images-discussions

[rocky drum]

I do get that. I tend to do about 3 different posts on a daily theme. But I see a lot of other accounts seemingly post more. But as long as I know the rules, I will just spread out when I do it more. I just was wondering if everyone has the same slow mode. But thanks for clarifying.

[teal timber]

Ya everyone has the same 30 minutes. I know some post every hour or so which is fine

[open wind]

Hello, will DallE 2 remain the same and available when dalle 3 becomes widely available ?

[quartz vale]

We don't know what will happen with Dall-E 2. All announcements will be posted in #announcements

[open wind]

Thanks for the reply

[rich star]

[dusty radish]

you say a paddleboard is not in the training data? how can that be? dont they use all images on google for training?

[fair crescent]

Not everything is in the training data, or maybe the algorithms do not give enough priority or something. Whatever the cause, it is easiest to imagine there simply is no training data available. Centaurs for example. We all know what to expect. In Bing you cannot ask for a centaur though. See the first picture. Prompt was "A centaur"

A paddleboard seems closer, bit I think it gets lucky because of associations. You see it generation a paddle and a board. As the paddle is used with aquatic boards like a surf board type things, it produces that. See the second picture. Prompt was "A paddleboard", though I tried many photographs against a wall too for example with sort of the same resulh. Thus for whatever reason, DALL·E 2 in Bing doesn't "know" what a paddleboard or centaur is.

[dusky grove]

Will DALL·E 2 still be available after DALL·E 3 release?

[wispy storm]

I think yes like MJ keep old model and chatgpt old api for a while

That's my guess

[dusty radish]

Ah okay, so i might use midjourney then? I will only need to generate photograph-like images of products, such as tennis racket, paddleboard, hiking shoes, ski helmet. The reason i dont take photos myself, is because it is an automated process for a company we're trying to achieve. We don't have time for that. We'd rather buy API access to speed up the process.

[fair crescent]

I cannot speak for MJ. I do not have enough experience to say they have such training data. From how they operate I would say they cannot do so either. Maybe you can use their trial credits to try and create it to see for yourself?

I would sooner try with DALL·E 3. The model is getting an absolutely massive upgrade, wich much more training data. Centaurs were present for example. A paddleboard probably be too. DALL·E 3 isn't out yet. It is rolled out early October for ChatGPT Plus/Enterprise and later this fall for labs (website). Bing is also getting DALL·E 3 in the future in some capacity.

The failure of MJ for me is that they are style over content, while DALL·E tries to follow your prompt to the letter. That is why I think it's better suited for your needs.

For your purposes you might also want to checkout MS Cockpit in the Enterprise or the Office 365 version if you want api control. They will get DALL·E 3 integrated in the future. I do not yet have experience with either, but from the promotional materials it seems it can do all you want and more.

[vagrant zodiac]

It definitely seems to have paddleboards in the current (not dalle 3) bing image creator 🙂

And in Dall-E 3 😉 same prompt

I will say, just in a general PSA though... is that you have to follow openai guidelines when creating/using these images. So it's not allowed to pass off ai-generated images as something real. So if the purpose is for advertising products, you'd have to be clear that the images were AI-generated. Would also have to make sure you comply with local advertising standards.

[lone mantle]

If DALL E 3 gets a plugin for PS it would be a big plus as well

[vagrant zodiac]

Well, I will say that isn't going to be first party from Adobe as they do already have built-in ai generation with Firefly

[lone mantle]

Yeah

I mean there is a SD plugin for PS as well but its not as good as a native Integration of their in house model

But there are people who dont use Firefly and would eventually use DALL E 3 + Photosho9

Photoshop

[vagrant zodiac]

maybe someone here will end up building one if/when a de3 api becomes available!

