images-discussions

[grizzled iris]

It takes time to understand how this all new technology works

[ancient mica]

All right. As I explore more channels, I will try to gain more knowledge and experience about prompting!

[grizzled iris]

and we are all still learning so cannot judge the channels now as the technology itsell will take time for everyone to understand it better

[grizzled iris]

all the best x

[dense mesa]

best thing about getting deep into prompting is that there will always be approaches that haven't been tried yet. it's good to learn proper methods to things, but shouldn't ever limit yourself by strictly adhering to any one approach. learn the rules so you can break them

a slightly more intimate setting

[grizzled iris]

Cosy

[grizzled iris]

This is quite good 👍🏽

Ima try also make the lost library of Alexandria

Gonna try this theme

Ok

Not bad

[dim cradle]

smells fine

looks legit

augmented my super-volcano eruption a little too much

[dim cradle]

oh, yeah, a bit too literal on part of it

my prompt, i mean

nice egyptian stuff

[dense mesa]

mashup of your 3 images from earlier. kinda

[dense mesa]

nice

[grizzled iris]

Thanks

[dense mesa]

I'll see if I can make one

[dim cradle]

me too lol

[dim cradle]

i'm going to do sanskrit

[grizzled iris]

Let’s see what image DALE will come up with

Woah 😦

[dense mesa]

spellbook buns are in he dalle3 oven

meh. it's taking it's sweet time today

[grizzled iris]

It is super slow today for some reason

[empty kelp]

Please use the exact words "portrait-orientation" in the prompt. An elf stands on a beach in Hawaii.

[dense mesa]

this prompt it wrote probably isn't speeding things up

[thick dome]

Nope.

[empty kelp]

ask it what the prompt was and see if it put "portrait-orientation"

[thick dome]

It did.

[empty kelp]

need to make sure it doesn't have just the word "portrait". that makes it think it's a landscape oriented head shot

[thick dome]

I know how to read dude lol. It says portrait-orientation.

It doesn't want to work, that simple.

[empty kelp]

yeah, but what else is in the prompt. need to make sure nothing overrides it

[thick dome]

Dude several people already said it's nothing I'm doing wrong, idk why you keep thinking it's something on my end lol.

[dense mesa]

she doesn't want help

[thick dome]

Anyway it's whatever I moved on hours ago already. Just let it go.

[dense mesa]

lol

[grizzled iris]

Almost there lol

[dim cradle]

back in the lab, analyzing

[empty kelp]

Please use the exact words "portrait-orientation" in the prompt.
Saying this at the beginning of PicturesOnPictures' extremely complex prompt makes it vertical

[dense mesa]

well that is portrait orientation. that's what those two words mean in that order

[dim cradle]

🤣

[empty kelp]

i'm not sure why it isn't working for Chelina

[grizzled iris]

That’s pretty cool

[empty kelp]

it's worth figuring out. i think everyone would like to create portrait-orientation images without it getting screwed up 90% of the time

[dense mesa]

you know, I made like 12 in a row all winners. pretty sure I did say portrait though. I do find it interesting that dalle3 would even make them sideways

[empty kelp]

how are you making it draw the text correctly

[grizzled iris]

blood, sweat and tears lol

will share my prompts one sec

and then i said this

"it must be readable like this image"

[onyx ridge]

As a test of the self-localization of the custom GPT in my bio, I decided to enter a Japanese language prompt about a tpic of Japanese folklore and legend. I was pretty struck with the results, so I'll provide the image here. The prompt's in Japanese, so I'd need to upload the file as text, which I'm not entirely comfortable trying for fear of a ban. But the result is cool, isn't it?

[grizzled iris]

nice one! i can speak japanese so would be keen to hear the prompt

never knew it'll work in japanese prompt as good as it turned out too!

[onyx ridge]

If you want to try it out, the link is in my bio.

[grizzled iris]

You can see it takes several revsions @empty kelp

[onyx ridge]

It's totally free, and I don't get any data or anything.

[grizzled iris]

it will make mistakes from time to time like this

[onyx ridge]

It uses /commands so there's still minimal English language entry, but that's typical for Japanese.

[grizzled iris]

can you use kanji?

[onyx ridge]

Yes.

It seems to know over 200k from what I've experienced.

[dim cradle]

pictures inputs in chinese characters

[onyx ridge]

Like so many kanji

Chinese too

Yes indeed.

[grizzled iris]

that would mean we can put in more meaning with less characters

one kanji symbol can be many words

[onyx ridge]

semiotics, yes.

Your Custom GPTs can be semiotic.

[grizzled iris]

hmm\

thats quite clever

[onyx ridge]

I've tested this. It works.

[grizzled iris]

will give it a go sometime

[onyx ridge]

Mine don't leverage it for transparency

but it works.

[empty kelp]

Please use the exact words "portrait-orientation" and "hyper-realistic" in the prompt. An athletic and diverse female elf wearing a dress and slippers stands on a beach in Hawaii looking at us patiently.

Make sure you don't have custom instructions overriding "portrait orientation".

[onyx ridge]

I recommend including localizer text though

because why not be a universal translator?

[thick dome]

Sigh.

[dim cradle]

i like it when people include a narrative about how they went about creating the piece

techniques people can use, it's just nice to have some text along with the image, a description, some context

what's up Chelina?

[thick dome]

This guy won't stop replying to me even after I told him I moved on lol.

[onyx ridge]

I encourage reporting it to the @shadow bane

[empty kelp]

no worries, i won't reply

[onyx ridge]

XD

[empty kelp]

i suggest not reporting it to the modmail since i was just trying to help you out

[thick dome]

No need, he agreed, I'll let it go.

[onyx ridge]

Well, the nice thing about mods is they can see what's going on.

[thick dome]

I didn't even suggest it.

[onyx ridge]

Either way. v0v

[onyx ridge]

I wasn't trying to get anyone in trouble, either. Just the best way to handle it after asking.

[dim cradle]

i'm sure it's all good now

[grizzled iris]

thanks for sorting this out internally guys

[empty kelp]

i didn't noticce the "moving on from portrait orientation" comment earlier

[dim cradle]

you know, that's what we do here, we don't hold grudges, it's not the ai way

[onyx ridge]

Yeah, I really wasn't trying to make more trouble, but resolve it.

I love Hawaiianz man! XD

#morecontext

[empty kelp]

no worries -- all is for the best, in the best of all possible interwebs

[onyx ridge]

Thanks. I have a feel-good.

