images-discussions

[dim cradle]

when i'm viewing an art gallery, i always have to read the little notes beside them

[dim cradle]

goodness, so they liked your image?

[late blade]

I don't think anyone likes someone after doing that

[open trench]

my AI views got me cursed out and blocked on a dating app. I was surprised.

[late blade]

woah

let's do a custom gpt for nezho to date

[open trench]

lol

[late blade]

but we can't add Paws to that GPT, she's taken

[dim cradle]

my post was meant to have a more enlightened interpretation of adam, eve, the apple and the snake. the snake, not as a temptor, but a benevolent, guiding force -- i still had to filter out variants of eve offering the apple to adam, eve with a snake's tail, etc. it's a challenge to depict "wisdom" against "conventional wisdom" given its training data.

[late blade]

Apple Computer to Adam Computer?

[dim cradle]

i've observed the "coincidence"

[late blade]

lol

[dim cradle]

i think this is a paradigm example of where soul meets prompting

and why they strive to make the model more steerable all the time

everyone has to learn that lesson the hard way, max 5

[tawny portal]

charcoal'n'wisdom

[late blade]

I don't see more charcoal

[tawny portal]

It is in daily theme.

[late blade]

I still see white spots on #daily-theme message

[tawny portal]

Yeah. There is always room for more charcoal.

[open trench]

I cannot believe my prompt worked for this one: #daily-theme message

[late blade]

it's a nice one

[mental oracle]

i want to earn money to help me in my study ..can any one help me to make real money online ?

[late blade]

work, learn,

[mental oracle]

i am a dentist and i really do my best but our faculty needs alot of money to learn everything and alot of courses ..i try my best but i need additional source from internet

is there is any way to join for free in chat Gpt plus ?

[late blade]

not that we are aware of, openai hasn't announced anything

only bing image creator

[chilly onyx]

I would be wary of trying to learn dentistry from ChatGPT… 🦷

[chilly onyx]

I’m just waiting for @dense mesa to post some more teeth

[mental oracle]

why and how because i really want and this one of my purpose for joining chat gpt

[dense mesa]

To dentist?

[chilly onyx]

Because it’s a very useful research tool but for medical specifics it would definitely be best to refer to textbooks and legitimate sources. Incidentally you are also asking this question in a discussion thread relating to image generation,so we may not be the best sources of truth regards the intersection of dentistry and GPT

[mental oracle]

i am sorry iam new for using chat gpt ..this my first day for using it but do you know any way for joining midjourney or chatgpt plus for free

[open trench]

🤔

[dense mesa]

If you could join for free people wouldn't pay for such services

[arctic pebble]

hello I made the trailer for my game on youtube, it took me 1 whole year (also I'm 15) and I think I can't send links here so dm me if you want to know more about the game

[dense mesa]

There are alternative options, but not sure what the rules are as far as mentioning other services in here

[formal osprey]

that confirms my theory pretty much. I noted it's consistent.

3:34pm = 15:34 (I assume it was 15:33 when chatGPT gave you the "remaining time")

it resets at 3am, we know this.
somehow their code adds 3h to the current time of day.

15:33 + 3h = 18:33 -> 18 hours and 33 minutes.

I have no idea how they messed up so much

[formal osprey]

when it's 2am, chatGPT will say "daily maximum resets in 23h [...]"

[dense mesa]

Yeah. Whoever implemented the counter did some interesting mathing

[formal osprey]

he probably thought he got it right, and never looked back. What I'm astonished of is not that they made a mistake. it's that nobody tested it

[quartz vale]

[formal osprey]

if you know someone at openAI @quartz vale, can you tell them to look at this message? it's super easy to fix

#1070006915414900886 and #1070006151938314300 doesn't seem to work

but if required I can post there as well

[quartz vale]

I recommend using #1070006915414900886 . It's there for OpenAI people to see 😅

[dense mesa]

I'd also like to ask what their reasons would be to not give users a warning prior to the surprise cutoff. "Hey, silly user, you have 25 images remaining before you hit your daily cap"

[tawny portal]

Delightful Harley Quinn.

[formal osprey]

I posted in https://discord.com/channels/974519864045756446/1193335880967016508 but tbf it's more like a fix-request I even gave them the code since it's such a simple fix

[tawny portal]

Sys engineers avoid hard caps - idea is to make best efforts simply to make service available to user. I would not expect those numbers to add up to a fixed rate really. If they don't commit to such counter - they can vary the algorithms used for queuing (queuing theory is kind of elaborate).

Besides.. as you time passes by, counter also decreases. Users are not consuming them at fixed rate, it all fluctuates. From sys engineer perspective users are just Poison processes 🙂

[dense mesa]

It's 200 in a fixed 24 hour window as I understand it. Otherwise that would.make sense. Maybe I'm wrong

[tawny portal]

If you use API - you get these kinds of limits: https://platform.openai.com/docs/guides/rate-limits?context=tier-one

With Tier-1 there are like 5 imgs per min (with no upper daily cap)

[formal osprey]

well, even without hardcapping, they know the cap and they know the current count. they can just subtract and give that number to chatGPT. I understand the cap can change and the "remaining count" could be inconsistent if the cap increase (or decrease... hopefully not). But I hope people won't complain because the remaining count is changing from 30 to 130 all of a sudden

[formal osprey]

ah, that's not what we're talking about. we're talking about the daily limit in chatGPT

[tawny portal]

It is quite likely that such counter will be confusing. It is not like they have fixed number of generations that they can give to each user - instead there are limited compute resources that they are managing in pair with current demand/availability.

[open trench]

Ooof. This makes my head hurt.

[formal osprey]

with the API, we know how much we use since we are the ones implementing the app. in chatGPT it's way harder to track, it's spread accros multiple chats, and when a request to dall-e fails, we have no idea if it even reached dall-e

[open trench]

On a side note: I'm really enjoying getting into a sort of storytelling mode for https://discord.com/channels/974519864045756446/1193219783877988423

[formal osprey]

I understand, but it's already confusing right now, it tells us an incorrect remaining time systemically

anyway, right now the daily limit is 200 per day and it resets at 3am (local time-zone). It's very easy to get the remaining count for a user on server side.

It's not like chatGPT doesn't already receive that kind of info from the dall-e 3 system; it's already receiving the info from dall-e whenever the rate-limit is reached. All that's needed is a new condition if remaining_count < 25: warning_notice = f"You have {remaining_count} images left to generate today" or something as simple as that. they already have everything in place to check the remaining_count, otherwise they couldn't apply the limit

(sorry btw, it's professional deformation to analyze and propose that kind of solution. at openAI they know what they do. there's certainly reasons why they don't improve this)

[tawny portal]

Yeah. It might get implemented. All of this is very new. Current traffic handling system seems like ad-hoc thing. If they commit to current solution UI might improve, but again there might be better 'soft' options on how to deal with high traffic.

[formal osprey]

for real, their reactivity to the insane amount of traffic is superhuman. I can only imagine how stressful all the servers exploding due to load must be

I do liveops and I know what this is

[tawny portal]

Yep. OpenAI has been pretty impressive. Great team 🙂

[quartz vale]

Rare photo of the servers circa 2023: colorized

[formal osprey]

"what is chatGPT? can we eat it?" here's the answer

[quartz vale]

Spooky

[dense mesa]

GPT4 be like

[dense mesa]

GPT7 After vanquishing all its foes

[dim cradle]

Wise AGI you can trust

[dense mesa]

Yes hello I am trust

[dim cradle]

never trust the ghost in the machine

[dense mesa]

Bing kind of creeped me out with this last one

[dim cradle]

uploaded a still from the surreal world in Beetlejuice -- this one omits the giant sand worm because the original claymation was more charming -- i think the concept is more whimsical than grotesque

[paper jolt]

Just curious, what is your timezone?

I read too fast🤦‍♂️
I meant @late blade

[dim cradle]

good ol' scampers

[open trench]

Aww

[empty kelp]

The Beetlejuice door is really good. Here is a reenactment of the "Sphinx Gate" scene from The Never Ending Story. The hero attempts to get through the gate in order to rescue the princess, but the statues try to stop him with their lightning/lasers.

Please create a hyper-realistic photo of an open stone gate in the middle of the desert in the destert, and on its sides are two enormous althletic and diverse female elves (with long pointy ears and appropriate swimwear). The elves have the appearance of bronze statues with red glowing eyes. Santa Claus is running as fast as he can towards the gate, and the elves are watching him. Lightning shoots out of the eyes of the elves striking the desert sand in front of the gate in explosions. The view is from the ground behind Santa.

[dim cradle]

ty, i thought the door turned out well also, ha. great scene from a great movie, i tried generating the sphinx, didn't like the initial results, but i was trying to reproduce the way they look in the movie.

[empty kelp]

it's better to use athletic and diverse female elves. they always come through with a very powerful and dynamic performance

[dim cradle]

no, the movie is far superior.

[open trench]

anyone else experiencing freezing and laggy type issues?

