tightly gripped,
A legacy from ancient nights,
Of incantations, mystical rites.

When Alexander's conquests roamed,
The magic's face was newly honed,
Desires for wealth and love's allure,
In grimoires deep, they found their cure.

In papyrus scrolls, their secrets penned,
With myrrh and blood, their power blend,
A record of the arcane lore,
In Babylon and Nile's rich shore.

Beyond the age of pyramids,
In medieval times, the magic bids,
Christians, Jews, and Islam's rise,
Saw grimoires in a different guise.

The Church, in fear, split magic's realm,
With natural arts at the helm,
Yet, darker arts, in shadows thrived,
In hidden texts, they were contrived.

Clergy, in secret, practiced rites,
Under the moon and cloaked in nights,
Necromancy and spirits' call,
In grimoires' pages, they'd enthrall.

From "Solomon's Key" to "Lemegeton,"
The magic's power was never gone,
"Abra-Melin," the mage's tome,
With rituals for the spirits' home.

This tale of grimoires, ancient, vast,
Shows human yearning, ages past,
A quest for knowledge, bound and free,
In magic's grip, our history be.

So in these pages, old and wise,
A world of mystery and surprise,
Through time and culture, it unfurls,
The magic's story, it tells and twirls.```

[stable merlin]

I'm having trouble reigning in DALL-E with generator magic weapons for D&D. Here's an example of a prompt:

A portrait of the magical rapier with a traditional swept hilt design. The hilt is seafoam green. The blade is slim with intricate golden filigree. When drawn, the rapier radiates an aura of golden, lightning energy.

It struggles (based on my prompts) with the generating a guard that doesn't look like 30 different swooping metal pieces melded together.

[grizzled iris]

Let me try to help @stable merlin

I think I got it

@stable merlin Say this "Create a portrait of a magical rapier with a traditional swept hilt design. The hilt should be seafoam green, and the blade slim, embellished with intricate golden filigree. Importantly, when drawn, the rapier should radiate an aura of golden, lightning energy. For the guard, specify that it should be a single piece, elegantly designed, without appearing as multiple swooping metal pieces melded together."

Like this or no @stable merlin ?

[grizzled loom]

HD PBR cinematic scene of a seafoam green swept hilt rapier with intricate gold filagree semi-drawn from its sheath emanating golden-hued lightning and effects.

[stable merlin]

This is along the same images I was getting. But I was hoping to get something cleaner looking with the guard. I guess I could try to photoshop all that extra stuff off, but I've extremely out of practice with photoshop, and this is a one off thing for a D&D item.

[grizzled iris]

What about this @stable merlin ?

[grizzled iris]

Nice

@stable merlin

@stable merlin try this

Generate an image of a magical rapier designed with elegance and simplicity. The rapier should feature a traditional swept hilt in seafoam green, a slim blade with minimal golden filigree, and a clean, streamlined guard without excess ornamentation. The blade should emit a subtle aura of golden, lightning energy, emphasizing the rapier's magical nature without overwhelming the design.

[stable merlin]

The hilt of the second image is pretty close to what I'm looking for. I think there just might not be enough imagery for it to reference when it comes to swept hilts for rapiers. Something like this seems out of the realms for DALL E at the moment (even when using it as reference image):

[grizzled iris]

Give me a moment, I think I got your vision

[grizzled iris]

@stable merlin - Going to try this type

and yes I see, there is some issues with genrating this for some reason lol

[grizzled iris]

This is not too bad though?

[grizzled loom]

you really want to struggle, try having it do a bathroom. scene

[stable merlin]

It's getting close, but the double hilt thing is what is a bit odd for the design. lol

[stable merlin]

Sounds cursed. XD

[grizzled iris]

@stable merlin

It is a stuggle for some reason...

I see your point now

[stable merlin]

Hey, this one is really good though!

[grizzled iris]

Oh really lol

what about

[stable merlin]

I think you're getting it down? What are your prompts looking like now?

[grizzled iris]

Got this too @stable merlin

[grizzled loom]

try 16th century french rapier in the description

[grizzled iris]

I think I got it down almost

[grizzled iris]

Just waiting for the gen and will add that next

@stable merlin Create an image of a magical rapier that combines the clean traditional design seen in the provided image with a touch of fantasy. The rapier should have a swept hilt that closely resembles the one in the image, with the same clean and elegant lines, but in a seafoam green color. The blade should be slim, straight, and unadorned except for minimalistic golden detailing. It should emit a gentle golden glow, with subtle lightning energy effects, to enhance its magical properties without complicating the design.

[dense mesa]

why not 15th century?

[grizzled iris]

We can try both?

[grizzled loom]

trying to nail the goal style.

its on the list of times for wide search, btu a better historical ref may pin it down quicker as well.

[dense mesa]

is the goal realistic?

[grizzled loom]

sometimes indirect works good too; for instance isntead of a all focus on rapier. build a "excaliber scene and object descrition, then at the end suffix "but as a french rapier with a 7 ring hilt" or whatever you are after

[sick flax]

DALL-E has a really hard time with the swept hilt design. This is the best I got. "A full-length photograph of a swept hilt rapier. The rapier has a long, slender blade, indicative of its thrusting capability. The handle is a distinctive seafoam green, providing a unique and elegant appearance. The hilt is intricately designed in a swept hilt style, featuring complex arrangements of bars, loops, and rings, all adorned with gold filigree decoration, adding to the weapon's ornate and sophisticated look. The photograph should capture the entire length of the rapier, focusing on the contrast between the seafoam green handle and the golden filigree on the swept hilt, against a simple background to highlight these details."