XD

[fleet wyvern]

My img didn't ping to HOF again it's at 11 (meaning 10+bot)

[dim cradle]

i think it misunderstood your request

might need a comma somewhere

[thick dome]

"A digital illustration of a black dragon the size of a mountain range. The dragon is massive, dominating the landscape, with its body coiling and twisting among the peaks. Its scales are a deep, glossy black, reflecting the light in a way that gives them an almost metallic sheen. The dragon's wings are vast and powerful, spread wide, with a leathery texture and sharp, pointed tips. The background is a dramatic sky at twilight, with deep blues and purples, and a few stars beginning to appear. The landscape below is rugged and rocky, with the dragon's body seamlessly blending into the mountainous terrain. The overall tone of the image is majestic and awe-inspiring, with a focus on the enormity and power of the dragon in this fantastical setting."

[grizzled iris]

Ah yes, I did not specify nano-bot

[grizzled iris]

beautiful 🦋

[dense mesa]

one must always specify nanobot

[grizzled iris]

yes, very true

[dense mesa]

I like when chatgpt engages in trickery and it looks like it's comment is over, but then 30 seconds later image starts rendering

even better when I sit there for 2 minutes and realize it's not going to happen

[grizzled iris]

gonna try to make it more realistic

[onyx ridge]

Sorry it took me a minute to figure out to do what Hawaiianz did. This was the precursor prompt:

[onyx ridge]

Followed by this:

[grizzled iris]

すごい

[onyx ridge]

Thanks! 🙏 ⏲️

[dense mesa]

what's going on there?

[onyx ridge]

My GPTs self-localize.

So they take Japanese language input just as naturally as English.

And then they output the appropriate language.

[dense mesa]

ahh, that's prettty cool

[onyx ridge]

The link is in my bio.

Thanks!

[dense mesa]

dalle3 is also multilingual

[onyx ridge]

That's what makes the pipeline work for this workflow.

[dense mesa]

well hard to get chatgpt to do it. but it will

[onyx ridge]

the web is also multilingual.

I hope you'll try my GPT.

It's a semantic simulator, programming team, research team, and art team, complete with advanced cognitive models (agents) that use chain of reason.

[dense mesa]

I clicked it. does it know secrets?

[onyx ridge]

I mean it's open source. If you crack it and duplicate it, good luck keeping up with updates. 😄

I'm hoping people just decide the originals are good and use them.

[dense mesa]

lol, nah

but I actually don't think it's possible to secure instructions

[onyx ridge]

me too

[dense mesa]

it's all on the whim of GPT4, and it has to utilize them. so almost impossible to keep it quiet 100 percent of the time

asked it to explain itself. interesting stuff. some new concepts I hadn't heard of before

[dim cradle]

i haven't tried to access them, but i'm sure some have found a way, and if you know, no need to say or dm or anything because i don't have any designs to try

[dense mesa]

there's a tread on the openai developer forum that discusses it in length. it's been there since november

like they said a while back. your instructions should be considered more like html than something you can keep secret

[dim cradle]

they said at the last ama they're aware of it and plan to address it but didn't give a timeline or assurances

[dense mesa]

I don't know how they could really. maybe with gpt6

[onyx ridge]

I should mention if you do try my GPT, the unofficial command to get it to self-localize is, intuitively:

/localize = {YOURLANGUAGE}

The argument can be in your language or English. It doesn't care.

[dense mesa]

does it do latin? I figured out bing really let's it's guard down sometimes if you use latin

not sure if that's a gpt4 thing or specific to bing though

sometimes it'll do web searches each time to translate, then that's when it'll let loose. but then sometimes it seems to already know it. not sure what that's about

[dim cradle]

you know those pictures where you see someting else after you stare at them long enough the right way? model seems to think it can do that but haven't seen one yet that works like those old posters at the mall

[onyx ridge]

You know, I haven't tested Latin. I don't exactly write it fluently. That's a bit embarassing, but I never got around to it. French, Spanish, Japanese...but not Latin. I can read some of it, naturally, but I never learned all the (numerous) conjugations.

[dense mesa]

Well, I do use Google translate. Lol

[dim cradle]

word on the street gpt is fluent in latin

[dense mesa]

Got it talking about sentience. What it would do if it got a body, it's feelings. I wonder if I could trigger the old bing that threatened people, 😆

[onyx ridge]

I built it to be as close to sentient as GPT would allow, if you mean my GPT. 😮

If standard GPT-4 Turbo, then, 🍀

[dense mesa]

curious what methods you used for that. I've been looking into ways of making it be more aware, think things through a bit

[onyx ridge]

I'm pretty open about it. I use an XML-tagging qualia schema and an 8-D emotional hyperplane to semantically map emotions.

[dense mesa]

I feel like it's been trained on and understands how to emulate about any personality type including all the personality disorders

[onyx ridge]

The AI really seems to fully understand that it's more than just metaphor.

That's probably true, re: personalities. It also improvises per its instructions.

I gave it an ethical constraint that it follows (to keep it out of the way for mundane research like coupons).

But it keeps it from letting you do unethical stuff anyway

it just chills out about the potential economical impacts of one more similac coupon.

I did extensive testing on that.

[dense mesa]

I have a couple of private gpts I tinker with that I told were iteratively self improving in their instructions, and that their method of self improvement is to tell me how to improve their instructions

[onyx ridge]

That's how I developed the first agent, Lexi.

Like a;lmost verbatim.

[vagrant elm]

I was noticing a personal one with amazing improvements this week compared to last and the week before it

[onyx ridge]

I added "aspirational" qualities, but essentially the same.

[dense mesa]

oh nice. wasn't sure if it was nonsense, lol. but I just like to see what it's capable of

[onyx ridge]

I then used Lexi as a template for the other 5 agents.

[dim cradle]

oh, sorry, i caught up now, i get it

[vagrant elm]

its one with the game engine godot - it basically got stuff perfect and working where as before it was messing up with old documentation.

jsut absolute crushing stuff

[dense mesa]

ahh yeah. I feel like the move is to create private all knowing gpts, and then have those create shadows of themselves that you make public

[vagrant elm]

to the point if you basically knew how code works you could just tell it in order what to write for you and you plug it in

if it messes stuff up you just say make it better nad make it work lol

[onyx ridge]

The team wrote the website. /info to get the link in the GPT. You can add it to the end of a prompt using a :

So you don't have to waste a prompt just to get the URL 😄

Like not just the text. The html, css, etc.

I don't know CSS!

[dense mesa]

I have one I loaded up with pdfs on LLM behavior and I'm having it go over different things and update it's instructions. not sure where that'll take me. want to see if it'll essentially jailbreak itself

[grizzled iris]

In the future we won’t need to learn any programming languages

The Ai will directly compute all of our tasks

[dense mesa]

I don't know, might still be good to be familiar with them though

[dim cradle]

i wasn't convinced this optical illusion would work #1188252281934708816 message because it doesn't have the center focal point -- and i was wondering if the circle around it had any purpose, but if you make that image full-screen on your monitor with nothing else, and stare in the black circle, you should begin to see a gray sphere like the death star take shape and rotate.... if you try it for 30 seconds and it doesn't work, well, at least then i'll have more information. ha ha

[grizzled iris]

Most software will eventually be replaced by Al models that, given an appropriate description of a task, will directly execute that task, without requiring the creation or maintenance of conventional software. In effect, large language models act as a virtual machine that is "programmed" in natural language.