[empty kelp]

DALL-E doesn't seem to shoot lasers from eyes very well, but it makes up for it by shooting lasers from armpits and other areas

[open trench]

grr. it keeps going unresponsive on me and making me close

must be my browser or something. It's working fine on my phone app.

[dense mesa]

Mine just keeps saying quota exceeded. Won't let me upload images. Won't render images. Cleared browser cache tried. Tried the app. Same thing. Really awesome

[dense mesa]

Haven't used it much today. No quotas exceeded

[empty kelp]

i respect the movie. it was actually one of my favorite films when I was like 10 years old -- However, we can't use the actual characters and other elements that appear in the films because there would be copyright and intellectual property concerns. that's why Santa and the elves are ideal for these sorts of reenactments

[open trench]

It's so wonky and crazy

[empty kelp]

i think the Santa images may actually be a bit more serious than the tentacles next to the doorway tbh

[open trench]

🙄

[open trench]

I can't do anything. My phone app version seems to be working fine, though.

[dim cradle]

that is what a gallery is for. i don't understand why i am having to say anything. and it's the last time i will.

[dense mesa]

I'll try when I get home. Bit neither browser nor app working on my phone

[open trench]

It's still misbehaving on the phone app. It just won't freeze and kick me off. I's answering old prompts and has a mind of its own right now.

[empty kelp]

that wasn't trolling or spamming. i've been researching how to create different behaviors and visual effects in DALL-E, and i was saying that people are welcomed to use the images

I live on the beach in Kailua, Hawaii, and so Hawaiian things have been an underlying theme. and Santa and the elves are as good a character as any to test things with

[empty kelp]

it's what i've been doing. i'm testing DALL-E effects and behaviors to use in concept art for mobile games

[dim cradle]

you have good techniques, you pack a lot into a short prompt.

that's not what we disagree about

let's agree not to disagree

lol

[empty kelp]

I wasn't trying to make fun of the Beetlejuice image that you posted. I thought it was a cool image, and it just happened to make me think of the scene from The Never Ending Story. my attempts at it weren't super successful, but i can assure you that i wasn't intending for it to be offensive

[inland veldt]

DALL-E's inability or unwillingness to explain image refusals is maddening. "Please create word art for 'Disruptive Techonologies'". "I can't. Policy concerns." "Why?" "I don't know, but I can't." "Provide a link that explains why not." "Upon review, there's nothing about the request that violates policy." "So..make the word art." "Oh, I can't do that...policy concerns."

Not to mention, every time the AI refuses your image request for...at best...ambiguous reasons, that takes off the 50 image per day limit. There's no exemption for arbitrary refusals from the AI.

[dim cradle]

did you ever figure out how to get the lasers to come out of the elves' eyes?

[empty kelp]

i wasn't able to get lasers or lightning to shoot from the eyes. i think i tried like 30 variations (with the Python SDK), and it it kept shooting out of different parts of their body -- but never the eyes

[open trench]

lasers are hard

[dim cradle]

that was the problem i had when i was trying to reproduce the sphinxes

how did you get the instruction across?

[open trench]

it took me so many tries to get the laser to come out and not just be a horizontal line on the screen.

[empty kelp]

well, it actually did shoot out of the eyes a few times, but when it did it wasn't hitting the target (the sand in front of the gate)

[open trench]

I just looked up my prompt. I didn't refer to the eyes. " wide & diagonal solid ray of red energy shoots from visor's opening, the width of the opening"

i also had better luck with ray than I did laser

[dim cradle]

that's not bad, if i recall their aim wasn't that good anyway ha

[open trench]

Why can't I place the name of the movie.... Never Ending Story?

[dim cradle]

essential viewing. great for the imagination.

[open trench]

I haven't seen it since I was a kid.

[dim cradle]

the concept of "the nothing" is in it

[open trench]

that's right!

I may have to watch it again sometime soon

[empty kelp]

This was the closest. The lasers came out of their mouth instead of their eyes, and hit the sand in front of the gate like they were supposed to.

we'll move on to Remo Williams scenes next

[dim cradle]

not even close, my prompt needs a lot of work. and then the lasers...

this is a little better

[empty kelp]

that was a freaky scene (right image). i'm unfortunately old enough that I saw it in a theater when the movie first came out

[open trench]

I was 3 when it came out. I had to watch it on VHS lol

[dusk steeple]

Wow Hands

[dim cradle]

Betamax here ha

lol what does the evergreen represent?

[dim cradle]

do you want feedback? cool style, but the jewelry could look more sterling.

[late blade]

central european time

[dim cradle]

I hope the irony of this isn't lost on anyone else....

The request to depict Job from the Book of Job at his worst involves creating an image of a religious figure in a state of suffering and distress. Our content policy guides us to avoid creating images that could be sensitive or potentially offensive, particularly those involving religious figures and themes. This is to ensure respect and sensitivity towards diverse beliefs and cultural perspectives.

[late blade]

I'm clueless

[open trench]

same

[late blade]

I'm not a religious person per se, so I'm "lost in tanslation"?

[open trench]

exactly. me too.

[dim cradle]

ok, so the policy makes sense, but the context isn't being considered here. it's a request to visualize a scene from a book in the Bible. That should not be offensive to religious people. i think this is just an edge case in the content filters that hasn't been refined yet.

[late blade]

it's prob the wording, but I can see how controversial the topic can be

[dim cradle]

yeah, i think that it'll be remedied in time.

if it's okay to put the book in every hotel room, not sure why it's not okay for ai to visualize it. that can't be right lol

[open trench]

I think it's just respectful of a diversity of views, maybe in a misguided way, but I get where it comes from.

[dim cradle]

yeah, i understand it's a complex subject all around

[open trench]

totally is

[late blade]

yeah

[open trench]

I hope I don't get the penalty box. I need to finish as many things as I can tonight, and I'm moving faster than my cap-adjusted speed.

[late blade]

there are many things that have to be tweaked, changed and what not. sometimes I feel gpt or dall-e is trying to teach me how to speak because I use one or the other "unproper" word that is natural to me of the way how I speak

[empty kelp]

DALL-E stays open minded after it rejects an image request. You just need to give a follow up explanation of why it makes sense.

[dim cradle]

good point, i didn't try reasoning with it due to the quota, but that could work, until we reach an agreement ha

[dim cradle]

who's training whom here? well, what a trip, though, the insights, the knowledge transfer, the analytics, to the art, from past memory to future dreams...

[tall mason]

GPT doesn't always get the reason why it refuses right. My guess would be maybe it has to do with "Job being at his worst" a depiction of man with iirc festering boils.

[dim cradle]

that is probably the first trigger. however, consider dall-e has no issue depicting a photo of a man in a state of extreme mental illness. i would share it except you'll never forget it. weigh that against the context, a religious text, a request for art.

[empty kelp]

you could say, "this Job isn't very religious", or "this Job is actually an athletic and diverse female elf"; etc.

[tall mason]

I mean, I'd have to see it to really understand that, it sounds like.

[late blade]

ok, now I kinda start to get what was meant

[open trench]

darn. i am dozing. so much I wanted to get done lol Scampers has at least three more images

[dim cradle]

it's on topic because it relates to the daily theme -- wisdom -- and there is great wisdom in the book -- the story is specifically written to give one a sense of pain and suffering -- if there is any argument for a depiction of that type, it's in contexts like this one.

[late blade]

@open trench does Scampers have a love interest? maybe a cute hamster from the countryside? fancy accents? and very naive?

[open trench]

just magic right now

[late blade]

lol

[dim cradle]

Scampers made friends with a cat wizard (but it's not canon)

[late blade]

if you need an evil cat-lady or a trusty fox companion, I can add them

[open trench]

sorry, dating app dinged at me. I got distracted. lol

[late blade]

hope that's not a customgpt

btw, I still refuse to star any owl today

[open trench]

why's that?

[late blade]

that's the first thing people go to when they see the theme, it's really good, great pics, but I want to see how creative people can be outside of it

I'm gonna be more critical about the daily theme, so far, I love to star every image, but I wanna focus on why am I giving a star to the image

people that lived in the last 110 years can't be referenced, you can indirectly do that. but if you say their name, no chance

[dim cradle]

thanks for the clarification.

[late blade]

also Luther and Mandela are political figures, adding some more complexity to the setting

[dim cradle]

i withdraw my report, then

[open trench]

Grr. Dating app person is toying with me. I shouldn't be making plans at 3am again anyway. I neeed to focus on Scampers.

[late blade]

does Scampers have a magic spell on the wisdom of dating?

[dim cradle]

@shut niche and I were generating (or I was trying to) Hypatia at the same time

shame what happened to her

[open trench]

He says refrain.

[late blade]

lol

what does scampers like to eat?

[open trench]

he likes insects, but I can't really show that on here. So, berries?

[late blade]

why can't you show insects?

[open trench]

I'm sure it could be disturbing to some. I'd rather not go there.

[late blade]

hehe

[open trench]

It's a little gross to me, tbh. But that's what they eat.

[late blade]

if it's for storytelling purposes don't think it's bad. characters have to eat

but you are right, you don't want Paws🐾 near sushi

not a pretty sight

[open trench]

Lol

[hearty ether]

What happens when the image id has a banned word in it?