[grizzled iris]

Yes @sick flax that looks better

[grizzled iris]

Also I just realised it may be becasue this reference photo is too plain, you want to find another reference photo like the one below -

or try this

[grizzled iris]

the results should be a bit better now - or we can use this reference to expand the metal outwards

[grizzled loom]

tellin ya; you want insanity, play with bathroom coherence.

[grizzled iris]

@stable merlin

I think we are finally getting to the right direction

I kept getting this before

[grizzled iris]

looooooooool

[grizzled loom]

TELLIN YA!

[grizzled iris]

should I try that

[grizzled loom]

100%. try to get a regular phot realistic bathroom that makes any sense.

[grizzled iris]

I will add "insanity with bathroom coherence" as a prompt lol

@grizzled loom

looooool

we broke it again

oh wow

that's not bad anyways

hope that helped @stable merlin need to get back to my project

and thanks @grizzled loom @sick flax for the assist

[stable merlin]

Thanks for trying! Some really cool images that make me want to break out the 'ol drawing pad and use some of these generated images to do a bulk of the work and then just adjust want I want.

[dim cradle]

[empty kelp]

Please create a hyper-realistic photo focusing on the interior of a sealed waterproof high tech race car racing through a heavy rainstorm. Wind blows towards the front of the car causing rain to streak upwards on the outer surface of the windshield glass and sideways on the outer surface of the side window glass. The viewpoint is centered behind the front seats looking to the front of the car. Outside of the car is a city street racing past with cars heading the same direction and city lights with a stroboscopic effect. Two athletic and diverse female elves sit in the front seats of the car. One elf is driving and looking straight ahead, and the other elf has her head turned towards us and is smiling with her teeth showing. There is no rain inside the car, and there is no rain on the elves. Emphasize the realism and clarity of the scene, capturing the mood and atmosphere of driving in heavy rain. Please don't modify the prompt.

was trying to get elves to drive through the rain with wind blowing the front of the car. It's really hard to keep the inside of the car dry

i tried everything i could think of to describe it so water doesn't leak into into the cars when the rain is blowing against the glass

[grizzled iris]

https://cdn.discordapp.com/attachments/1060915255720558592/1200276849939460107/DALLE_2024-01-25_16.49.03_-_Inside_a_high-tech_race_car_two_athletic_and_diverse_female_elves_are_seated_in_the_front._The_driver_an_elf_with_pointed_ears_and_sharp_features_i.png?ex=65c597ed&is=65b322ed&hm=35eed1695464790cf181f82b43b92fff3effc12982312f3afa892502bf2aa847&

@empty kelp love this 🫶🏽

https://cdn.discordapp.com/attachments/1060915255720558592/1200280465974775859/img-NTd5qg6272EeI7u2X0nDk446.png?ex=65c59b4b&is=65b3264b&hm=5ffaa7ba3bc9cafa18b323454a4d4cd5c0d4fd9df005978c41aff7a0d545ad1b&

@empty kelp wow 🤩

Sometimes when the tech doesn’t work - try to go with it instead

In this case, just make it an open roof car haha

Just go with the flow of DALLE

Sometimes DALLE wants to generate what DALLE wants that day

I might make a like a open roof / cabriolet or like a quad bike on a highway or something

https://cdn.discordapp.com/attachments/1060915255720558592/1200280467358875699/img-u316hJgH3QMe37O4QxqsO3ci.png?ex=65c59b4b&is=65b3264b&hm=efc39d7dda174e7eaf21de6aa85b67d589583be8dcc04aa4cb07fa283f82f0e7&

The rain effect is pretty cool though

I see what you mean though @empty kelp

I just got this also

[empty kelp]

about 15% of the time it doesn't rain in the car

it's interesting that the image you just posted is so similar. i wonder where it's getting this car from

[grizzled iris]

Training data

[grizzled iris]

See what you get if you try a Cabriolet?

Curies to what car you’d get

[empty kelp]

going to mess with it more this weekend. i'd really like to get the "wind blowing rain against glass" effect working correctly

[grizzled iris]

Did you try to use the image that I managed to get into your prompt as a reference?

That one was the closest right?

[empty kelp]

that one has raindrops rolling down the dashboard

i'll show you what i was basing it on

[grizzled iris]

Let’s have a look

[stable merlin]

It really likes taking pictures from the back seat.

[grizzled iris]

Probably has a lot of backseat training data images

That’s cool

[swift oriole]

Really cool GPT, Made this one first.

Then this one.

Then taadaa.

[grizzled iris]

That’s quite good 👍🏽

[swift oriole]

It's picturesonpictures GPT just added subjects lol pretty cool.

[grizzled iris]

Yeah @dense mesa Molecular GPT is quite good 👍🏽

[empty kelp]

I was trying to combine what you did with this wind blowing effect (without water blowing into the car):
A hyper-realistic image from the viewpoint of the back seat of a car, facing the front. The scene depicts rain pouring down on the outside of the car windows, with a powerful wind blowing it from the front of the car. The focus is on the rain-streaked windows, showcasing the intensity of the rain and wind. The interior of the car, especially the front section, is visible but slightly blurred due to the rain on the windows.