[lone mantle]

Yeah

[acoustic crater]

Hey folks: Just a reminder there's no need to pay for DALL-E in Labs, since it's available for free (Bing Image Creator)

It has also been confirmed DALL-E 3 will soon be publicly available in Bing Image Creator

[wispy storm]

Last time I tried bing dalle it had limit to gen picture

[acoustic crater]

Bing image creator is completely free and unlimited. You get 100 free boosts a day - boosts speed up image generation from 20-30 seconds to 10-15 seconds. Once your boosts run out, you can still generate images without limitation for the rest of the day (until the next 100 boosts the next day), it'll just take slightly longer

And 100 boosts every day is a LOT, it means 400 images generated at high speed a day

Then you can generate thousands more if you like, would just be slower

[fair crescent]

Could you show the source? I've only found ambiguous information so far that makes it uncertain when it will come to the public. All I can say for certain is that it will be in MS Cockpit in the future, and that accompanying Bing image creator.

[fair crescent]

All the given information might be subject to change when DALL·E 3 is implemented as well.

[acoustic crater]

Can't send a Twitter link here because Twitter is not whitelisted

But it's a tweet from a few days ago

[acoustic crater]

Mikhail Parakhin, Bing CEO, said there are no plans to change this 100-boosts-a-day approach

[fair crescent]

Thanks. The last picture is the only one new to me. Helps a lot!

The rest is unfortunately ambiguous. The first only only tells about a rollout of DALL·E 3 within MS, but not clear where. The second tells that it will be in a few weeks, but is careful to mention new issues can push it back with uncertain amounts.

My experience with such rollouts is that nothing is definitive, unless unambiguously and clearly stated in official documentation.

Only the 100 credits are unambiguous, which is nice.

[acoustic crater]

People already have gotten DALL-E 3 on Bing Image Creator

So this is excellent news! @fair crescent

Ability to correctly render text is an indicator of DALL-E 3

Another example

[fair crescent]

People having this is not an indicator of a rollout either. People who 'have' DALL·E 3 on Bing often report that after a while they revert back to DALL·E 2. Some have it later again. There might be daily or rolling limits, suggesting it is a test and not a rollout.

When a true rollout happens we can expect announcements from the respective parties. OpenAI for ChatGPT and labs, MS for Bing.

I really want DALL·E 3, but we simply have to wait for it to launch

[acoustic crater]

Except in the first tweet I sent by the Bing account, it shows DALL-E 3 generating through Bing Creative chat mode, which is 100% free

I actually would like to differ and say I don't find this ambiguous. Numerous users on reddit have this already, and Mikhail seems to be pretty clear on DALL-E 3 coming to Bing Image Creator

[acoustic crater]

Losing DALL-E 3 after having it is normal. It's part of the testing phase. This was the exact same scenario with an earlier feature: Visual input. I got it and lost it too. Now it's available for everyone

The reason behind this is to select a different sample of random users each time, to avoid bias

So there is a rotation between random samples of random users to ensure wide-as-possible data collection

I get your skepticism, but I feel confident in saying we're all having free access to DALL-E 3 in October 🙂

Microsoft has deeply ingrained Image Creator in their products and won't let users access to an inferior, older version - just like how we're getting GPT-4 for free

[fair crescent]

I have my own timeline in my head, but we're not supposed to speculate. It can lead people to take it as truth.

[torpid sigil]

Day 4 of waiting every hour to see if i get dall e 3 access

Found this on reddit and i got more excited. Look at this barely looks like an aj did thus

I've also created 5 new accounts on bing just to see if one of them will get access

[torpid sigil]

Yeesh

[karmic cairn]

Is there a guide resource for dalle e?

[fair crescent]

#resources might get you started 🙂

[karmic cairn]

Thank you!

[karmic cairn]

I am trying to figure out how to use the design feature. I received the 4 free credits and apparently used them without doing anything 🤦‍♀️

[fair crescent]

Yw! Any other questions you might still have we can answer in the appropriate channels of #prompt-engineering #❌┃prompt-help #❌┃editor-help or here if it doesn't fit anything else

[fair crescent]

Are you using the labs website or the Bing image creator?