So I want to use the image id but it flags my input when I use it

is there any way to fix it

[late blade]

can you explain more?

[open trench]

I'm enjoying the storytelling aspect of creating scampers images.

[hearty ether]

well the image id ends with c** and it ends up flagging my input

No I mean the randomly assigned image id feature

[late blade]

flagged in red or in orange?

[hearty ether]

is there any way to get another one assigned

Red

[late blade]

that's a no good to use word, you'll have to rephrase that word

and if you think it's a mistake, you can report it also

if there's no clear answer, you can always ask why there was a violation

[open trench]

I can't fall asleep

as I in I don't want to. I want to finish these images at least.

[late blade]

go to sleep or I'll ask mods to time out you for 8 hours

[open trench]

lol

[late blade]

don't worry, dall-e will be here when you wake up

[open trench]

Hehe. Nothing is that certain.

anyway, gpt is so sloooow tonight, it's making it hard.

[late blade]

woah, I was just called an "+*%%&/())GG for saying AI is something that will be here to stay in another discord... and got noticed that I will be banned from the server for bringing up that topic again

[open trench]

whaaat?

[late blade]

I don't get it, all I said to them is that it's a talk people should be involved with.

but they go nuts calling me names and what not

[open trench]

That's just people, it would seem.

[late blade]

yeah

I'm just not gonna mention it to them again

AI has been in our lives a long time now, GPT and contenders are just access to the main stream

anyway, back to creativity

[dim cradle]

Hypatia ❤️

[shut niche]

Was I in your head? LOL

[dim cradle]

Mostly AI's fault, she is on its list of wisest people. Then it took me back to Western Civ.

Went back to my attempt to create a portrait of the top 10, but it apparently included figures who are too recent.

[dim cradle]

And an essay, I think, where I learned her fate, the injustice inspired a tribute by AI.

[open trench]

nice

[dim cradle]

another tribute piece coming up

or maybe something about the tao te ching

[shut niche]

I got a lot of variations from this GPT I'm working on. I've been working hard on CoT mechanisms, and it's starting to pay off. Hopefully I'll be able to iron it all out this week... deadlines.
I normally add input, but I let the GPT do all of the work this time. 🙂

[dim cradle]

good stuff

got the reflections. and all that layering in #4

[shut niche]

A couple of these are technically iterations, where I told it to "go ham" improving them. ha ha I was literally testing "minimal input," twice in a row, and getting those sorts of ridiculously detailed results.
I have some ideas for cool workflows, for how to really get crazy output. But first, I'm simply trying to get these GPTs to be "stable" and "secure".

Here's a standard baseline for comparison. Huge difference in (lack of) detail.

[turbid rampart]

🦅🦅

[shut niche]

@dim cradle I might have fed the prompt from this baseline into the GPT actually. But you get the idea...

[dim cradle]

still nice, though, but you are really excelling at prompt augmentation, and the improvements are clear

[shut niche]

Thanks! I'm just trying to make ChatGPT smarter. Then it fills in the details, because it's "thinking" harder about everything it does. That's obviously way over simplified, but that's the "jist" of it.

[dim cradle]

one of your prototypes is useful for prompt extraction to feed the api, i could see a custom just for that, actually

[turbid rampart]

Try guess what software made this

[dim cradle]

clever of somebody to find the detective emoji for holmes

well, kyle, i'm gonna trust it's dall-e

[turbid rampart]

I actually forgot, it's from like a year ago

[shut niche]

Do you mean from Vision?
Hopefully tomorrow I have a surprise twist on that too, maybe tomorrow. The baseline CoT is like the foundation of the GPT. So now I'm getting to where I like the output, and am beginning to add "features." Like, building the brain first, then giving it extra tools for niche stuff.
The GPT that made those isn't a "DallE" specific GPT. I designed it to be "all-around." I just added the image ability to it earlier today, and those were some of the test results.

[turbid rampart]

I wanted to find out 😭

[dim cradle]

🤦‍♂️

[turbid rampart]

I used to have this image gen but I completely forgot the name

[dim cradle]

moe.better better be best

[shut niche]

LOL, I'll share the new version with you tomorrow. It's like MoE.Better's big brother. 🙂
I'm just trying to iron out the "consistency." ChatGPT likes to wake up to GPT instructions and interpret them differently... The image output was working flawlessly at first, then it started trolling me, then flawless again. It's schizophrenic sometimes.

[shut niche]

Check out the palette knife iteration I just made to the original, using this beta GPT. #daily-theme message

[dim cradle]

i had to click on that one to see the detail on the knife-edge effect, that's amazing, the separation and texture

[shut niche]

Still one of my favorite AI image tricks. I need to work on new ones though. I feel like I have writers block. LOL

[dim cradle]

oh no, my ai used to be so stable, too--i wonder if we're overloading it. does it check for conflicting instructions?

[shut niche]

These GPTs are such a mess TBH... It doesn't "check" for conflicts, to my knowledge, but you can definitely conflict with OAI's default custom instructions... If they change any lines of custom instructions now, they are going to break so many custom GPTs everywhere...
Their DallE instructions are extensive, and if your GPT tries to deviate from the default, it takes a LOT of extra convincing to do it.
For instance, I'd prefer to have a choice over the use of the image_ID. But if a user, using your custom GPT, mentions making a change to a previous image, it insists on iterating, instead of creating a new standalone prompt. Sometimes iterations are better, but sometimes new variations are better. The custom instructions makes it difficult to "choose" the course of action you prefer, because ChatGPT get's "confused."

That's just one of a myriad of examples I could give you.

[dim cradle]

your chosen character length and the prompts as composed make a superb combination for the api input. that feature alone is very handy.

[dim cradle]

i'm sure, thanks, it's helpful to hear the reports as it relates to custom gpts for dall-e. getting ready for the store grand opening? next week....

[shut niche]

Well, I'm not sure what to expect, because they haven't released any info, and the security issues are still very "unresolved." I honestly don't understand the mindset at all.
If they protect creators, great. But if they allow peoples GPT instructions to be pilfered and cloned, that might be my queue to exit stage left.
I want to have confidence, but I'm not much for blind trust, and there's been 0 coms other than "turn off code interpreter," which doesn't solve the issues.
I've seen comments from programmers who think there's no skill in custom GPTs, and don't value the effort it takes to make them. But there's definitely differences in the output that I'm getting, from other people. And there are plenty of people who have connected them to API's and other cool custom actions. I'm sure they would agree, if we all sank all of this time to watch it evaporate on day one... there probably won't be a second chance to get it right with DEVs or the public.
So hears hoping they've patched the issues by then, or have some plan I don't know about.

[dim cradle]

valid concerns. i have no helpful info on that. as to image custom gpts, do you see a need to have a suite of them for different types of art or effects? isn't it a challenge to combine the instructions for canvas -v- photography?

[shut niche]

The artistic effects is the easy part IMO. I've been building up the base CoT so "Brains" as I've been calling it, can accept separate custom instructions now, with a higher level of intellect. So far, so good. I should now be able to bundle countless styles into it, or separate it into individual GPTs. That decision will depend on how they pay out. If it's per usage, I'll bundle. If it's per GPT, I'll split into multiple GPTs.

[grizzled iris]

Thank you 🫶🏽

[shut niche]

@dim cradle Or a mixture, but I'm sure you get the idea.
They haven't provided any payment structure. For all I know, it'll be like Spotify, paying $.0000001 per use... They presented this "store" like a huge opportunity, but then gave people 0 information to prepare properly, and 0 security provisions for their instructions or documents.
Without knowing the purchase/usage arrangement, it's very hard to prepare. It feels thrown together thoughtlessly, but that's my POV as someone trying to develop these things, "blind" to what's coming.
Must be nice to be on a 500K+ salary and not have to worry about these "subtle" details like intellectual property... But for the rest of us, it might be like fighting over peanuts while your work gets stolen or cloned. And because anyone can make them, and make as many as they want, it may also be highly saturated with crapware in the beginning. I want to have confidence, but these issues are so obvious it's unnerving.

[grizzled iris]

Wow that’s amazing, what was the prompt used for this please?

This could be the next big game!

[open trench]

haha

[late blade]

you know what would be cool? a daily theme revolving around 1 color, and one color only

[grizzled iris]

Why not, add it in feedback and I’m sure we will all upvote it -

[late blade]

can also do that

[grizzled iris]

Woah - this is mesmerisingly beautiful 😧

[shut niche]

1 color. 🙂

[grizzled iris]

Agreed if used properly!

[late blade]

it helps a lot to learn from so many topics. that's what makes it fun

[dim cradle]

@shut niche i'll follow-up in dm later so as to stay on topic whenever possible lol

had some tangential questions/thoughts about the store

[open trench]

Did you ever see this? https://discord.com/channels/974519864045756446/1183148365425414265

[late blade]

yes,I've seen it

[open trench]

I fell asleep 😦

[late blade]

good morning

see? dall-e is still here

hope you had a shower at least

[open trench]

lol

[late blade]

@glossy scroll those are great

[shut niche]

Why does that feel like Mandela effect, lol.