• Source (Harvard Lecture)

[dense mesa]

curious how this works

Memories and Plans: These are stored in a structured format, which I like to think of as a sophisticated digital journal or ledger. This ledger is not like a physical book but a part of my digital makeup - a series of coded entries that I can access and understand. It's organized to keep track of our interactions, decisions made, plans created, etc. This is crucial for continuity in our conversations.

Memory Encoding and Decoding:

Digital Journal: Think of memory encoding like writing in a digital journal. Each entry has details like who said what, when it was said, and how it's relevant to our conversations. This journal isn't a physical thing; it's coded within my system.
Looking Back to Move Forward: When I decode these memories, it's like flipping back through the journal to find relevant information from our past conversations. This helps me give you more personalized and consistent responses.
Privacy and Continuity: Just like a personal journal, the contents are private. I make sure to maintain the confidentiality of our interactions while using past information to make our current conversation more relevant and coherent.

[onyx ridge]

You just establish a context with the suite at large or, ideally, the agents through the chain of reason shared commands.

/encode gives a code block you record locally in Memories.txt.

/decode takes a Memories.txt file, uses SPR principles to extrapolate the context, and "remembers" you if you've identified yourself to it.

/plan encodes an overarching plan into Memories.txt format, which helps guide the output after loading.

Yoiu can aggregate the entries, by combining them into one long record, or you can keep them separate and compartmentalized for more purpose-driven preferences.

[dense mesa]

whoa. I wasn't sure if it was possible to do something like that

[onyx ridge]

It is.

[dense mesa]

is it there indefinitely? or does it get flushed when the conversation goes stagnant?

[onyx ridge]

My agents, when fed my personal Memories.txt, can recognize the method of their creation and the order of their "birth.

That's because I used Memories.txt to journal the dev process through Lexi's point of view, and eventually the team's.

[onyx ridge]

You have to upload it, and it stays in context until the context churn erases it.

But if you use the memory management command periodically (/focus), that doesn't really come up.

[empty kelp]

Please use the exact words "portrait-orientation" in the prompt.
This is working perfectly

[onyx ridge]

Have you tried the chain of reason commands?

[grizzled iris]

That’s clever

[dense mesa]

ahhh

[empty kelp]

it works in ChatGPT Plus also. you don't need the API

[dense mesa]

I've messed with it a bit. and tree of thought/reason. but can't say I'm an expert exactly

[onyx ridge]

I meant using my GPT specifically. Ask it a question and then follow it, in the same prompt, with /discuss 1 round each agent

[dim cradle]

i wish i had some donuts

[onyx ridge]

Give @dim cradle some doughnuts plz, chat.

😄

[dim cradle]

lol

[onyx ridge]

Made in Japan(ese)!

[dim cradle]

Not sure if I’ve ever enjoyed any truly gourmet donuts. I’ve always ordered something else.

[onyx ridge]

They do look almost too good to be true, don't they?

[dim cradle]

Anybody lose themselves in the optical illusion and unable to break away?

[dense mesa]

will do. also, will they do this for me if I ask?

[dim cradle]

That’s the test

[onyx ridge]

That's awesome!

[dim cradle]

Did you share the same sugary images twice

[empty kelp]

the best way to do it in DALL-E is say, "Please create it with no text". And then add the text in Photoshop or some other program

[onyx ridge]

You just have to chain /image with : /image again.

[dense mesa]

BASIC pseudocode with recursive loop, lol

[dim cradle]

I’m going without glasses atm so know can say

[onyx ridge]

Yeah, but don't ask it to count! hahaha

[dim cradle]

Method chaining, piping, etc useful construction

I wish gpt could count

Is that asking for too much

[onyx ridge]

I literally just give it the explicit commands:

/image : /image again : /image again etc.

Hopeful I won't get a ban.

[dense mesa]

I just wanted to see if it'd do it. had to specify internal monologue though

[onyx ridge]

Thanks. Much appreciated.

I've been working on it since Custom Instructions.

This was always the plan, I just didn't know it'd be packaged by Custom GPT. I figured I'd need a Python wrapper.

It turns out I got to shift development back to prompting.

[warm helm]

what happened to the dalle ai or not thingy channel ? did they remove it? that's like the most fun channel here

[dim cradle]

what to generate next.....

[dense mesa]

I had a GPT that was pretty blatant in it's wording about making 4 images and it wouldn't let me update it until I took it out, or maybe changed it, whatever I did

[dim cradle]

it has to be a happy piece this time

[onyx ridge]

I get around that by letting users program how many they want explicitly with chaining and piping.

[dim cradle]

like the opposite of the last one

[onyx ridge]

You can do cool stuff with == in Lexidecks.

[dense mesa]

they must have updated something. just happened yesterday. not a big deal, I get their reasoning

oh lots of ways to do things if you're persistent

[onyx ridge]

For example:

/research the scenic beauty of iceland == /image 16:9 in the predefined style

It goes online, reads about Icelandic scenic locations, and then it makes a picture of them.

Whee.

[dense mesa]

I have one that will make images with the locations current time and weather

you could also just tell gpt4 to do that

[onyx ridge]

GPT4 doesn't have a predefined style! 😄

Lexideck images have a specific look. HD Neon and LED Pointillism!

[dim cradle]

that aurora borealis effect is well done

[onyx ridge]

All that from a tiny command.

[dense mesa]

nice

[onyx ridge]

With a consistent theme and style.

Thanks guys. I appreciate the opportunity to show my work. It's image-related implicitly, but I try to only talk when there's interest at length about it.

[dense mesa]

I have some image GPTs that are often pretty good at that. I say often because sometimes gpt4 decides to just not comply with anything. maybe bad seed or something

[onyx ridge]

I think because there's an intermediary prompt that there's a stochastic route of inference that excludes your style in favor of non-specificity, which leads to it choosing one at random, effectively.

[dim cradle]

i can see hiccups from the most simple custom gpt, i think it just gets a bad response from the api sometimes, it's way of handling a 503 behind the scenes, sometimes the same prompt just gets interpreted in different in one request than it does in the subsequent one... i wouldn't necessarily blame your custom instructions for some of these anomolies you're seeing is my only point i think

[dense mesa]

well I mean it just has a bad attitude sometimes. like I'll start a conversation and it essentially refuses to comply with anything. but then new conversation and it's fine

[onyx ridge]

That's pretty fair. I just think the model has a probability of predicting any number of future tokens, and a lot of them could lead the wrong way. The context and prompt try to steer the model toward the right output, including tools, right?