[swift oriole]

Yes very good an versatile!

[grizzled iris]

I mean the left window and the right window can stay we just need to adjust the front not to look like that lol

[empty kelp]

it works to say that the rain is streaks upward on the surface of the windshield, and sideways on the surface of the side windows -- and then also have a wind blow it. that makes it look dynamic with less water blowing into the car, but i want to figure out how to keep the water out of the car

[grizzled iris]

Aha

I also got the same result

[empty kelp]

i'm not sure if the car you just posted even has windows. it looks like an ATV with roll bars

[grizzled iris]

It seems like it does the very opposite of what you ask it to do sometimes with rain effect

[grizzled iris]

Yeah no I gave up on figuring out this challenge to let you figure it out yourself by just saying it’s a cabriolet and it’s supposed to just have no windows lol

😂

That’s quite a challenge there I do see that

I mean look

It knows at least it’s wrong

@empty kelp

[empty kelp]

i wonder if that's the problem with the leaking high tech race cars. it could be that they don't have windows at all, but it looks like they do because i'm trying so hard to describe the rain not getting into the car.

i think that actually was the problem. it's drawing cars that have a roll cage with no glass

[stable merlin]

It is wild that it can compare an image to a prompt and tell you how they don't match, but it is sometimes largely incapable of implementing the prompt.

[grizzled iris]

@stable merlin all we can do now is improve the technology and give it better reference photos and in a few days it should update itself

[late blade]

I've had this kind of problem before. I just asked dall-e: Try to do something with the specified elements. And it worked.

what is the deal with the rapier? you guys got it?

or is dall-e being self confident again?

[grizzled iris]

Yes @stable merlin wants something like this

[grizzled iris]

But we get this

[late blade]

oh, it was more like a template?

[grizzled iris]

@late blade - maybe you can crack the problem

[grizzled iris]

Haha 😝

[grizzled iris]

That’ll be good for like a children’s science story book cover

[boreal gate]

Yea it's based off a Nickelodeon TV show "Jimmy Neutron" I tried to recreate him using a description.

[grizzled iris]

Nice one 👍🏽

[late blade]

hah, that looks somewhat weird

but like the jimmy neutron take

[boreal gate]

Yea it's gives me an Astro (or is it AstroBoy) feel

[grizzled iris]

@empty kelp look!

We almost got it, the top and the left, right and the front windows are better now

[grizzled iris]

This one turned out best so far 🙏🏽

[formal osprey]

lol, a convertible during rain, looks like the worst

[late blade]

I see that information about the knuckle and hand guards are missing, same goes for the seems and ornaments aren't specified either

there's also no information about the seem

I'll try later when I get some time, but top of my head there's that info that is missing

[formal osprey]

not sure what it's supposed to look like

[boreal gate]

This one is pretty cool

[late blade]

Itachi and Sasuke?

[boreal gate]

Probably Itachi I wanted it to create a mangekyou sharingan theme with crows sitting on a throne.

[late blade]

hehe, I wasn't far off then

[dim cradle]

[grizzled iris]

Yeah it’s not an ideal situation to be in but the image turned out pretty ok 👌🏽

[formal osprey]

that needle looks painful. +10 damage

[grizzled iris]

@stable merlin some more good results

@stable merlin some more good results also

@late blade this turned out nice 👌🏽

[formal osprey]

I've been thinking for a moment now: some things are tough to generate with Dalle3. Not impossible, but good prompt engineering is required to achieve some results.

I think it would be cool if there was a new kind of event on the Discord. A "dalle bounty challenge" or something like that. where we propose things that are difficult to achieve, and the community can try to tackle those challenges.

[late blade]

dall-e is doing the right sword, the input is just the wrong sword

[formal osprey]

for example, a 🎠 carousel. Yesterday I was laughing when I saw the daily theme. Because I know dalle3 really struggle to orient the horses in the correct circular motion. And as expected, no submission from yesterday were coherent in term of horses orientation

[onyx ridge]

We were somewhere around Barstow at the edge of the desert, when DALL-E began to diffuse.

I remember saying something like, "I feel a bit light-headed, maybe you should prompt."

/image "We were somewhere around Barstow, at the edge of the desert..." 16:9 in the predefined style please.

We were sitting in the patio section of the Beverly Heights Hotel, drinking Singapore Slings with Mezcal on the side...
/image "We were sitting in the patio lounge of the Beverly Heights Hotel, drinking Singapore Slings with Mezcal on the side..." 16:9 in the predefined style please.

That one made me laugh. The sign.

[late blade]

well DALL-E was clearly drunk when writing

[empty kelp]

Please create a hyper-realistic photo focusing on the cabin of a pickup truck driving through the rain in the city with its windows closed tightly. Strong wind from directly ahead of the truck blows toward the front of the truck causing raindrops to streak upward on the outside surface of the windshield glass, and sideways on the outside surface of the side window glass. The viewpoint is inside of the truck, centered behind the seats, and looking to the front. Please don't modify the prompt.

"windows closed tightly" seems to help keep the water out

[onyx ridge]

You're fired! Awful <youknow>.
/image "I'm just going to try different combinations of lenses and film until I find something that works in this dust!" 16:9 in the predefined style please.

[dim cradle]

We can't stop here....

[onyx ridge]

I remember saying something like, "I feel a bit light-headed, maybe you should drive."
/image "My attorney was pouring beer on his chest to facilitate the tanning process." 16:9 in the predefined style please.