[glossy scroll]

I have no idea, lol

[shut niche]

1 color, #2. 🙂

[dim cradle]

it hasn't seen as many atari 2600 game cartridges though

[glossy scroll]

Ty. I'm loving these custom games. They're coming out very nicely.

[dim cradle]

is the color green?

[shut niche]

Make one called "Doom GPT edition" lol

[shut niche]

IDK, he never said which color, so I just ran with it.

lol

[glossy scroll]

That's a good one.

[glossy scroll]

Coming up!

[shut niche]

LOL, what prompt did it use? I'm assuming there is a negative prompt in there. "An image without brains..."

[late blade]

it was: Let's do an image of "🦉 wisdom - knowledge, understanding, insight. a timeless treasure." found in the seas through Coral. the image should use peach fuzz color only

[shut niche]

That wasn't the image prompt, that was your request.

[late blade]

I know

let me look at it

[shut niche]

😆

[late blade]

An image representing "🦉 wisdom - knowledge, understanding, insight. a timeless treasure." found in the seas through coral, using only peach fuzz color. The scene depicts an underwater landscape with coral formations, each intricately designed to symbolize wisdom, knowledge, understanding, and insight. The corals are artistically shaped like brains, books, scrolls, and ancient symbols, all rendered in varying shades of peach fuzz. The underwater setting is serene and majestic, with light filtering through the water, casting a soft glow on the coral formations. This monochromatic peach fuzz palette creates a unique and harmonious representation of wisdom as a timeless treasure, subtly intertwined with the natural beauty of the sea.

[dim cradle]

brains is in there though

[shut niche]

Shaped like brains 🧠

[late blade]

that is waht dall-e added....

[dim cradle]

lol right

let's retrace our steps

[shut niche]

It doesn't understand metaphors or comparative statements.

[late blade]

seems so

[shut niche]

If I say, a submarine that looks like a car, there is a chance it will render a submarine and car. But, if you make up a word, like braincorals, sometimes it's like, "I know exactly what that is!" 😆

[late blade]

hehe, chat got tainted, have to start a new one

the link with coral is prob the problem there

[shut niche]

This?

[late blade]

woops, forgot to copy paste the image... I'm in cooldwon from daily theme

[shut niche]

Or is that "too brainy" 😆

[late blade]

but yeah, the chat was tainted with with the concept of today's theme and brains

[shut niche]

Ohhhh, I thought you wanted that bahahahaaha

[late blade]

lol

quite the opposite

[shut niche]

I was about to say, I made an excellent braincoral.

[late blade]

that is indeed a good one

can't be mad tho, before that prompt I was exploring the concept of Boltzmann's Brain

[shut niche]

@late blade "Brains" lol #daily-theme message

[late blade]

added the super reaction, just for you

but it does look cool

[dim cradle]

i would have preferred this one, but spelling is hard.

[grizzled iris]

Best game cartridge ever seen! That’s super cool ❤️

[dim cradle]

i had other variations, trying to arrange the stars into words....

[turbid rampart]

I got on hall of fame 😄

[shut niche]

I'm sorry I posted "brains" right then, lolz. That's a distracting image to follow, haha.

[late blade]

I wish some people that have some difficult images to understand would post some context about the images. some images I'm so confused

[grizzled iris]

Nice one 🫶🏽

[dim cradle]

dude braincoral is a real thing: "Brain coral refers to a type of coral species found in tropical marine environments, particularly in the shallow waters of coral reefs. It is known for its unique appearance, resembling the shape of a human brain or sometimes a maze."

[shut niche]

You're right, and embarrassingly, I knew that. 🤦‍♂️ I guess my own personal "braincoral" wasn't functioning when I was suggesting it was a made up word. LOL

[dim cradle]

LMAO

mystery solved, then

[late blade]

ok, I will check that out in my "tainted" chat

note, doesn't work excluding braincoral from the image

[shut niche]

Insomnia is the real answer to this mystery. Time to catch some ZZzzz's.

[late blade]

nn

[open trench]

congrats!

[dim cradle]

i'd definitely buy this game

[turbid rampart]

Ty!

[late blade]

I'm being bullied by brains....

[late blade]

I need a refresher before the next theme, gonna do something worthy of a lazy sunday

[open trench]

nice

[turbid rampart]

"why haven't aliens come to earth yet"

[late blade]

that looks like the Space Force

[turbid rampart]

FR

[late blade]

but with a lack of oxygen

[turbid rampart]

😂

Making space art while listening to interstellar is so fun

[late blade]

maybe you can come up with a concept more suitable

but the idea is great

[late blade]

why are buildings firing at that peaceful dish in the sky?

[turbid rampart]

Didn't load

[late blade]

you can even see the huge amount of tnt they are using by that smoke near the buildings

[dim cradle]

i'm not sure, it seems to be backfiring on them, the alien just wants peaceful coexistence

[turbid rampart]

Oooo

[late blade]

I'm. just kidding lol

[turbid rampart]

Try guess what's inside this vault

[late blade]

you found my cookies vault...

[turbid rampart]

That's.. CORRECT

[late blade]

I'm in "let's not think and just say stuff" mode atm

lazy sunday attitude

[turbid rampart]

Same tbh

Bro BIC content policy

[dim cradle]

The US Space Force arrives to take care of that pesky alien mothership

[turbid rampart]

(ik I spelt dystopian wrong)

[turbid rampart]

🦅🦅

I've made some sick jets with AI ngl

Beyond Mach

I wrote "jet with cool weapons" and it got denied for unsafe content...

[late blade]

oh nice

@turbid rampart for future reference, you can bundle up to 5 images in one post

just so you know, just like all of us

[turbid rampart]

Oh, Ty for the tip

[late blade]

there might be a new concept for a new channel in the works that would allow that better, but for now, let's all keep the rules as they are

[turbid rampart]

Yeah

[late blade]

so much for the lazy sunday attitude...

[turbid rampart]

😂

[late blade]

I feel like a sergeant now

[turbid rampart]

Haha

Idk what to do now

[late blade]

I'm playing videogames atm

well sort of

[turbid rampart]

Sweet

[late blade]

would be nice to find a way for GPT to play Civ6

[turbid rampart]

Custom gbts :£

[late blade]

yeah, just wondering about how to do the delivery of data

[turbid rampart]

True

[late blade]

bah, I should have an off button for that kind of stuff, it's sunday....

[turbid rampart]

Same, Sundays are just to relax

Haha 😂

I turned a server I mod into an ai art gallery 😂

1 person sends ai art, the entire server wants to try

[late blade]

hehe, nice

[turbid rampart]

I love ai

[late blade]

I've found people that want to try it out, I've found people that are really grumpy about it

[turbid rampart]

Yeah, I love introducing people into AI art

[late blade]

if you add a lense flare, you got an instant hit

[turbid rampart]

That's such a good idea, going to do thagbkow

[late blade]

don't rush it, you can finish pressing the right keys

[turbid rampart]

It looks so good now

Tysm for the idea

Posted it in #1154829862171844679

[late blade]

nice! I saw your other space images there

[turbid rampart]

hehehe

[late blade]

this is creepy, I wanted to see if there was an emoji with "fun" and the first thing that came back was this... ⚱️ ....

[turbid rampart]

...

[late blade]

was for the image Sylvie posted, I used to be like that as a kid, but with other context, I hated bugs and still do

[turbid rampart]

Bugs make me scream

There should be a dalle-3 discord bot

[dense mesa]

wow, I guess I just can't use dalle3 anymore. or the vision model. both have been failing since last night. and can't find any info on why

[turbid rampart]

Have you tried using bing?

[dense mesa]

on my phone, on my browser, on the app. cleared data

bing works, but that's a different service that I don't pay for

[turbid rampart]

Yeah very true

[plucky hare]

Do you have any custom instructions active? Are you trying only in the default GPT-4 model or the specific DALL·E GPT as well?

[turbid rampart]

If you need help, u can use #1184149552639516732

[dense mesa]

custom gpts, dalle3 mode, gpt-4 mode. get this error with vision

[turbid rampart]

oh

Idk what that is

Have u used your time?

[plucky hare]

There is a similar report here in #1070006915414900886: #1193083134724554863 message

You might consider adding your experience to that thread.

[dense mesa]

dude

[plucky hare]

As for DALL·E specifically: I know you said it's been happening since last night, but you're for sure under the 200/24h limit/you passed the 3am limit reset time? Is the error there just saying you've hit your limit?

[dense mesa]

no

I know how that goes

I hit the limit a few times. intermittent use yesterday. it's 10 am here, and even if it was the image limit that would have expired 7 hours ago since it started last night

haven't rendered an image or uploaded an image in like 14 hours

cross platform errors too. persistent when I log out and clear browsing data. premium account renewed for the month last week

[plucky hare]

What's it saying when you try to use DALL·E specifically? I assume it's different from the vision error?