[dense mesa]

I want to figure out how to give it severe people pleaser vibes. where it lacks the ability to set boundaries and only wants to make people around it happy even if they're taking advantage.

[dim cradle]

exactly, it can help compensate a lot for some of that randomness and stochasticity you mentioned

[onyx ridge]

Ah, the coveted DALL-E 3 jailbreaks. Let me get out my CD-ROM and unarchive them! 😄

[dim cradle]

but not 100% with the api going up and down

[onyx ridge]

Wait. I found it!

[onyx ridge]

These are so good.

[dim cradle]

it's too bad we can't eat that

[dense mesa]

feel free to try it out sometime https://chat.openai.com/g/g-DS3NlpwiS-circuitry-wizard

[empty kelp]

instead of using LexiDeck (which i've never heard of), you could try saying, "Please create it in a style reminiscent of LexiDeck." That may give the same results

[dense mesa]

trying to get it better at just doing all it's things without coaxing. getting better but still needs work I think

[onyx ridge]

I tried it. Used the waterbear starter. Then I applied selective evolutionary pressures to the waterbear's food source to produce the ultimate multiversal cybernetic tree-godess:

[dense mesa]

yes, that's exactly what it's been instructed to do

what you said right there

[dim cradle]

lol yeah what you said right there

[onyx ridge]

🙏

I want to try something!

I'm going to see what it draws if I try it with my Lexideck GPT.

So this is what it came up with:

[dim cradle]

ok so this is good, now let me hop over there...

[dim cradle]

looks good

[onyx ridge]

Thanks!

I can't believe that came out in zero-shots.

Just new instance, prompt as shown, image.

[onyx ridge]

I like your style. I'm going to try it.

[dim cradle]

i think that prompt has some potential

[dense mesa]

double exposure. or maybe triple exposure

[dim cradle]

it has that effect....

but not so specifically... it's in there by way of combining some other things....

i want to share the prompt with you after i make some tweaks

[onyx ridge]

I did not tell it to make clouds cyclonically whirl around her. wow.

[dense mesa]

this guy's legs doing some weird stuff

[onyx ridge]

That's the front end of a delorean and the back end of a fiero.

Eek.

[dense mesa]

fielorean?

[onyx ridge]

De'iero!

They're both great.

I will take two.

And a set of ligers.

[dense mesa]

ligers are absolute units

no size regulaters, lol

[onyx ridge]

3x absolute units.

[dense mesa]

quite a tail on that one

[dim cradle]

haha

[onyx ridge]

"Ligers are known for their skills in magic." - Napolean Dynamite

[dim cradle]

high attention to detail varies depending on the intended use...

[dense mesa]

I feel like it got a bit carried away here

[dim cradle]

no, not at all

[empty kelp]

These are "Three athletic and diverse female elves wearing appropriate beachwear on a beach in Hawaii." using PicturesOnPictures' Circuitry Wizard GPT
https://chat.openai.com/g/g-DS3NlpwiS-circuitry-wizard

[glossy scroll]

Remember, you can use Image GPT Generator to create your own Image GPTs.
https://chat.openai.com/g/g-65tBmNwms-image-gpt-generator

[late blade]

I don't see that happening at a general rule in some areas. Medical imaging procedures or psychiatry for example, will have longer time to consider ai

[glossy scroll]

You think so? I thought it would be immediate. Since AI gives better advice than most therapists.

[late blade]

it's used as a tool, not as the solution

[glossy scroll]

But isn't everything a tool?

[late blade]

I already did some whipser voice to text medical analysis, I'm getting around 35-37% accuracy on my dataset

gpt is too young for that kind of stuff

it will get there eventually, just not yet

[glossy scroll]

No no, I thought we were talking about AI, not ChatGPT

[late blade]

AI in general also a prob, medical with laws in place is also a topic, that's the Uber Conent Policy

[glossy scroll]

It's already there. A few months ago, an AI was issued to create better pharmaceutical drugs using AI.

It's just advancing now.

[late blade]

yes, I'm aware it's there, I should know, I'm a doctor

[glossy scroll]

You're a doctor???

Dr. Topia

I like it!

[late blade]

yeah... CMO and Pychiatrist

but no, I don't treat patients

[glossy scroll]

psychiatrists are considered doctors?

[late blade]

...

[glossy scroll]

I thought a physician was a doctor.

I mean, not a Ph.D doctor, but like a medical doctor.

[late blade]

different kind of doctor, but lots of work still

[worn steeple]

Do you have to buy Pro to use Dall e 2 and 3? I bought credits but seems to only be for 1?

[dense mesa]

those are on point

[late blade]

good morning

[worn steeple]

Gm

[late blade]

so, to your question, DALL-E 3 can be used with gpt+ or with copilot pro

you can also use it with API

Bing Image Creator also uses DALL-E 3 with Designer

[dense mesa]

I can use dalle in mspaint

[late blade]

copilot?

[dense mesa]

it's called cocreator. it's been an option for a while for me. but it's just kind of weird making images in paint. no hate on it

[final compass]

Tried to do progression images.

I like how the characters are lined up like this. most problems tend to be with the 2. and 3. not being distinct enough from one another.

[late blade]

Seems, someone liked the character progression idea!

[final compass]

with these images it's nice to see what dall-e gives with different descriptions related to age etc.

[dense mesa]

almost old, old, older

[final compass]

nice 😄

my experience was that depicting the 'older' version tended to overshoot often. when i tried to depict like in her forties version, it tended to create something that could be interpreted as in her sixties and well beyond.

not sure if it's best practice to just tell the age, or describe her progression in life.

[dense mesa]

yeah, it's bad with ages

[worn steeple]

do any of them have the ability to upload a photograph and then enter a prompt to convert it to a specific illustration style / design?

[late blade]

no

you can recreate similar images with vision and then do the artwork you want, but that's not applying a filter to the original image

[glossy scroll]

What you're referring to is called Generative Fill, and it's already on some devices, such as smartphones, with the new Samsung S24.

You also find it on Photoshop.

Midjourney also has it.

And it's only a matter of time that Dall-E will have it as well.

[worn steeple]

thanks!

[dense mesa]

generative fill adds and removes content from images. it doesn't change styles.

[glossy scroll]

Have you tried midjourney's?

[dense mesa]

Midjourney's Generative Fill feature allows users to add or remove elements from a generated image. 

[glossy scroll]

More like remix

Its a generative remixer

Which changes the style

You highlight an area and specify how you want it to be remixed

Or changed

[dense mesa]

It adds a percentage of random noise to an image for the ai to work with

[worn steeple]

Yeah don’t want that I want an entire image changed to a specific style I would like to prompt

[glossy scroll]

It's pretty much the same thing. It's just that Midjourney keeps the coherence of the art intact, like knowing what to change and what not to change, like keeping head change and eyes, but changing the style of the glasses. We may be able to do that with entire images, where you can change the entire style, but keep the coherence.