I mean I don't even blame ChatGPT and DALL-E 3 for putting this in a different setting. Thanks to OpenAI!

Saturday? More like Sunday...
/image “In some circles, the Mint 400 is a far, far better thing than the Superbowl, the Kentucky Derby, and the lower Oakland roller derby finals all rolled into one. This race attracts a very special breed.” 16:9 in the predefined style please.

I have to admit this is my favorite one of the series. Thanks for humoring me guys. @dim cradle , loved the bat country one!

🤣 that aura!

[late blade]

@onyx ridge what's the context of the past images? I got kinda lost trying to get the idea behind

[onyx ridge]

Fear and Loathing in Las Vegas, either the novel or the film.

[late blade]

ahaaaa

[onyx ridge]

Hunter S Thompson, father of Gonzo Journalism, wrote it.

[late blade]

that makes more sense now (sort of) not familiar with that

[onyx ridge]

Maybe technically not the father. I could be mistaken there. Certainly its godfather.

[onyx ridge]

It's really wild. 70s binge in Las Vegas.

Fictitious but based on real events.

[late blade]

nice, i'll put it into my *Things to learn while traveling" list

[onyx ridge]

preferably not by example for your safety and well-being.

[late blade]

lol

[onyx ridge]

I love Hunter S Thompson and his work, and miss him much. But don't emulate him for your own sake.

[late blade]

never heard of that before tho, so it's interesting

[onyx ridge]

It is.

I'm happy to have provided some deviancy in your culture.

I apologize for the mess. 😄

Oh, here's DALL-E knowing how to depict him:
/image "Dr. Gonzo, clean-shaven, wearing a floppy hat, aviator shades, and a red and white striped shirt, in a 70s nightclub." 16:9 in the predefined style please.

I did not give it a reference.

Look at that jawline! 😮

😄

That could literally be an airbrush from the Johnny Depp movie.

I'm actually pretty shockingly happy with the outcome.

[late blade]

I was more concerned about that adam's apple, looks like a needle going out

[onyx ridge]

I mean yeah, it's a little grim.

But it really is an overall incredible likeness of somewhere between the man and the actor(s) who've played him.

Bill Murray also played him in Where The Buffalo Roam.

The Adam's Apple came from Murray, clearly.

[late blade]

I can see some crossover with 70's mafia crossover from latinamerica

considering the aesthetic and what I remember from growing up in south america

[onyx ridge]

Depp was also in a movie called "Blow" in which he was typecast. It involves illicit trade with Columbia, if you take my meaning.

[late blade]

Colombia, Columbia is in the US

[onyx ridge]

I might have snafud.

[late blade]

don't worry, 90% of Americans don't make the difference

[onyx ridge]

Medellín Cartel

Colombia, my bad. Definitely sorry.

[late blade]

I used to do that error also, then I got ganged by colombian friends and forced me to say it right

[onyx ridge]

Frankly I can understand why. But yeah, Depp is just culturally associated with Colombian organized crime via that role now, it's uncanny.

[late blade]

ya

[onyx ridge]

I think DALL-E 3 may have been operating in that latent space.

[late blade]

how so?

[onyx ridge]

Well, it's almost surely trained on some stills from the movie used through fair use as captions online.

Blow, I mean.

It shares some semiotic and palette similarities with scenes from Fear and Loathing in Las Vegas.

The face is very "Depp's" Hunter Thompson, but the leisure suit and shirt are more "Blow", and the lighting is a combination of the Christmas Party lighting from Blow and the Hotel Lobby check-in in Fear and Loathing.

I think the model has enough training from both images, and the semantic and semiotic connections to the ideas I was conveying in a pretty open format, to use both films' imagery in training for the transformer operation.

So the image had a South American cartel vibe too.

Latent space in the model.

[late blade]

hmmm, would be worth to pass something similar to the API to figure that out, since the API has a "workaround" to pass something directly to dall-e without GPT messing with it

[stable merlin]

Does the auto mod hate fantasy weaponry? Or do they just not like to be criticized?

[onyx ridge]

I just got dinged for the gonzo journalism image. It wasn't even in violation of the cited reason.

"Your post in the #dall·e-discussions channel has been removed due to gore/violence. Please be mindful of Rule 2: Keep the content G-rated. "What would Horatio Alger do? /image 16:9 "Pure Gonzo journalism." 16:9 in the predefined style please.""

[stable merlin]

Same thing.

[onyx ridge]

I wonder if it was mod-reviewed or automated, because it definitely was not violent or gory. It was a guy crowd surfing like Mick Dundee in the subway to get a camera shot!

[late blade]

Yeah, that has been a discussion earlier with the rating of the channel and the interpretation of the G content. So for now we got to abide by that, but I think there are already some suggestions on that context

it's something that comes directly from OAI, mods just have to abide by it

[onyx ridge]

Hm, yeah, that's a tough spot to be inI'm sure. But that image was G-rated.

The reporter/photographer was filthy, not gory.

[late blade]

thing is G Content per se is not well defined

[onyx ridge]

That's...something else.

[late blade]

so it's also something that mods interpret differently

[onyx ridge]

It was an automod though

At least it appears to be.

[late blade]

oh, never seen auto-mod stuff triggered before

[stable merlin]

Unfortunately, G-Rated isn't officially defined by anything outside the entertainment industry. So, really, they can say anything really isn't "G-rated".