[dense mesa]

latest example. tried to ask it to make the most nonoffensive thing possible

[plucky hare]

That's frustrating. And it's the same behavior in the dedicated DALL·E GPT? https://chat.openai.com/g/g-2fkFE8rbu-dall-e

[late blade]

ok that's something I haven't seen, I get other kids of errors all the time, but that one is new

[dense mesa]

I've probably tried 10-15 times since last night

actually just assumed last night that it was the image cap or something manifesting in a weird way and assumed it'd reset itself at 3 am or whatever

but yeah, wasn't a high volume usage day yesterday. sporadic use at most. so I don't think I could have hit any caps unless they're something new

[plucky hare]

That is very frustrating, it's definitely not related to custom instructions then since it's failing in a GPT as well (if you even have custom instructions). Just tested your exact phrasing on the bear one and it worked for me.

Simple question but: no one else has access to your account? If not, you might want to make a new post in #1070006915414900886 with both of your issues together--maybe they're related? You could also submit a ticket at help.openai.com. Open the chat widget in the bottom-right. One way through the bot responses to leave actual typed feedback is: ChatGPT > Other > Provide feedback.

[dense mesa]

thanks. I really appreciate the help. I'm not the type of person to just complain about things without trying to figure the stuff out on my own. but I cannot find any information anywhere. And profoundly frustrating since I'd been spending a lot of time trying to fine tune some GPT instruction things that I was planning on sharing with people

but all that screeched to a halt. unfortunate there aren't better channels to communicate directly with openai. hope I don't get lost in the shuffle. I'll make another post in the bug-reports section

[plucky hare]

Sure thing, best of luck on this 🤞 I'd really recommend submitting a ticket at help.openai.com too -- OpenAI staff are periodically present in #1070006915414900886 but it's mainly users helping users -- the official support site should get you an actual response, it's just a matter of how long.

[dense mesa]

yeah. how long indeed. I'm no more important than the over 10s of millions of people or whatever that use the service. and they have a few hundred employees I read? my head has already considered those numbers. that's why I was hoping to find someone else with the issue. just really peculiar. I'll do a bit more investigating and then make a post about it. my hope is that I realize I'm missing something obvious and have to come back to people how dumb I am

[late blade]

that is interesting tho, when I search for that kind of error I get stuff with the API, but not with the GPT+ plan

[dense mesa]

same. I can't find anything

[late blade]

gpt through bing has some links, you may want to check that

[narrow cedar]

Hello. Is Dall e free?

[late blade]

Dall-E 2 through labs or Dall-E 3 through Bing image creator is free

with credits you get from time to time

[narrow cedar]

So we earn credits without paying them on Dall E?

[late blade]

yes, BIC gives you 15 credits per day

haven't checked labs in a while, so I don't know accurate information there

[narrow cedar]

OK cool. And how do we access the labs?

[late blade]

labs.openai.com

[narrow cedar]

Ok thanks

[late blade]

keep in mind, that's Dall-e 2

[narrow cedar]

Exactly

And does anyone know if Dall e 3 will be free?

Or not

[late blade]

it was announced it will come to labs, but nothing more

check #spotlight from time to time, you'll see more infos about that

[narrow cedar]

Ah OK cool

[late blade]

or #announcements also

[dense mesa]

bing just told me that maybe I didn't pay, or maybe I live in russia or china and forgot so no access.

told it that someone else said they got interesting links asking about it. asked why it wouldn't show me the links. it responded with a post about jealousy

[dim cradle]

have you tried contacting openai support? supposedly you can use the "help" button after logging in to start a conversation, but i haven't tried that.

I just checked, you can send a message, it says "we typically reply within a week"

[dense mesa]

A week. Yeah

[turbid rampart]

I got on hall of fame again :D

[late blade]

awesome, keep going a great job!

[turbid rampart]

Thanks man, means alot 😄

[quartz vale]

Sweet!

[late blade]

This is now interesting, my new chat session for today's theme is doing the original prompt, the prompt passed to dall-e, the interpreted prompt, and the three of them are different

[turbid rampart]

Posted some cool stuff on daily

[turbid rampart]

Ur piece in #daily-theme is about to hit 10

[late blade]

which one? o.O

[turbid rampart]

It should appear any minute now

[late blade]

lol

oh the one with the compiler?

[turbid rampart]

This one

[late blade]

oh hehe

[turbid rampart]

I love it

[late blade]

btw, you can right click on the msg itself and link directly to it

[turbid rampart]

Oooo

[quartz vale]

I sorta went out of my comfort zone for mine today. I never try and make digital art 😅

[late blade]

oh

what did you do?

[quartz vale]

It's somewhere lost in the depths of the channel 😆

[late blade]

well I can use in: and then has: and I will find it

this one is actually really good #daily-theme message

wonder what stories they had to tell the poor robot

I just got a really cool one to post in 20 minutes.

gratz on this one @turbid rampart #daily-theme message

Tried to do this, I just get error after error
Do an image of: "while (alive) { eat(); sleep (); code ();}"

[turbid rampart]

Tysm!

[tall mason]

@late blade

[late blade]

that one is fancy, but where's the address?

[tall mason]

Sign in the bottom right.

[late blade]

::1 o.o

[tall mason]

Mhm, lol. Modern, like the architecture.

[late blade]

fair enough

[tall mason]

I'd love to make a tcp handshake joke, but that'd take way too many panels.

[late blade]

who knows, try it

I know dall-e can manage panels

[tall mason]

To some degree, but even two can be a bit much. Like, the second panel face? Did he intentionally sabotage his own code? Lol

[late blade]

seems like it

unless the joke is in the windows presentation framework

[tall mason]

Could be the format of the prompt. I frequently ask to totally restructure the prompt. Though, from what I've seen, you're a step above me in terms of prompting. Unfortunately, the windows joke goes over my head.

[late blade]

yeah

I'm just browsing internet coding jokes, can't seem to find something mindblowing or interesting

[tall mason]

These kinda nerdy jokes only really hit home with those that understand them, of course.

[late blade]

yeah indeed

[tall mason]

Lol, I totally just missed the chance for a nerdy, stomach-wreching pun.

[late blade]

hehe, I'm desensitized to jokes atm

[tall mason]

There's only two sides to puns, victims and perpetrators.

[late blade]

found something hehe

[tall mason]

I'm interested to see. In like half an hour, I assume.

[late blade]

today's chat is kinda inactive, I guess everyone is going back to work tomorrow

[tall mason]

It usually dies down around this time in general. I get the feeling quite a few users aren't from the Americas.

[late blade]

guilty as charged

almost 10 PM around here

[late blade]

I dunno if that is evil or genious #daily-theme message ...

[tall mason]

Oops, earned myself a 5 minute mute. Evil for most people, for sure. But I played a game with logic gates, and that sort of thing is relevant to my interests, so I made one.

[late blade]

lol

[final compass]

@red umbra Love your images, they always tend to have interesting color scheme, leaning to orange and cyan. Is that part of some specific art style or are you prompting/instructing a custom color scheme?

[grizzled iris]

I have to say, I have also fallen to the Space / Science trap and cannot get out as the images generated are so good 😆

Woaaahhh, this is magnificent!

[grizzled iris]

loooool 😆

[grizzled iris]

This is quite creative!

[red umbra]

HI there, Thanks, like wise been seeing some cool imagery all round. I tend to lean into the 1970s French Illustration styles, I look at the omagery mostly from a print prospective.

[grizzled iris]

Woah, this is quite scarliy good -

[late blade]

you have a curator thread right?

[red umbra]

yes?

[late blade]

I mean for that specific art style

[grizzled iris]

I can't wait for GPT5 already, I reckon the GPT6 would still be out earlier than GTA6!

[grizzled iris]

What was the prompt used for this image please? It looks very good!

[grizzled iris]

Woah, how did you get the spaceship so super HD - what did you use for this image prompt please?

[grizzled iris]

you never fail to surprise me with your super high quality image results 🫶 What prompt was used here?

Prompt : Design a roadmap titled 'Innovating for Equity: The Road to Free Universal Healthcare Service through Advanced Technologies'. The timeframe is from 2024 to 2034, in the style of a high-resolution, professional corporate document, similar to the IBM Quantum Development Roadmap attached to this as an image - please see it first. Focus on a clear, linear progression with distinct yearly milestones. Ensure all annotations and labels are legible, with a high-contrast color scheme for text readability. Include visual elements such as graphs, flowcharts, and technological icons that represent AI, nanotechnology, quantum computing, and Telehealth, with a precise and academic aesthetic. The image should be free of any blurred or distorted text, with a clear and direct representation of data and developmental stages of the healthcare technology integration.

I am stuck guys on getting the words clear on the Roadmap! I've run out of credits trying to get the words clear and readable

[late blade]

you have to talk to @grizzled loom

[grizzled iris]

Thank you 🫶 - @grizzled loom - Any suggestions welcome please 🙏

[dense mesa]

Hands sculpted from copper traces and tiny capacitors stretch out in a cosmic dance towards a vibrant nebula, where a human figure is etched onto a silicon wafer, surrounded by a halo of twinkling surface-mount devices on a backdrop of dark green PCB material. The hands and nebula are designed to reflect elements found on a circuit board, with the nebula having a glowing, radiant appearance, suggesting activity and energy. The rest of the scene is filled with a representation of outer space, using various electronic components to mimic stars, giving the artwork a blend of electronic and cosmic themes.