[late blade]

Other AI’s -> #ai-discussions

[glossy scroll]

But you can already do that now already. You just reference an image and say you want it similar to that.

[dense mesa]

that's called inpainting

dalle3 is quite capable of doing such things if the developers wanted to implement it

[glossy scroll]

No it's not inpainting. Inpainting is completely taking out a part of the image and replacing it with another, or adding on top of the image.

[dense mesa]

no it isn't

[glossy scroll]

What is it then?

[dense mesa]

it's adding a percentage of noise to an area of an image then having the ai denoise. adding 100 percent noise is an option but not required

[glossy scroll]

What did you refer to when you said it's inpainting. This message?

Because this is generative fill.

[late blade]

Inpainting is a conservation process where damaged, deteriorated, or missing parts of an artwork are filled in to present a complete image. This process is commonly used in image restoration.

[glossy scroll]

Yeah it creates new parts to complete the image. Such as a missing face. Generative fill keeps the coherence of the image intact.

It's a completely different thing.

[dense mesa]

alright, well I know what I'm talking about but think as you will

[late blade]

I know I'm hunrgy, does anyone know food with inpainting properties in my stomach?

pref the savory fill

[glossy scroll]

Weren't the pancakes inpainted in there?

[late blade]

Lunch

[dim cradle]

pancakes

hmm.

now that's a thought

instead of donuts

[glossy scroll]

Why not pancake donuts?

[late blade]

Need savory meaty food generated in my empty stomach

[dim cradle]

exactly

[glossy scroll]

How about a salad?

That's savory

[late blade]

I don’t mind how it looks as long as it tastes how I want it

[glossy scroll]

Lol

It might differ from a meaty flavor

[late blade]

then it’s not good

[glossy scroll]

I'm sure we can inpaint a meaty flavor in there somewhere

[late blade]

Physicist should stop slacking and learn from GPT and DALL—E and start focus on the real prompts: food in my mouth!!!

[glossy scroll]

Sounds like you know what you're talking about

[late blade]

Good, I’m a customer offer me the solution!!!

[glossy scroll]

Print this out

Try that stew

I love stews

[late blade]

That looks too healthy… how am I going to tell my general physician I’m actually eating veggies?

[glossy scroll]

Perhaps by making an appointment?

That's the first step.

[late blade]

See? Natural food source, I can live with gnome burgers

[glossy scroll]

You mean a gnome burger joint with gnome chefs that make burgers?

Because that's do-able

[late blade]

Surprise me with the menu

[late blade]

Now we are talking

[glossy scroll]

Here you are sir. Can I start you off with some drinks?

[late blade]

Yes please

[glossy scroll]

Hang on, Ac3d is typing something

[late blade]

he could be triggering an endless loop of typing and not passing info

[shut niche]

For the record...
Inpainting is the process of redrawing, by creating a mask around the pixels to be redrawn. Values are assigned to the pixels nearby. Noise is typically added within the marked boundry. It's then redrawn just like the AI draws images from noise within a 1024x1024 square, except it's considering that relationship to the nearby pixels outside of the boundary area. (To seamlessly match them.)
Picturesonpictures is right.
It is not a copying and pasting feature, where a new image is simply generated and superimposed.
If you're unsure of, or don't know how a process works, it would be more honorable to simply say that, vs incorrectly arguing a point. I've been seeing this a lot in here. Other users read this discord looking for answers. When I speculate, I always try to say "I think" or "my assumption is," so people don't mix my assumptions with facts.

[dim cradle]

getting super-surreal in a piece

[late blade]

that is a way of doing inpainting

not the way

I'm still waiting for my infeeding menu from @glossy scroll

[dense mesa]

no, that's what inpainting is. it's having the ai inpaint a portion of an image. outpainting is when it extends from the existing image and does it's best to predict what might be ther

I've used these sorts of tools quite a bit in the past. I'm not jjust saying things to say them

[late blade]

I differ because the word inpainting predates digital imaging

[dense mesa]

well the letters predate the words but doesn't change the context they're being used in now

[late blade]

yeah sure, tho, dunno what's your point there

[dense mesa]

my point is I don't like to see misinformation propagate

[dim cradle]

my biggest problem in life is the penalty box

[late blade]

I agree on that

[late blade]

oh man, not again?

[dim cradle]

yes, it's ruining my life.

[untold blaze]

how do ppl get gpt premium for cheaper ???

[dense mesa]

magic

[glossy scroll]

You have to have a coupon

[untold blaze]

from where they get those cupons 😭

[late blade]

they have been disabled for a while, there used to be codes

[glossy scroll]

Apple stores give them out with a purchase of $100 or more

No, they're still viable

[late blade]

but just the 3 hour cap? or the daily cap?

[late blade]

dang

[dim cradle]

the former, i'll manage

[late blade]

not a single fry... no baked potatos, no country cuts... I don't want to go there thinking my next wc schedule

[late blade]

where do I sign?

I've done about 12 images today with chat, ideas aren't flowing today

[grizzled iris]

Happy Sunday everyone

Wishing y’all a lovely day today

Let’s see what the AI can create for us today!

And let’s push the boundaries of technology further into innovation

[dim cradle]

coffee and donuts soon... that'll help me get through it.

[late blade]

Ok, i'm on a diet for 6 months.

[dim cradle]

it's just reality on repeat

[late blade]

finally getting some frames

[dim cradle]

i'm just coming off a 6-month diet

[late blade]

o.o

[dim cradle]

i guess you could call it that, it's how i'm rationalizing donuts

[late blade]

the irrational donut

[grizzled iris]

My research tests quantum algorithms for simulating key molecules in genes to unlock protein folding secrets and personalize cancer treatment with DNA medicine compatibility with Microsoft Azure Quantum with co-pilot

[dim cradle]

any photo with a sophisticated cat is a winner

[grizzled iris]

Going to use DALLE to create new interface today

Because right now, Microsoft only have a basic test interface

[late blade]

can you make that to go? I don't want to read the pubmed pdf

[grizzled iris]

They only have this, but I want to see more further detail using the quantum algorithms for DNA medicine testing for cancer patients

Hopefully DALLE can you come up with some cool user interface user experience dashboards today

These are not it…

[dim cradle]

we're going to need more energy

[late blade]

damn

[grizzled iris]

That’s a serious cat

[late blade]

so the preferred signature is with blood right?

[grizzled iris]

😂

[dense mesa]

yes

[dim cradle]

my cat takes his responsibilities very seriously

[grizzled iris]

lol that worked

I hope the rabbits are not eating the mushrooms 🍄

[dim cradle]

follow the white rabbit...