[karmic cairn]

https://labs.openai.com/

[fair crescent]

Try Bing. Until DALL·E 3 gets out many people will prefer the Bing image creator. Subjectively better in most things, still powered by DALL·E 2 and only requires an account to use.
https://www.bing.com/images/create

[karmic cairn]

I will try thx! I want to upload and image and have it provide alternate designs

on a house specifically

[fair crescent]

If DALL·E 3 comes to inpainting/outpainting in the labs website you should definitely come back to labs though.

[fair crescent]

Ah that it isn't designed to do.

[karmic cairn]

okay, sigh haha

[fair crescent]

It is solely a generative AI. Not transformative. You can do something a bit like you want in Bing and ChatGPT soon, but it would use the house to make a description, and from that description draw the house.

As a metaphor, think of someone looking at your picture, making an accurate description and then giving it to 4 random artists who only get the description each time you generate. You'll get similar houses, but not the house you want with your adaptations.

[karmic cairn]

Thank you!

[open wind]

in really hope so, i am looking forward to dall e 3 but hope that dalle 2 will remain in its currrent state and available to use

[dusky grove]

It will, in fact, still be avaliable

[clever jungle]

It will come to labs.

[red arrow]

Is there any other way to use dall-e other than buying credits, I want to make a ai poster for school but dont want to spend money

[fair crescent]

That doesn't mean it'll include inpainting/outpainting. I'll hope for it, but believe it when they confirm/I see it in Labs

[clever jungle]

I doubt it won't include those.
Such a great feature won't be discarded.

Dalle 2 exp. Missed those but It was probably due to being beta

[sinful heron]

With the ChatGPT plus for 20 Bucks, will there dall-e 3 be included or do we need credits plus a subscribtion?

[vapid elk]

plus subscribers will have access to it on chatgpt, but oai didn1t stated anything aboug costs

my educated guess is that it will either be free or link to your credits on the labs, but that is just my guess

[open wind]

wow , ok thank you so much for this info, ive been mastering my craft for a year now on dalle 2 , cant wait for 3 but i also love my current outputs just where they are now

[velvet fiber]

@quartz vale how do you have access to dalle 3? your images in daily theme are awesome.

[quartz vale]

All of my images made with Dall-E 3 have the text "Made with Dall-E 3" in the message 🙂

[velvet fiber]

i know, I mean how did you gain access to dalle 3?

Oh I see you are staff. apologies

[tepid adder]

patience... patience...

[quartz vale]

was this image made with Dall-E ?

[subtle pendant]

Seems a little too high quality for DALL-E

[thick smelt]

pokemon in dall e looks kinda good

[boreal coral]

anyone here

[thick smelt]

i'm here!

[willow crag]

that's not true anymore, is it? I haven't used it for a few months and I'm still out of credits.

[fair crescent]

It is still true, but I do see people who haven't used it in a while complaining about it. I think long inactivity removes this, treating your account as 'new' when you login again. The account might be suspended, archived or similar with too long inactivity.

I still have the free credits every month, though I use bought credits as well.

[proud saddle]

anyone know what I'd have to prompt to get an image similar to this

[vagrant zodiac]

I'd recommend asking in #❌┃prompt-help !

[proud saddle]

thank you, my bad lmao

[vagrant zodiac]

no worries. Lots of interest coming back to dalle so wanna make sure people know where to go 🙂

[acoustic crater]

Just use Bing Image Creator for DALL-E 2. Completely free

[acoustic crater]

Upload it to Bing AI using the Visual Input feature (camera icon button) and ask it "Please try to recreate this image"

It will construct the prompt

[torpid sigil]

Finally got dall-e 3. the wait is so worth it the image quality is insane on par with midjourney

[torpid sigil]

a painting of a ship docked in a harbor, a digital rendering, cg society contest winner, victorian london, boston, ships in the harbor, colonial exploration, amazing wallpaper

[tulip patio]

@torpid sigil When you got Dalle3 access how did you find out?