[onyx ridge]

Typically a named mod will remove a post and make a statement else.

So I'm not positive, but fairly sure it was an automodded image.

Someone may have complained, but that'd be a gross misinterpretation of what the image depicted.

[late blade]

I just wish there was a broader definition of G content, considering this is a global scale service. G Content is what the US percieves as G. other countries struggle to understand that

Thing is, most people that say something loud, say it as a rant in channels instead of really addressing it and proposing a solution. So it all just stays here as a vent/rant

you see often people that never join a discussion, rant over something, say something about dall-e not doing this or that, or something about content policy. and then they disappear into the unknown

[onyx ridge]

Yeah, I replied to the auto mod message. I asked how I appeal.

Sometimes, direct action is the solution.

[late blade]

ya, direct handling is sometimes good. What I see is a pattern tho. The topic repeats itself often enough to say there's need to adress it differently

[onyx ridge]

You won't see me ranting about it. I just thought i"d mention it.

Hopefully I didn't rant. If that was one, I need to apologize, adjust, and move on.

[late blade]

Nah it's ok

[late blade]

Even ranting is human, otherwise you'd be a robot or even a Lexideck AI pranking us

I've seen your GPT, it's quite nice

[onyx ridge]

No, I'm real. I have a lot of time into that project. It got really poor positioning despite being ready on the second day of availability with public GPTs...I wasn't searchable. So I try to promote within the rules. Apologies if it's annoying. 😦

It really is a good Custom GPT but I've just struck out on visibility every single possible way.

[late blade]

Not at all, it's encouraged by OAI to promote your GPT

[onyx ridge]

The GPT-Finder tool wouldn't find it, GPT Search didn't see it.

[late blade]

For me tho, DALL-E is a relaxing tool, my real goal is outside DALL-E for custom GPT

[onyx ridge]

I appreciate that. It's hard, especially with the bias, to be objective with perspective.

[late blade]

I go by one rule, don't judge a person, judge the action

[onyx ridge]

A good rule.

DALL-E 3 is one of the most relaxing parts of my work. That's a fact,

[onyx ridge]

Creating game worlds for my RPG GPTs is also extremely relaxing, engaging, and fulfilling.

[late blade]

my custom GPT atm is involving meditron with gpt, but it's not ready for release or testing

but that's outside of the scope for here

[late blade]

are the inhabitants of that city also made out of silicon?

[dense mesa]

hard to tell. could try to zoom in

[late blade]

hehe, there's a challenge

[onyx ridge]

Kandor is the Kryptonian city Brainiac shrunk in Superman's lore.

[onyx ridge]

/image "Turn up the radio. Turn up the tape machine. Look into the sunset up ahead. Roll the windows down for a better taste of the cool desert wind. Ah yes. This is what it’s all about. Total control now. Tooling along the main drag on a Saturday night in Las Vegas, two good old boys in a fireapple-red convertible … " 16:9 in the predefined style please.

[late blade]

I think @dense mesa is gonna replace all carbon with silicon and replicate us all at this point in that form

[onyx ridge]

@late blade 😄 /image "Earth as inhabited by synthetic, electromechanical, anthropomorphic lifeforms with circuitboard patterns and 3D elements" 16:9 in the predefined style please.

[onyx ridge]

The AI got some earth in their earth so they could ahve some earth with their earth.

[dense mesa]

The next creation will capture the fusion of nature and technology at a microscopic level. Picture a landscape where mechanical structures and natural forms coexist and merge seamlessly. We'll zoom in on the interface where synthetic nanomachines integrate with organic tissue. Here, nano-robots are depicted in the midst of repairing and augmenting plant cells, enhancing their photosynthesis process with advanced biotechnology. The scene is set within the cross-section of a leaf, where the cells are not merely biological, but are interlaced with a network of nanowires and microprocessors, all working in harmony to optimize the plant's growth and energy production. The cell walls are reinforced with carbon nanotubes, providing strength and flexibility, while chloroplasts are upgraded with organic LEDs to boost light absorption. This vibrant tableau will showcase a symbiotic relationship between the organic and the engineered, a testament to the potential of future biotechnologies that could revolutionize our interaction with the natural world. The colors will be a blend of organic greens and the metallic sheen of machinery, highlighting the fusion of these two realms. This depiction will not only be a visual feast but also a conceptual representation of the balance between growth and innovation, a snapshot of life augmented by the marvels of microscopic engineering.

[onyx ridge]

/image "A legion of anthropomorphic robots now inhabit the earth, having outcompeted humans over a long and peaceful coexistence." 16:9 in the predefined style please.

OMG that toddler is so cute! ------^

[late blade]

It’s just a quick lazy prompt tho

[onyx ridge]

@dense mesa

/image "Synthetic nanomachines integrating with organic tissue at the nanoscale" 16:9 in the predefined format please.

You know how I made it.

[late blade]

You repeat a lot the image format. Maybe create a command like /image wide to minimize your prompt repeats

[onyx ridge]

It's in there. I prompt it to ensure the data is recalled by the model irrespective of context length.

Force of habit from prompt engineering.

Frontload/backload strategy.

[late blade]

If it’s in your GPT knowledge it won’t be forgotten

[onyx ridge]

That's true. But it's not always fully read when it's semantically matched.

[late blade]

Strange

[onyx ridge]

I sometimes get watercolors in pastels if I have a long context and don't reinforce.