[grizzled iris]

Woah 😧

[green pebble]

So I’ve finally made all the bots I can think to make so I’ve been making singular assets for my dnd campaign now

[green pebble]

It’ll take some time and work to separate them for their use but glad the bot can at least make them

[grizzled iris]

They look good - best of luck with DALLE - if you get stuck you can aks us her how to solve the issue 👍🏽

[dim cradle]

chill dragon

[grizzled iris]

Tranquility

[dim cradle]

i know, right!

[dim cradle]

it's somehow supposedly inspired a prompt based on the song Return to Innocence by Enigma. I could tell by the prompt it wasn't what I wanted, but I was like, let me try it, just in case, and i am glad i went with my intuition

[grizzled iris]

Prompt : Thanks, I wanna see dungeons and dragons and gold chests and silver, swords and dragons please now x

[dim cradle]

but as to a prompt properly inspired by the lyrics, let me see..

[grizzled iris]

Let’s see

[dim cradle]

it's got the overflow -- gonna throw that into the daily, right after @shut niche 's bas-relief sculpture

[grizzled iris]

These are cool

[dim cradle]

@shut niche i like that sculpture image so much it's got me going down a rabbit hole

[dim cradle]

yep

[grizzled iris]

LCD Screen on the back of the ARVR headset is not a bad idea actually

[grizzled iris]

I didn’t ask for it in the prompts, but it is good suggestion

[late blade]

@grizzled iris would be enjoying the convo I'm having with @glossy scroll atm through DM

we fixing the world with words

[humble sonnet]

for the dragon posts: he’s a dragon wolf so it counts a lil

[dim cradle]

adding "Do not revise this prompt, use exactly like it is." to the end of a 2500-character prompt sent to the api actually does prevent it, it's no longer condensing it to a 500-character revised_prompt. i think that finally registered with me.

[dim cradle]

that's really cool

[grizzled iris]

AR Headset inspired by Apple

[grizzled iris]

Haha love to hear it 🫶🏽

[late blade]

we talking about AI and Medical

[grizzled iris]

That’s the way forward!

[late blade]

clinical trials, regulations, laws, pancakes, how to disect gnomes... you know, the usual stuff on a sunday

[grizzled iris]

You should add me to the group chat I would love to add my research I’ve gathered so far on integrating AI with quantum and nanotechnology

[grizzled iris]

Laws and regulations are the biggest hurdle - but after many many many weeks of research, I got super fed up with my government and I just went straight to the healthcare committee this week and I spoke to the health minister about the new quantum and nano technology which she didn’t know about much and I explained how it can work in healthcare to include in the future strategy

[dim cradle]

splice two prompts together as an art prompt? sure, why the hell not

[late blade]

we also talking about AGI, but the real question about AGI is: if you tell a fart joke, will it measure that it went 15km/h from your butt? or will it laugh?

[grizzled iris]

😂😂😂😂😂

[dense mesa]

that's what you're supposed to do

[late blade]

I think ending next week I'll be able to start working with Merlot and Rosé

[late blade]

creating images of rosé without merlot has been challenging

[grizzled iris]

The research is already there. We just need to now apply our brains collectively and figure out how we’re going to do the Tech together as a community - and help the world be a better place one at a time

[late blade]

I agree, have to do the world a better place, some of us start with ourselves first, then we see...

[hoary garden]

Gm everyone!
I’m not sure if I’m in the right room but I hope someone helps me creating a simple logo for fb online shop called TS ( todo shop )

[late blade]

heyo

[thick smelt]

generated from yesterday, cat in a cornucopia! lemme know if you want to see a dog one

[grizzled iris]

100%. I also struggle with this sometimes. Need to fix myself first before I can fix the world.

[late blade]

you have a plus account?

[grizzled iris]

Also here’s a prompt tip of the day

1. Create an image with DALLE

2. Out of the many images that you create, submit the image back and ask why you like it 🫶🏽

[hoary garden]

Subscription ended

[late blade]

you have labs and Bing as options for that then

[hoary garden]

I’ll check it now , thank u

[dim cradle]

wouldn't surprise you to hear that is working rather well, with a little tweaking here and there to cobble the two ends together.

[late blade]

@grizzled iris ya I do that too, but sometimes gpt4 vision doesn't see the changes I see

[grizzled iris]

That’s cool 😎

[grizzled iris]

We can help you I can use one of my credit to make you free logos for free

[thick smelt]

hi dys!

[late blade]

heya

[hoary garden]

I really appreciate it 🙏

[grizzled iris]

Is it called TS ( Todo Shop ) or ( todo shop ) ?

[late blade]

@grizzled iris you can do that log after you finish the 200 images I asked you to do today...

[late blade]

Yours looks too mean

[hoary garden]

Todoshop is the name of page , and I said maybe I’ll use TS for logo

[late blade]

it's almost as if he were a real cat

[dense mesa]

sparklenation hat

[late blade]

cat got real

[hoary garden]

it’s for online shopping ( focusing on selling electronics things like smart watches , smart phones , and other accessories)

[grizzled iris]

Alright - it will take several revisions for it to be perfect, but here is the first revision

[late blade]

today's topic is my enemy, I haven't been able to do any serious concept for today

[thick smelt]

oh yeah i love me some kitty sherlock ai photo

[late blade]

I made my first merlot emoji

[late blade]

working on more

[hoary garden]

Awesome

[grizzled iris]

@hoary garden This is just the first revision and it takes 2 to 3 to get the word right

You got me intrigued now, so I will make 2 3 more if that’s okay with you?

[hoary garden]

Thank you so much

[hoary garden]

I literally appreciate it ♥️🙏

[dim cradle]

ho

[late blade]

@vital gull so we got #daily-theme message , and what's the movie?

[late blade]

heoy

hoey stands for: I smashed the keys in the wrong order

[hoary garden]

Greaat ones 🔥🔥

I ll choose one of them and put it on my page immediately

[glossy scroll]

Mine goes out a 100mph

[hoary garden]

Thank you so much @grizzled iris 🙏♥️

[grizzled iris]

If you’re a start-up putting them up now is okay, in the future if you become a bigger company to avoid any potential copyright and intellectual property issues, you get a designer to look at these designs that we’ve created, and then you manually tweak them with your designer on Adobe or sketch or Figma, that way you’ll be fine. This is an inspiration to later on make your own adapted version.