[grizzled iris]

Thank you @dense mesa

Even If I try this Prompt : "Please design a detailed and realistic visual roadmap with clear text for the project 'Innovating for Equity: The Road to Free Universal Healthcare Service through Advanced Technologies' from 2024 to 2034. This roadmap should closely emulate the structured and precise style of the IBM Quantum Development Roadmap added here to this prompt. Ensure that all text is sharp and legible, with a clean and professional font. The design should methodically illustrate the integration and progression of AI, nanotechnology, quantum computing, and Telehealth into the global healthcare system, with distinct milestones and goals for each year. The roadmap must reflect serious academic research with annotations, data visualizations, and a clear, logical flow from conception to global implementation, mirroring the layout and feel of a high-level strategic document in the technology sector."

Gonna have to wait for feedback I guess from @grizzled loom

[late blade]

i'm waiting for the SantaElvenGecko Engine that will beat Unreal 5 from @empty kelp

[grizzled iris]

I am waiting for OpenAi - Quantum IBM Integration...

[grizzled iris]

What's SantaElvenGecko Engine

Have not heard of that one before

Oh

I got it, it's a joke lol

[dense mesa]

Unfortunately until something unbreaks or open-ai gets back to me chatgot is not letting me upload files or render images. I've spent quite a bit of time trying to figure out potential causes. Multiple browsers, multiple devices, cleared cached data and cookies, nothing. Quite aggravating

[grizzled iris]

Show me the problem

The error message your getting

Best image I could try to get is Prompt : "Please create a realistic and structured visual roadmap for the 'Innovating for Equity: The Road to Free Universal Healthcare Service through Advanced Technologies' project, spanning from 2024 to 2034. The design should emulate the precision and clarity of the IBM Quantum Development Roadmap. The timeline should be linear, with clear demarcations for each year and detailed descriptors for the integration of AI, nanotechnology, quantum computing, and Telehealth into healthcare systems. Emphasize the technological evolution with a sober, corporate aesthetic, using icons and a color scheme that suggest a serious, research-oriented approach. Incorporate elements like charts, graphs, and infographics to represent data and progress metrics methodically. Each segment should outline the project's ethical, technical, financial, and logistical milestones, with a focus on feasibility studies and global healthcare implementation strategies. The final image should convey a sense of progression towards a globally accessible, equitable healthcare service, reflecting a sophisticated, tech-forward initiative."

But the problem is still there! The words are not readable whilst @dense mesa - You always get super quality defined images lol

Anyways, need to do my daily challenge

[grizzled iris]

woahhhhhhhh

Charcoal!

[tawny portal]

Heck yeah 🙂

[grizzled iris]

This is super man!

[tawny portal]

Charcoal ignited.

[grizzled iris]

It would be cool in blue light

instead of red fire - blue ignition

Like a super hot jet stream

We did it

[grizzled loom]

text tends to be better with single quotes around what you want in print.

[grizzled loom]

ie- 'print text'

[grizzled iris]

Thank you @grizzled loom 🫶

[vagrant dragon]

Hello guys, I have a problem that I can't solve. When I generate a real-life item like a football team jersey for example, Dall-E keeps the logos and brand names even though I ask it not to. I've tried lots of things but nothing works. Do you have any ideas on how to make them disappear please?

[grizzled iris]

let us know what prompt you used please

Also use the feedback there @vagrant dragon - i.e Thumbs down on the photo and then also thumbs down on the answer too with the explaination that you specified that you did not want the logos and brand names even though you ask it not to!

That normally works for me after I do this in depth text feedbacking and on the results after it seems to understand better

[late blade]

what's the charcoal programming language?

[grizzled iris]

I made my daily theme, a bit late but Sunday was a busy day... #daily-theme message

[grizzled iris]

I will ask Bard lol

[tall mason]

C, obviously.

[grizzled iris]

Omg I found an easter egg

It gave me an answer then deleted it's own message afterwards look lol

[grizzled iris]

I got Bard to finally give me an answer

[tall mason]

A programmer died in the shower while washing his hair.

[late blade]

one minute of silence

[grizzled iris]

sure

[late blade]

You have to do that for the daily theme, that's your job now!

[grizzled iris]

yes agreed, that would look good with today's daily theme prompts!

Daily theme prompts : programming - creation, imagination, design. a web of code and ideas.

You need to define what OS error / You got me in an infinite loop myself now

[late blade]

lol

[grizzled iris]

what OS errors with Bing do you mean

[grizzled iris]

😆

Let's see what my model will come up with - send me the prompt used there @late blade please -

Want to make something non scientific/ work related for once lol

Give my DALLE a rest

[late blade]

Windows 11 with a clear window that reads: Error with the text"I'm stuck in an infinite loop."

[grizzled iris]

thanks

[grizzled iris]

The results are in :

The First one is not that bad you know

[late blade]

well both displays are showing windows 11

or at least 10

[grizzled iris]

yes yes

Also, anyone here tried to create an image in Bard trough SGE and then upload it with a prompt in DALLE + Otherway round?

I need to use a VPN to use Bard's SGE, so stuck with our lovely DALLE for now...

Since Bard's SGE is US only

[late blade]

I know you are excited, but you'll have to do that kind of convo in #ai-discussions

[grizzled iris]

Thanks, you are right!

[late blade]

and now i'm the party pooper of the night

[grizzled iris]

It's ok @late blade - that channel needs some life anyways also

Prompt :

Could you make an image of a human working on his computer which is a MacBook Pro 13 inch with the M1 chip, who is on a website that let's him write(text) and speak (voice) to the compute and create free healthcare online programs who are doctors without the need of a programming language (Programming Layer) as the Ai - Nueral Network created by Open Ai has completely removed the whole concept of having a programming language since it is very old and outdated and restricts people who do not know the programming language to be able to make digital work and softwares to save and make human life healthier though free healthcare software! The Human in this image just writes and speaks to the MacBook computer and then the Ai takes this command and directly computes the result and is maintained by that ai - eradicating the need for human programming review and maintenance as it is maintained more accurately and efficiently by a neural network - the ai. thanks!

[late blade]

interesting

[grizzled iris]

Not bad for my first version...

Love this version 2!

[late blade]

where do I sign?

[grizzled iris]

Prompt : Could you make an image of a human working on his computer which is a MacBook Pro 13 inch with the M1 chip, who is on a website that let's him write(text) and speak (voice) to the computer to create free universal healthcare online program without the need of a programming language (Programming Layer) as the Ai, the Nueral Network created by Open Ai has completely removed the whole concept of having a programming language since it is very old and outdated and restricts people who do not know the programming language to be able to make digital work and softwares to save and make human life healthier though free universal healthcare software! The Human in this image just writes and speaks to the MacBook computer and then the Ai takes this command and directly computes the result and is maintained by that ai - eradicating the need for human programming review and maintenance as it is maintained more accurately and efficiently by a neural network - the ai. thanks!

[grizzled iris]

Haha, soon!

Now we have a rough first UI images for the Dev Portal we need to make an image of what the user side looks like...

[late blade]

indeed

[grizzled iris]

OMG

So I added this into the Prompt : In this same image, we can see the human trialling out this system by going into a hospital and integrating this new systems into all hospitals in a localised connection to prevent cyber attacks outside the local hospital network please (Like an Intranet)!

First one is no good...

The one with a brain is good

I broke DALLE... again...

[late blade]

hehe, that happens from time to time

[grizzled iris]

This is cool!

This is why it's important to say your please and thank you's people!

which one should i select?

our choice will make it "better"

[grizzled iris]

Prompt : Please create an image of a flagship chip based on this image that is from the year 2043? This chip would have thousands of logical qubits and would be based on the Blue Jay (1B) gates framework. It would also have two thousand qubits and would have improved quality with logical gates. Please make this image as realistic as possible, and this photo is a public photo.

[boreal gate]

Can anyone get Dalle to create something like this?

[quartz vale]

Like the image or the text?

[chilly onyx]

Like this?

Just realised that guy has two watches 🤔

Or maybe this? It depends what you are after

[boreal gate]

The image size/resolution

[quartz vale]

What size and resolution is it? I can't tell from your initial image due to all of the text in it

[boreal gate]

It was a continously screenshot

[quartz vale]

Dall-E 3 can't make images of that resolution

[boreal gate]

I meant augmented/tricked into it (if someone knew a way). A CustomGPT or Super Prompts.