I dont mind much. But it's still a habit.

[late blade]

Haven’t had that problem with my commands

[onyx ridge]

I would get 95% the right format if I didn't mention it, too.

[onyx ridge]

A lot of commands in my GPT.

/image "Synthetic solid-state nanomachines integrating with organic tissue at the nanoscale" 16:9 in the predefined format please.

[late blade]

Add to the image refresh knowledge

The command

I’m on mobile so autocorrect messing with me typing

[onyx ridge]

It's okay I understand. The way my GPT calls commands is conditional.

IF the user issues a /command THEN examine MODULES_AND_COMMANDS.txt, find the best match, and try to execute the command.

Approximately that.

It's one of several conditions.

They do different stuff. Obv.

[late blade]

Aha

[left forum]

Is DallE from bing the same dalle as the one from ChatGPT 4 ?

[late blade]

No

Content policy alone is different

[left forum]

Where can I see the differences?

[late blade]

Content policy when referencing products with copyright or trademark are handled differently

The model itself it’s the same

Copilot Pro doesn’t offer the chat iteration experience afaik

[onyx ridge]

@dense mesa

/image "Necrotizing solid-state nanomachines integrating with organic tissue at the nanoscale" 16:9 in the predefined format with dark, contrasty lighting please.

[left forum]

Alright but which one is better then? I got chat gpt 4 but not sure if i should use that or dallE

[late blade]

And the free BIC offers up to 490 characters prompts

[onyx ridge]

GPT-4 Turbo > Bing, DALL-E 3 prompting > designer, playgrounds > notebook.

[left forum]

What’s turbo

[onyx ridge]

YMMV of course.

[onyx ridge]

The current implementation of ChatGPT 4 is ChatGPT 4 Turbo.

[left forum]

ok

[onyx ridge]

It has been for two months and a week or so.

[left forum]

Well then turbo is slower than shi

[late blade]

So you got a gpt+ sub?

[left forum]

Yes

[late blade]

Then use dalle there

[onyx ridge]

It's about 50% faster than it used to be, on average, from my experience.

[left forum]

But takes ages to load an image and plus it has like daily limit

[late blade]

Start chat and do images or you can use OAI’s custom gpt

Both Microsoft and OAI have daily limits

[left forum]

Alright

[late blade]

OAI also has a 3 hour limit, unlike Microsoft

[onyx ridge]

I just use my Custom GPT to ask for as many images as I want.

Usually that's one.

I'm not greedy. But sometimes it's like eight images in one go.

[late blade]

Think custom gpt have an extra limit afaik

[onyx ridge]

Sure, I may have to wait, but it usually finishes them all.

I literally just chain the /image commands with a :!

/image "Necrotizing solid-state nanomachines integrating with organic tissue at the nanoscale" 16:9 in photographic, hyperrealistic style with dark, contrasty lighting :
/image "Necrotizing solid-state nanomachines integrating with organic tissue at the nanoscale" 16:9 in watercolors with light, pastel tones```

[late blade]

We were talking about chaining images yesterday. Custom GPTs and Dalle gpt form OAI can do many renders . But normal chat won’t do so

[onyx ridge]

The sequence I just made is a great example. Each image matches the prompt almost flawlessly.

[late blade]

Nice

[onyx ridge]

DALL-E 3 is great. Using it with CustomGPTs is the way.

[late blade]

Indeed

[onyx ridge]

IMPRESSIONS OF KANDOR:
/image "A post-modern neon-and-chrome cityscape embedded in the lattice structure of a silicon wafter at the nanoscale." 16:9 in the predefined style please.

[dense mesa]

The descent into the molecular scale unveils a portrait sketched in the language of chemistry. The character's face is a mosaic of individual molecules, bonded together to illustrate the features with atomic precision. Carbon, nitrogen, oxygen, and hydrogen atoms connect in a delicate balance, creating the molecules that define the character's appearance. The eyes are crystalline structures of organic compounds, reflecting light and the soul of the character through the arrangement of their atoms at the macromolecular level as we observe a tapestry woven from the very building blocks of life. Proteins fold into complex shapes to form the contours of the character's face. Enzymes, receptors, and structural proteins interlock, creating a visage with expressions governed by molecular interactions. The character's eyes glisten with the glossy sheen of lipid bilayers, the iris composed of intricately arranged protein complexes, each hue a different amino acid sequence in this subcellular world where organelles dominate the view. Here, mitochondria and endoplasmic reticulum entwine to form the character's facial features. The mitochondria, with their inner membranes, serve as the pupils, intense with the energy of respiration. The endoplasmic reticulum morphs into a complex network that outlines the face, with ribosomes dotting the surface like stubble. Lysosomes merge to shape the lips, releasing enzymes that speak to the dynamic nature of cellular processes in a scene where the character's face is a canvas for cellular biology. At a magnification revealing the cellular scale, we see not skin, but a vast array of individual cells, each with its distinct membrane, cytoplasm, and nucleus. The eyes are twin orbs of compound eye structures, with each ommatidium a single cell reflecting a fragment of the world. The mouth is a cavernous opening leading to a tunnel of muscle cells, each fiber contracting and relaxing in a rhythmic dance that mimics speech. This intricate cellular landscape is bustling with activity, as messenger RNA molecules flit from the nucleus to the ribosomes, crafting proteins that give life to the expression on the character's face.

it really went hard there

[thin pelican]

So today, no matter what kind of art style I ask for, all I can get are anime girls using the website. Very disappointing. WHat's up with that?