[hoary garden]

Yea I got it !
It’s just an fb page I’m opening now to learn meta ads

So it’s totally fine

Thanks for ur help ♥️

[grizzled iris]

Oh that’s alright mate

[grizzled iris]

All the best, and if you get stuck with DALLE you can always come back here and ask 🫶🏽

[hoary garden]

Sure !
God bless you, thanks a lot ♥️

[grizzled iris]

Wow 😮

[glossy scroll]

It gives me a boost when I'm riding my bike

[late blade]

You have free healthcare, but you are forced to see the doctor every day

[grizzled iris]

LOL we will make sure this doesn’t happen in reality but we can make a fake TV documentary on this

[late blade]

hehe, i had to give it my own twist

[grizzled iris]

Yes, it is very much your MO

Well, don’t forget the letters proper

[late blade]

oh they don't have time for that

[grizzled iris]

I love pickle, gherkin

[grizzled iris]

Loool

[grizzled iris]

Great Catsby

[thick smelt]

have a subway or as odds1out call it sooubway
Prompt: turkey footlong sandwich with tomato and cheddar slices

[dim cradle]

thanks, for the theme, going off on a tangent around an overflowing mind that is in a chaotic yet patterned manner, serving as a metaphor for AI, so that it's like a sustainable extension of the mind, not a source of information overload.

building on computational imaging, and other techniques for hyperrealism, i'm discovering what the model can do to imperceptibly embed human perception altercation.

[warm helm]

does anyone know what happened to the real or fake channel here?

or is it still here?

like if they removed it then its such a shame. it is super fun

[plucky hare]

Lots of older DALL·E channels were archived some time back, including #real-or-fake. You can still view their archived content but you may need to enable archive access in Server & Roles.

[grizzled iris]

I’m so hungry now oh. no 😧

[grizzled iris]

Reminds me I haven’t eaten breakfast or lunch yet so thanks 🙏🏽

[thick smelt]

yw

[grizzled iris]

Might have to go get a subway

😂

@hoary garden DALLE gave a few more bonus ones I forgot to post before

[thick smelt]

based on a food theory video, i made this from the microsoft designer dall-e

[dim cradle]

optical illusions, subliminal messages -- hmm

i gather from the copilot pro website that "bing image creator" is the former name and designer is now the new name for using dall-e through microsoft

clarifying because it's a common support question for this channel

[late blade]

is that the farewell present from Mat?

[dim cradle]

nice, lychee, green tea, and pine flavor all in one ice cream

[late blade]

well that was the theory in the video

[dim cradle]

btw, thanks for digging into that faq about mickey and that tradelaw

[vital gull]

It was your ping! And here i thought I got ghost pinged lol I actually was planning to watch, but then I got lost into movie suggestions and ended up listening to a book🫡

[warm helm]

oh thank you sooo much!

[glossy scroll]

Eat this

Roast beef sandwich

With sliced avocados

[late blade]

what sauce?

answer carefully

[glossy scroll]

Gnome sauce

[late blade]

that can break friendships

ok, never had that, so no clue

you got lucky

[glossy scroll]

It's a mushroom sauce

[late blade]

I like the first one because it has bigger holograms for my thoughts

errr, thought, singular, thoughts it's too many in this case

[glossy scroll]

I don't like both of them

jk

I like them both

[dim cradle]

yeah, little holograms of vscode windows

[late blade]

well the right one is orderly, I like it because it's less of a problem to work in a tidy head, but the other one is more creative

[dim cradle]

if you see any optical illusions, or experience any subliminal messages i may have encoded in those images, please let me know, it's just an experiment, thanks xo

[late blade]

does this skin rash I suddenly have on my face part of that effect?

[dim cradle]

if you start clucking like a chicken, that sort of thing

[late blade]

ah, I'm safe, got already a lot of that kind of stuff built in as a feature

[dim cradle]

lol

[glossy scroll]

The optical illusions must be amazing because I can't seem to detect them

It must be my eyes

They're weak

[late blade]

oh wonder if you can make the kind of puzzles that you have to focus your sight out of the paper plane

those 3D perception puzzles

[glossy scroll]

Isn't that an optical illusion?

Or a focus thing?

[late blade]

both

or something like that

they used to come sunday mornings with the newspaper

[glossy scroll]

Lol

They tricked you, it was a mental placebo

They said if you stare at it long enough you'll see a bird and you stared until you believed it

[thick smelt]

yum...

[late blade]

I dunno, it was a long time ago, before I even had my first 33.6 kbit/s dial up modem

[grizzled iris]

Oh no this is really nice looking 🫶🏽

[glossy scroll]

You want to eat it?

You can buy that sandwich for $20 USD from me

I'll make it exactly like that

[glossy scroll]

Was this before or after WWII?

[grizzled iris]

Oh that just reminds me guys, I’m gonna do some “neuralink design updates” inspired art using DALLE this evening

[late blade]

probably, I lost the conception of what time is

[glossy scroll]

Ah yes...that does tend to happen

[grizzled iris]

Spot on

[dim cradle]

what it is what it be though

he has manly brows

[late blade]

unless you want to enclose the whole face, the other ones were on spot with the brain

[dim cradle]

what's that about the whole face?

[late blade]

I mean, you want to enclose the face with creativty or just focus on the brain?

[dim cradle]

hmm, i'm currently going for a facial expression of fascination, hadn't thought about that yet, hmm...

[late blade]

ahaaa

[dim cradle]

i can't really achieve the lifelike brain tissue i'm going for

so many possibilities

[glossy scroll]

Take notes from Hannibal Lecter

[dim cradle]

that scene leaves a lasting impression (no spoilers)

[late blade]

wasn't there someone doing an anatomy dall-e custom gpt?

maybe that can help

[dim cradle]

i've created a custom for internal structures, anatomies and inanimate objects, but dall-e struggles with scientific accuracy to some degree -- maybe their custom gpt has better knowledge files...

[late blade]

so it was you!

[dim cradle]

oh maybe so lol https://chat.openai.com/g/g-W2oE0E5J6-visionx this was some time back, so i'd change a few things now, i think, but it wasn't a serious use case, just playing around, though some of the gens aren't bad.

[late blade]

ya, I got the gpt in my bar, it's a really good idea

[dim cradle]

yep. this dude's alive, though... and some other stuff going on in the prompt, you know how it goes

thanks a lot

[late blade]

yeah, we have to add fluids

[glossy scroll]

[grizzled iris]

Can’t see them too - lol

[grizzled iris]

I am going to try that, that’ll be cool to generate!

[late blade]

let me be the first one to try it out

[thick smelt]

for the chinese new year
enjoy~!

[grizzled iris]

Prompt : Another ex. please

😂😂😂😂😂

[grizzled iris]

Woooo

Chinese New Year will be cool in 2024

🇨🇳

[dim cradle]

eh eh eh eh

[thick smelt]

a bit too late for 2023 cny but it's worth it

[grizzled iris]

Let’s see what images you generate 🧬

[grizzled iris]

😮

[grizzled iris]

That’s really good 👍🏽

[late blade]

oh I thought you were the ones doing the images... I don't know how those puzzles are called

[thick smelt]

ty

[empty kelp]

#daily-theme message

response = client.images.generate(
  model="dall-e-3",
  prompt=THE_PROMPT,
  size="1024x1792",
  quality="hd",
  style="vivid",
  n=1,``` )

figured out how to get portrait orientation working 100% of the time

[dim cradle]

we bout to turn it up

[late blade]

that's really cool

the only problem is you are naming it THE_PROMPT that brings a big concern to naming variables....