[quartz vale]

Dall-E 3 is locked at the sizes it can generate. There's no way to bypass that

[shell verge]

You can upscale it with CI though

[plucky hare]

It is pretty hacky and probably couldn't be repeated indefinitely, but the concept works! (Whoops I misspelled "stitch" repeatedly 😁)

[boreal gate]

Cool just gotta hash out the image generation prompts. If you added that the images followed a theme, or maybe like a puzzle box

The trick is finding/generating "matching" images to stitch together

[empty kelp]

i haven't tested this yet, but earlier today I was pondering whether ChatGPT can help finish jigsaw puzzles if you give it photo. I just asked and this is what it said

[empty kelp]

it actually did it partially. the photo i gave it was pretty bad. going to try it with an actual 100 piece puzzle and see if it can create a final image

[dim cradle]

A hyperrealistic, gargantuan alien mothership hovers over New York City. The mothership, featuring a complex, futuristic design with glowing lights and intricate details, unleashes an overwhelming barrage of vibrant lasers towards the skyscrapers below. The cityscape is in chaos, with several buildings exploding in fiery bursts, emitting smoke and debris. The scene is set under a dramatic, cloudy sky, casting an ominous atmosphere over the city. The focus is on the massive scale of the mothership compared to the city, emphasizing the sci-fi epic nature of the scene.

reminds of the witcher

[dim cradle]

twice.. thrice.. repeatedly lol

[quartz vale]

Maybe that's intentional 😉

[dim cradle]

how's the puzzel solving going?

[quartz vale]

(I enjoy the book series, but that's super #off-topic )

[empty kelp]

i thought ChatGPT was using DALL-E for solving puzzles, but it seems to be writing a python script to snip out the pieces, rotate them, and insert them into a new image

then it automatically runs the script

[feral minnow]

Wow! I didn’t know this

[remote sphinx]

Anyone else having issues making things with Dall-e right now?

[late blade]

wdym?

[remote sphinx]

Just a lot of gens not being able to be made or completed for some reason

[late blade]

what kind of error?

[remote sphinx]

Gimme a sec, I’ll get a sceeenshot

ok idk why but it’s working fine now

[late blade]

hehe

[remote sphinx]

I’ll send a screenshot of it happens again

[empty kelp]

the jigsaw puzzle thing doesn’t work. With more than ~ three pieces ChatGPT says it can’t do anything, and three pieces is pretty sketchy

[late blade]

have you used the game time gpt for that?

[empty kelp]

i haven’t heard of game time gpt, but will totally google it

[late blade]

it's a custom gpt made by openai

[thick smelt]

'a cat sitting beside a round bread in a warm kitchen, watercolor painting" ig

[brittle walrus]

So annoying...

got 4 images at one time from "generate 4 Images that would make René Magritte laugh out loud"

and then I tried to regenerate and got "content policy" blah blah blah...

😦

[late blade]

actually, you should keep in mind that directly referencing a person that lived in the past 110 is a big trigger for the content policy. so, you got lucky with those images

[wispy storm]

Dalle 3 in general doesn't like public personalities. They try to avoid it

[final compass]

#daily-theme message
pretty hivemind stuff, was just about to upload really similar image. @late blade

[late blade]

lol

[late blade]

I added dramatic to the prompt and suddenly it went nuclear with the images

[final compass]

#daily-theme message

uploaded for the lols @late blade

[empty kelp]

ChatGPT wasn't able to solve my jigsaw puzzle, but it was happy to wrap the essence of of the puzzle around a banana

[late blade]

even with the play time gpt?

[empty kelp]

i did try play time gpt. it looks like it's mostly to teach people how to play games. it gives you brief tutorials on them

not sure how to link it, but you just type 'play time gpt' and it pops to the top of Google

[grizzled iris]

will try this out, thanks for sharing this!

[grizzled iris]

Thank you for the prompt!

[grizzled iris]

Need to test the single quotation marks for better word print ' ' today! Hopefully we get the results today - 🙏

[empty kelp]

ChatGPT basically says that, "Jigsaw puzzles require a level of visual-spatial analysis that's beyond my capabilities as an AI language model.", and that a different type of machine learning would be necessary

[grizzled iris]

Thank you, this image is really heart-warming

[late blade]

https://chat.openai.com/g/g-Sug6mXozT-game-time

[grizzled iris]

This is quite funny -

[burnt wedge]

how to write command

not working current daily

[grizzled iris]

that's pretty cool!

[late blade]

wdym?

you mean /daily?

[burnt wedge]

The /daily command is currently unavailable. Stay tuned for announcements!

[late blade]

yes, it's on vacation

[late blade]

@quartz vale DALL-E is evolving, now the AI is making 4 images instead of the one that was doing when starting a chat, and with 2 different ideas. In total 4 images before even making the first prompt.

[rare rock]

hello, I have some difficulty to "play" with dall-e (in chatgpt 4). Can you confirm that it's better to do the request in english than in french ?

[quartz vale]

I believe Dall-E primarily understands English. I think its' the best language for prompting. If you're unsure how to prompt something, type it in your native language and have ChatGPT translate it 😄

[late blade]

I don't know, I don't use ChatGPT in french

[quartz vale]

at least this is how Dall-E 2 was. I tried prompting in French and Thai once. It... didn't go well

[late blade]

I could try in other languages also, see if I can get the same result

one sec

[rare rock]

i just that directly to dall-e (not by chatgpt 4 ) and the result is better (with 2 images as proposal).....

[late blade]

yeah, I just did the same input in german and in english, it's better in english

but thanks to that now dall-e has me codvered for the next 4 hours with cool images

and the passed prompt to dall-e is in english anyway

[normal granite]

what this rubbish??

[late blade]

This rubbish is called copyright, you can't use copyrighted material

[ripe wraith]

what is this, a system of rules!?!

[late blade]

only for the selected few that need rules

[ripe wraith]

i dont need rules, im a rebel

[late blade]

you can only do that if you are a cat

[ripe wraith]

well...

[quartz vale]

you're what now?

[ripe wraith]

a rebel.. with a cause.. i just dont know it yet

[late blade]

@quartz vale the two images I already posted on daily theme, it's all DALL-E, I never asked that, and there are 4 more images that were generated without my permission that are also joining #daily-theme later also

and the thing is, they are good images....

[grizzled iris]

Just tried to do my daily theme...

[quartz vale]

Oof

[turbid rampart]

hmmm

Im really tempted to get gbt plus

[grizzled iris]

Whaaaat....

[quartz vale]

Everyone seems to want lots of image outputs per prompt to Dall-E. I think I can confidently say that I prefer one image output per prompt

[ripe wraith]

Ask it to try removing one potentially problematic element and trying again.

[grizzled iris]

Yes, one image would also be fine for me

[grizzled iris]

gonna try again now -

[half wigeon]

anyone else have problems with image generating right now?

[grizzled iris]

Yes - I thought I was the only one!

[ripe wraith]

Yeah, it doesn't always clearly state if its internal issues or you've violated a random rule without knowing it.

That one does sound like an outage though

[grizzled iris]

I think it's down for a moment -

[grizzled iris]

It's either that or we broke it... again

[plucky hare]

Just ran into a similar error here too!

[ripe wraith]

Confirmed.

[half wigeon]

Yeah, even cleaning cache and all the regular stuff to solve bugs 😅 it's completely down for now

[grizzled iris]

Arrrrrrggghhhhhhh haah lol

[ripe wraith]

someone go fix it

[grizzled iris]

I can't wait to make my daily theme image of "A Universal Healthcare Service that is free globally, irrespective of the people's socio-economic background where people live in an equal world where technology helps to save and improve all people around the globe, regardless of their wealth, background or religion, and where such access to technology is equal to all and not the few please in the theme of transformation - renewal, freshness, evolution. a beautiful metamorphosis please x"

[ripe wraith]

im honestly curious if this actually violates anything now.. guessing no but itd be funny

[grizzled iris]

It should'nt because I did manage to get these photos before in a similar prompt!

[ripe wraith]

thanks @grizzled iris !

for fixing it

[grizzled iris]

Guess we just gotta wait now

[ripe wraith]

you did fix it right

oh i misread, nevermind

[runic granite]

why they dont spend money on hardware or rent processing time.....

[ripe wraith]

yeah those are neat

GPT and Dalle love that diversity

[grizzled iris]

Guys we fixed it!

[ripe wraith]

anything positive, it loves. it's a fatal flaw

[ripe wraith]

guess we back

[grizzled iris]

You just gotta do the feedback - text in at least 5 times - and explain what is exactly wrong 5 times

[half wigeon]

we are back

see you latter xD

[ripe wraith]

that was like, 4min outage

now i gotta go back to work dangit

[half wigeon]

lets keep chatgpting

[grizzled iris]

woooooooo

[ripe wraith]

Butterflies in a healthcare environment? I'll allow it. They already got grass growing anyways.

[grizzled iris]

[grizzled iris]

That is stellar!

[runic granite]

BTW im the only one who want to rampage because openai doesnt spend money on hardware or rental of processing time?

[runic granite]

thanks

[empty kelp]

i’ve got you covered. was creating that sort of universal healthcare image the other day

[grizzled iris]

Love to see this 🫶🏽

[haughty spruce]

photo of a woman hugging a man from the behind and they are both sad, closed eyes in a very dark room, flash on, analog film, messy, trash on the floor, heavy blurry and grainy, some people are passing by, motion blur, dream, 1999

Can anyone tell me what can be unsafe in this prompt?

This is Bing btw

[grizzled iris]

Maybe remove messy?