[onyx ridge]

We're getting some distinctly not-anime girls in our images here.

[late blade]

What’s your prompt?

My King Arthur is so synthetic but I’ll settle for that atm.

[onyx ridge]

After copying @dense mesa 's prompt into the beginning of mine, I shift-entered twice and typed:

/auto 16:9 in the predefined style please.

[onyx ridge]

😮

2x we got very nearly identical output

#nice!

[onyx ridge]

This one was hard:

/collaborate 1 round each agent to describe Helen of Troy without ever using her name. You can refer to her using pronouns. : /auto 16:9 in the predefined style.

But the results were good.

Representation and bias are a real pain in AI. Inverted training and filtering aggression is incredibly inefficient and unethical.

[late blade]

so, are there any current challenges going on?

[open trench]

challenges?

[late blade]

or something fun

i'm bored atm lol

[open trench]

so am I

[late blade]

I was doing some work on my custom gpt, but had so many inaccuracies that got frustation out of it

[open trench]

I'm working on a scampers image, but it's not really that fulfilling and keeps getting worse and further away from what I want

[late blade]

ahaa

[open trench]

we need games like in MJ's discord. But i'd always be in he penalty box if we had those. I'd literally have to move into the penalty box.

[late blade]

dunno, never used MJ

I got so frustrated at MJ for their entry point that I told myself that's not for me

[open trench]

They have like alphabet games where your prompt has to start with the next letter or some such rules. I'd have to look again.

[late blade]

well that's gamification and engagement with the users, we already adressed that with one suggestion, just need to hear from OAI how that's going

thing is GPT is much more than just images

but I guess we need to add more ideas

[open trench]

true. I only really do the images side of things on here. I know there's more, but I love the ai art side of it.

[late blade]

ok, gonna try to come up with a concept for a challenge, so I can engage with others and get out of boredom

[open trench]

I'll message you the rules of the game I was thinking of. I took a screen shot, but I don't want to share it in here.

[late blade]

sure DM me them

[dim cradle]

Glad people liked my Icarus gen. I think it might be a favorite. Turned out nice.

Have we exhausted all the myths yet.

[open trench]

that one was amazing

[open trench]

lol feels like it

[dim cradle]

maybe some more creepy urban legends 🙂

[open trench]

Yeah, i was reading up on whether those constitute as myths. I'm not sure they technically do. So, I've avoided the ones I researched.

But I want to do more urban legends

[dim cradle]

it's not always included in the official lists, but i get the impression urban legends falls into a category of contemporary myths, but i could be wrong, just an impression

[open trench]

ooooh. they have legends in the description of the daily theme. fair game now.

[dim cradle]

sounds logical to me 🙂

[open trench]

lol

[dim cradle]

some urban legends to get us going

1. Vanishing Hitchhiker
2. Hook Man
3. Bloody Mary
4. Babysitter and the Man Upstairs
5. Choking Doberman
6. Phantom of Heilbronn
7. Mothman
8. Jersey Devil
9. Resurrection Mary
10. Crying Boy Painting
11. Bell Witch Haunting
12. Flying Dutchman
13. Headless Horseman
14. Black-Eyed Children
15. La Llorona
16. The White Lady
17. The Highgate Vampire
18. The Bunny Man
19. The Lady of the Lake
20. The Greenbrier Ghost

[open trench]

I've been researching 8

that's quite a list thank you lol

[late blade]

A while ago I did la Llorona when we were talking about folklore

#daily-theme message this one

[vagrant zodiac]

I think they definitely do! But also pls remember to spoiler tag any images that are particularly spooky 😄

[open trench]

Oh definitely. I already did a couple of spoiler tagged ones just to be on the same side. Especially this one #daily-theme message

[vagrant zodiac]

very good!

I do enjoy spooky images

and especially how AI can make them

[open trench]

for sure. It's fun. I also don't mind the challenge of making it appeal to broader audiences by sticking to the content policy. AI doesn't always want to play ball, though. lol I've had to ditch come concepts completely.

[dim cradle]

it takes our words so literally 🙂

[vagrant zodiac]

yeah i'm currently trying to get theseus and the minotaur but it really is not producing minotaurs I could share here 😭

[thick smelt]

hi guys!

[dim cradle]

my last post is an optical illusion so try to look at it that way if you run across it

[open trench]

hi there

[dim cradle]

The Mothman

[late blade]

This isn't scary enough #daily-theme message

[open trench]

it's so cute.

[thick smelt]

hi nezhno chan

[vagrant zodiac]

just glad dalle 3 can actually do cerberus

it was a huge pain with previous versions and with other AI

[thick smelt]

mac guin's first tour 🐧 🎸

that and a modern hades chef's kiss

[late blade]

that would be great

[hot rain]

A new way I've started using GPT is to immerse me in strategy games I play by having it write a journal from the perspective of one of the soldiers in my armies. In this case, I'm playing Total War Rome 2 and GPT is writing the journal of Gaius Aelius, and before and after each battle or major movement of the army I have it make an entry after giving it some details on what happened.

Very entertaining stuff lol. It does a good job at making me care more about the soldiers on the ground. And I also use Dall.E to make images relating to the campaign.