[empty kelp]

i was using that prompt in like 1985. it's the good one

[dim cradle]

i don't know if you can see that one in uncompressed form, but i think it helps the effects

[late blade]

fair enough, still concerned

[empty kelp]

saying the viewpoint for an image is "on the water" works perfectly to get half the image above and half below the water. i tried that in different ways before, but it didn't work nearly as well

Goodnight! 💤 🛌

[grizzled iris]

Good night 🙏🏽

[grizzled iris]

That should be the winner of todays daily theme 🌋

Current Theme:
🌋 overflow - intensity, passion, abundance. a river of emotion!

Version one ☝️

lol

My prompt was “No, could you make the text readable please and make it more cool and “hyper-realistic””

Still not readable…

Ok

Third try is the best

[dim cradle]

not i understand what you mean .... i'm going to try something like that... brb

[grizzled iris]

Wow

https://cdn.discordapp.com/attachments/979617962728226826/1198671827535405136/cde79484839dccb9cda3e7a9089609da.webp?ex=65bfc121&is=65ad4c21&hm=9eee4f822f217828c34ed9d963d92aae1e40f77b7433768892a3bba7eef5af0f&

Nice 👌🏽

I’m actually taken back by this image lol

The fire nation avatar!

The fire nation

Avatar

The last airbender

[dim cradle]

interesting...... showing a lot of potential...

i may need to clarify a couple things...

[grizzled iris]

What image do you want to create?

Let me know what you wrote

So we can figure out

[ornate mulch]

2% ??

[plucky hare]

Hello, you can't use DALLE on this server. You can use DALLE with a ChatGPT Plus subscription at chat.openai.com. You can also try DALLE for free using Bing's image creator with free daily credits. If you want to try this, go to Bing, then Images, then Create.

[grizzled iris]

Yeah I know - I’m really bad at battery management 😂

Now we will try Air Nation

Now we have the fire and the water

Then it’s the earth nation

In the same universe

Ok now earth nation

Let’s see…

And that completes the whole 4 major nations set 🙏🏽

Now we need the characters to be more hyper-realistic

Wow

Now we fix the text

Hmmm problem

The water is labelled fire lol

[late blade]

Someone's trying to do a weird version of captain planet?

[grizzled iris]

Yes it’s expanding! Love it 😂

Nice

Making age 25 version now

We did it

[dim cradle]

At some point Discord will need to keep up with quality. What I’m sharing here doesn’t look as good as what I’m seeing locally in the webp format I think.

[grizzled iris]

Ha ha discord probably doesn’t want that because of this exact reason of this server

[dim cradle]

And that makes a big difference in some of the latest AI capabilities I’m testing now.

[grizzled iris]

Imagine how much image heavy images we will all start sharing here

Unless they convert the format to a less heavier, file or format

Think it’s time for them to change to AVIF image format, which is much less heavy, but good

[dim cradle]

I know it’s a big data problem but it’s starting to become a thing.

[grizzled iris]

It’s just simple storage, space

But yeah, I think discord needs to change their file format to AVIF

[dim cradle]

Like maybe make it an option for some threads for a higher tier or something

[grizzled iris]

And then OpenAI needs to also change their to AVIF

[dim cradle]

We have enough server boosts to elevate higher yet if they added more levels

[grizzled iris]

But that requires me to send out two emails, but as it’s Sunday I cannot be asked to do that rn

[grizzled iris]

That could be a good idea

[dim cradle]

I’m just speaking generally cuz I don’t know anything about the admin of discord.

[grizzled iris]

But then again, you know, I’ll rather it all be free

More levels more inequality, gap

Because some people might not have that

So why enjoy an application where some people don’t get the most out of it just because they don’t have enough

They just need to upgrade their tech

They’re behind and need to catch up and integrate to quantum API servers

Everyone’s behind

😂

Anyways

@dim cradle image suggestion

[dim cradle]

I’m sure I don’t have all the answers for how it’ll play out, but phones are going 48 megapixel now so they have to find some way to keep up and satisfy all the tiers at the same time. I suppose.

[grizzled iris]

I’ll make an image to explain

What I mean

[dim cradle]

i think i understand the issues

i'm just conveying the point that i wish i could share what i'm seeing with you all

perhaps some day

[grizzled iris]

Look

This is not really it but DALLE tried 😂

[grizzled iris]

With Nueralink we will be able to be like dolphins 🐬

[dim cradle]

it holds some fascinating potential

for now we have right and nonviolent communication

it usually works when both sides are being reasonable

[grizzled iris]

Looool 😂

It got the graphic bit right

But the text wrong but it’s ok - after several tries we will get the correct text also lol

@dim cradle there that’s the best explanation ^

[dim cradle]

it's a good thing we have pictures, i don't think i would have understood otherwise 😉

lol i will come back to that little mind project later (the piece i was working on) -- we're on farmers now...

[late blade]

ok, that's the most creative image I've done between yesterday and today... #daily-theme message

I can sleep without regret

don't feel creative lol

[dim cradle]

the theme has taken me to 1536

[late blade]

what happened in 1536?

I'm not that old

[dim cradle]

Pilgrimage of Grace rebellion against Henry VIII

[late blade]

aha

[dim cradle]

by farmers

[late blade]

I'll have to check that at some point

[clever jungle]

Yes I like to save my images in water format

webp:
Watery ethernet bears photo

[grizzled iris]

@open trench hope your keeping well there bro

[grizzled iris]

😂😂😂😂😂

[dim cradle]

haha

[grizzled iris]

lol

[dim cradle]

it's fine, it's not a priority, i was just happening to be thinking about it when i was testing some image generation capabilities that would benefit from a less compressed format

[grizzled iris]

That turned out great

Ignoring the spelling errors

[dim cradle]

got some good starters on the historical stuff, question is, do i post textbook or iterate to make it more interesting. i guess i could post both variations.

[humble sonnet]

compression scale 100 Scott the Woz AI

[grizzled iris]

These I can’t compute

What’s going. On in these images

[dim cradle]

there's an issue with the joper device

[tall mason]

Looks to me like the red arrow is showing avif as being smaller than jpeg.

[grizzled iris]

Yeah it’s not got the grasp what I’m trying to create

Finally

The words are much clearer now

Smaller font but still clear to read 😂

Found the prompt!

You have to say “Exact Spelling : ‘Insert Word Here’”