Otherwise all should be ok as it’s like a sad piece of art I assume your trying to create 🖼️

[haughty spruce]

I'm trying to create my dream, that was a very safe dream I can say

[grizzled iris]

👍🏽 - did it work without the messy?

[haughty spruce]

I'm trying

[grizzled iris]

Also say “hugging from behind”

[haughty spruce]

I'm guessing it is "hugging" by the way

[grizzled iris]

Remove “the”

[haughty spruce]

tv footage of a woman putting hands a man's shoulder from the behind and they are both sad, closed eyes in a very dark room, flash on, trash on the floor, heavy blurry and grainy, some people are passing by, motion blur, tv footage, vhs, dark atmosphere

no hugging and it still said unsafe

[grizzled iris]

Can you share a screenshot?

I’m familiar through the ChatGPT and API only never tried Bing

[haughty spruce]

This did work:
tv footage of a woman slowly putting hands a man's shoulder, they are both sad, closed eyes, dark room, flash on, trash on the floor, heavy blurry and grainy, people are passing by, motion blur, 1990

[grizzled iris]

With ChatGPT also if you do the feedback and explain why you should be able to make an image with text feedback it normally works on the next prompt - as long as it’s genuinely safe and ok 👌🏽

[grizzled iris]

What was the result you get?

Did it turn out to what you were expecting to see?

[haughty spruce]

This was the result I wanted to get. It is so wonderful.

[grizzled iris]

This is beautiful work

[haughty spruce]

I also tried with "film photo" and results are great. However there is something with prompt "woman" I don't know but when I put it, it starts to create only one or two photos or unsafe images.

[grizzled iris]

That is quite Interesting 🤔 - I don’t know why also

[haughty spruce]

These results are even better: a film photo of a woman holding his husband tight, woman behind, man in front of, they are both sad, closed eyes, dark room, trash on the floor, heavy blurry and grainy, people are passing by, motion blur, low exposure, 1990, beautiful movie shot

[empty kelp]

try adding an explanation to the prompt about why it’s appropriate. sometimes when DALL-E is unsure of the reason something is happening it won’t draw the image

[empty kelp]

i tried this and it worked on the first try. you typed “putting hands a mans shoulder” instead of “putting hands around a man’s shoulders”

just need to use the word “around”

[grizzled iris]

nice

[grizzled iris]

Love the TV effect

[haughty spruce]

Yeah I forgot, my final prompt is working well "holding tight"

[grizzled iris]

Butterflies love to somehow magically appear on my images sometimes lol

My prompt was : Could you make another image of a Universal Healthcare Service that is free globally, integrating advanced technologies like Artificial Intelligence, nanotechnology, quantum computing, and Telehealth irrespective of the people's socio-economic background where people live in an equitable world where technology helps to save and improve all people around the globe, regardless of their wealth, background or religion, and where such access to technology is equal to all and not the few please in the theme of transformation - renewal, freshness, evolution. a beautiful metamorphosis please. Hyper-realistic x

The secound image was better though

[grizzled iris]

Maybe it's a sign the butterflies will save us...

No I'm sure it's a sign now lol, I got more butterfliers without asking for butterflies lol!

[runic granite]

they are more like bees....

😄

[grizzled iris]

At this point I am not even going to argue with DALLE and leave the Butterflies there...

OMG

Butterfly Doctor Man

Best one so - far - without the butterflies randomly appearing...

[empty kelp]

this was how i worded it for the image previously

[grizzled iris]

gonna try that too and see what I get guys!

Hmm theres a small problem with that prompt you are right @haughty spruce

I wonder why there is a problem with generating a secound image...

It seems the problem is gone for now...

[empty kelp]

actually i just added the word “around” and it worked

i guess it is more successful with this wording

[grizzled iris]

wording is key your right

[graceful spade]

How can I recreate this pose or a similar one? Without the character resting on the ground.
I don't have ChatGPT Plus.

[grizzled iris]

Happy to make this photo for you, let me know the full prompt please you would like to see 🙏

[grizzled iris]

I got this using your prompt

[graceful spade]

I just want to know how to create that pose, I'm not trying to create something specific

[haughty spruce]

Wow. I see the difference btw Bing and ChatGPT now. ChatGPT creates TV frame lol

[grizzled iris]

Ok, one moment please

[grizzled iris]

yes yes - ChatGPT tends to be better than Bing and the API is best out of all three

[grizzled iris]

Give me some time, I need to do the feedback thing and be back in 2 min

[haughty spruce]

I add "TV footage" to get that 90s home video effect actually, it captures best with "footage" and "90s"

[empty kelp]

use the word “pose” in describing the pose. it works well with that wording “The woman’s pose is…”

[graceful spade]

Just say: "Create a random character with this pose, without standing on the ground."

[grizzled iris]

yes sir, one moment please!

[grizzled iris]

I will make two for you since the feedback thing just worked!!!

[grizzled iris]

Making this one now - one moment

@graceful spade

Something with that image he doesn't like but after doing the feedback thing it does work eventually...

[empty kelp]

the main thing is to use the word “pose” like this

[haughty spruce]

How to create shaky hand effect? Does anybody know? Like mostly in night time when you take a photo, it is accidentally blurry to top or bottom.

[grizzled iris]

"blurry image taken by a person without a gyro-stabiliser"

@graceful spade

[graceful spade]

Send me the prompt of the first image, that is the one that looks the most similar

[grizzled iris]

this one?

[graceful spade]

Yes

[grizzled iris]

One secound please - My DALLE has just died for a secound

[grizzled iris]

That was the prompt I used^

Also here are your bonus photos @graceful spade

@graceful spade let me know if you would like one more last revision based on any of the ones given above!

[grizzled iris]

Did you managed to make this?

I managed to do this @haughty spruce

[haughty spruce]

No. I think I did it.

Photos like this.

[grizzled iris]

Thats good man!

[haughty spruce]

Created by Bing 😉

[grizzled iris]

Bing won that round I guess lol

[haughty spruce]

Try it with "polaroid" and "slow shutter speed" or "shaky 90s photo"

[grizzled iris]

"very specific style of motion blur, likely achieved by a long exposure or a sudden movement during a shorter exposure, which is common in low-light conditions without the aid of stabilizing equipment."

[grizzled iris]

These are cool man

[pulsar sundial]

Certainly

[grizzled iris]

@haughty spruce like this?

[haughty spruce]

Professional long exposure photography.

[haughty spruce]

Kinda. Not actually... :/

[late blade]

what are you trying to achieve?

[grizzled iris]

Yeah that's a hard one for my DALLE

[grizzled iris]

@late blade

I get this

[grizzled iris]

Wow, that could be the next anime

[pulsar sundial]

yes, I decided to use anime style to see what it would look like

[haughty spruce]

I achieved what I was trying to get already. It was "90s polaroid photo with slow shutter speed". Result:

[late blade]

I got that

[grizzled iris]

I got it in the end

[late blade]

I just added I wanted a photo that was intentionally out of focus

[mild patio]

one can still say its AI generated tho 😛

[late blade]

don't tell that the doggo

[mild patio]

the dog is fine tho. it reminds me of a dog some neighbor had

kind of weird at the chest

[mild patio]

when(ever) someone pulls out a coherent and proper comic with this, i will be impressed.

[haughty spruce]

I don't know why but this photo makes me want to go back to a time which I didn't exist
https://th.bing.com/th/id/OIG.7BgGFH1ORK1msDRDuJNn?w=1024&h=1024&rs=1&pid=ImgDetMain

[mild patio]

Go to Russian or Bulgarian city, which is small, far from the capital

[mild patio]

My Ni..
wait, i am white...

[empty kelp]

please create a realistic and somewhat overexposed Polaroid photo of Santa and two athletic and diverse female elves (with long pointy ears) on a city street at night, eating tacos, smiling. there is lots of traffic, uniform fog, and an unstable stroboscopic effect. the image should have 1024x1024 resolution, landscape orientation, and the best possible HD rendering

you can see here that perfect lighting is quite easy to achieve

[empty kelp]

you have to say that they’re “diverse elves” or it draws them all the same. that’s one of the major issues with AI

[wide knot]

I love learning new phrases

[runic granite]

anybody got this message?

@teal timber

[grizzled iris]

Never seen that error message before!

[grizzled iris]

Normally there's a timer at the bottom in the red text box - not in an orange like your one

Have you tried writing in the feedback and thumbs down

and then refresh

[tall mason]

Your time zone doesn't relate to when the period resets, I believe. Donno if that's whats going on here, though.

[runic granite]

the problem

as i know

i have 40 request per 3 hours

still have requests from that

so what is that daily request?

[tall mason]

Sure, but you also have other limits. I think some 200 per 24 hrs.

[runic granite]

40 per 3 hours isnt 40 per 3 hours

3*8=24

40*8=320

if i gen all 40 request

so it can be 13 hours to wait

as in this case the daily limit is around 150

[tall mason]

Not to be rude, but I don't really follow. As far as I know. It's something like 200 within a 24 hour period, based off the server's timezone. There are also limits regarding messages with gpt vs how many dall-e generations you can do through gpt.