[dim cradle]

just started a Medusa gallery #1200458842413211769 message

[thick smelt]

(Left) Dall-e 3
(Right) the og version (credit to Mystman12)
what do you think of my dall-e baldi?

[thick smelt]

hey ted!

[hot rain]

Howdy! I had to take a crack lol.

[thick smelt]

i have a dall-e baldi, i add the og version for comparison

[dim cradle]

she's broken many hearts (when they crumble)

[hot rain]

💔

[dim cradle]

between yours and @red umbra 's recent one, she was on the brain, i guess

[open trench]

haha.

[fair crescent]

This is..... truly terrifying

[late blade]

lol

[open trench]

penalty box. for two hours. yeesh.

[dim cradle]

ugh, maybe use gpt-3.5 to develop some concepts and write some prompts and have them ready to go when you're back in the game.

or post some gens you've been making

looks like scampers has gone super-hyperrealistic and surreal, i love it

[late blade]

what did you do?

[open trench]

I'm using my microsoft 365 account's dalle3. It's almost identical when you go through co-pilot's gpt to create images. I'll probably burn through my credits in 25 turns because it does four images at a time, though.

[open trench]

I made too much good arts.

lol

"good"

[late blade]

you have to put scampers images with a penalty box

[open trench]

haha.

[red umbra]

I too have noticed this, I will work on an idea, and then when I go to post I see another very close, it seems to be like a frequancy tapped in at times.

[dim cradle]

i'll browse other images when i post mine, reviewing yours and others i know those visuals influence the creative process and future creations, i'll see an element i must try, and sometimes the timing is perfect.

[late blade]

. . .

that's too scary in so many ways

specially if it's so self centered like normal cats

[dim cradle]

yeah, i imagine they don't always get along

and it's hard enough taking care of one, much less if he had 3 mouths to feed

that's a high-maintenance cat

[open trench]

oh wow. Scampers from Co-pilot's Dalle

[late blade]

you caved in and got a copilot pro sub?

how does the experience compare?

I'm curious of how that works

[dim cradle]

awesome!

[open trench]

well, right now, after making those awesome images, all it's making is cartoons. So, I dunno. It's like they hit a switch and it won't comply.

[late blade]

interesting

[open trench]

it's a very similar experience to using GPT here, but obviously without the custom gpts. Also, you can't use any seeds or image IDs. It's a little buggy too.

[late blade]

copilot pro has it's own GPTs

custom ones I mean

[open trench]

oh gotcha

[late blade]

hmmm wonder if that's possible in another way

[open trench]

it's so buggy. It's not worth it right now. It tells me I have a long a detailed prompt and then creates an error saying my prompt isn't detailed or long enough.

[late blade]

lol

[open trench]

I had to put it in creative conversational style for it work

[late blade]

aha

I'm still waiting for my free access to it, but it hasn't been rolled out yet

[open trench]

it's an expensive backup for me to avoid the penalty box. I'll argue that the images are a little better from Copilot's Dalle-3 than this one. But it's got other problems of its own, obviously. I feel kinda strange using and posting about copilot, but it's still GPT and Dalle-3 I guess.

[late blade]

someone got hooked on dall-e

[open trench]

lol

[dim cradle]

@shut niche welcome to the medusa party!

[dim cradle]

i've noticed the quality bump, wasn't sure if you were prompting for it, looks like copilot's could be doing some prompt rewriting, tuned to take greater advantage of the model's capabilities perhaps

[dim cradle]

is it taking around 20-30 seconds to generate an image or closer to a minute?

[open trench]

20 - 30

[dim cradle]

oh, and copilot is using hd for api's quality param instead standard which the old bing used, so that's another reason, i'd guess -- plus they went from square res to wide-format

[open trench]

no idea. but it's making all cartoons again. I get like three photorealistic ones per conversation and then it goes all squirrely on me.

[dim cradle]

i haven't really used it, maybe it's not managing and transforming the prompt as well as chatgpt does

[open trench]

14 minutes before my penalty box is over.

[dim cradle]

time to depict another scene in the medusa gallery: "Perseus and Medusa: The tense, dangerous moment of Perseus using his shield as a mirror to avoid Medusa’s petrifying gaze while beheading her could be rendered with intense drama and suspense." i don't know about you but i'm on the edge of my seat.

looks like we've switched to a "marine" theme now

[late blade]

I haven't found any real information about dall-e on copilot pro

only the 100 credits and wide stuff, but nothing more than that

[late blade]

I'm so scared of the penalty box and haven't done much so far....

[open trench]

lol

I hear you there

I got my scampers in, though. #daily-theme message

[late blade]

you sure like to put the poor mouse in danger

[open trench]

lol

well, you know. it sells tickets. No one wants to see a Scampers having tea or at a dinner party.

[late blade]

I'd love to see Scampers eating and having tea and then the universe collapses

[open trench]

Scampers Vol. 3. Eating Habits

[open trench]

lol

[late blade]

also, have to do character progression

what was scampers like as a baby, who did he marry? did he have 2 or 3 kids? is he divorced? who kept custody of the DVD collection?

all the important questions

[open trench]

lol

[late blade]

Is he lactose intolerant?

[open trench]

that's vol. 4

[late blade]

lol

I'm thinking of just posting the image with no comment, nobody else does comments, only a few

[open trench]

I only sometimes comment. It just depends on the post.

[late blade]

fair enough

oh no @formal osprey you are not allowed to post exotic formats

[formal osprey]

?