[torpid sigil]

Because of this

If u have readable text and very complex prompt with almost no deformation u have dall-e 3

[tulip patio]

Where did you make the image?

[torpid sigil]

@tulip patio also this "A bear with Motorcycle and clown shoes fighting a tiger wearing a leader jacker in a unicycle, the bear stole the lion bike, digital art

[torpid sigil]

Bing.com

[quartz vale]

it's important to note that not all Microsoft accounts seem to have access to the new model yet

[tulip patio]

Yeah, I'm pretty sure this isn't Dalle 3.

[spare tangle]

It would be great if the upcoming image upload would be compatible with Dalle 3 (As starting point images with iterative prompt changes). Is that unlikely to be a feature?

[quartz vale]

I feel like once Dall-E 3 is in ChatGPT, you should be able to do that because of the context tokens
I'm not sure how it'll be implemented, so we'll have to wait and see!

The video with the announcement showed something similar with Larry the super-duper sunflower hedgehog: https://openai.com/dall-e-3

[acoustic crater]

I got DALL-E 3 in Bing!!!!!!!!!!!!!!!

So excited!!

[obsidian rose]

time to check out my Bing account

still on the old version for me

[lone mantle]

I dont even have a Bing account

[subtle pendant]

You don't have a Microsoft account?

[glad canopy]

Hi im a old user of dalle2 i see it can write now how its just luck ?

[fair crescent]

Accurate text is a feature in DALL·E 3. Though it isn't launched yet, some people have access. The most accurate release date is the OpenAI version in ChatGPT, saying "early October" in the announcement. It comes to labs "later this fall"
https://openai.com/dall-e-3
For the Microsoft side we are much more uncertain. They are currently releasing a slew of copilot iterations to Windows 11, MS office 365 and Enterprise editions that should have access "soon" to DALL·E 3 in Bing image creator. An unofficial X post tells it will come in "a few weeks". Best to wait until official accurate announcements, which should come before the actual rollout of DALL·E 3.

[glad canopy]

Damn lets go incredible

[fair crescent]

Soonᵀᴹ

I'm excited

[glad canopy]

You hyped me

[agile patrol]

when dall e 3 ? 😄

[fair crescent]

#images-discussions message

[lone mantle]

i might have gotten DALL-E 3...on Bing Image Creator?

the strange part is...it works on mobile version of Google Chrome but on desktop not really

[midnight mango]

Yeah I think they rolled it out to everyone on mobile

could be wrong

[rocky drum]

They rolled it out to me today on Windows on my PC. It shows up when you go to the Bing search engine, on Edge, of course.

[dim cradle]

i got it on chrome first try

[dim cradle]

use bing image creator, try a text prompt, if it doesnt work, try the same thing on mobile or other browsers

[hallow rock]

idk where to put this but i'm having an issue when i try to generate things on my computer (my phone works fine)... does anyone know why this might be happening?

[sonic rampart]

How does one use dalle 3 is that via website or something else

[ionic ore]

So I've been lucky enough to get access to dalle 3 and have been playing with it for a while- it's AWESOME

I downloaded Edge onto my iPad and used Bing- it didn't work on my computer using Chrome lol

I know its rotating- so does that mean if I close out of this session I won't have access anymore?

[hoary night]

how do I know if I got dalle3?

[subtle pendant]

Prompt an image to include text and if you can actually read it then it's DALL-E 3

[little walrus]

I think I gained access on mobile too (Bing app on iPhone). Note though that there are some limitations: you get a certain number of "boosts" for faster processing, which replenishes weekly. Not sure how many, I had 83 shortly after testing a few prompts and encountering a note about them. Without boosts, one request (four images) takes three hours (!!!)

though this was based on a quick search, that limit may be outdated

[fathom kiln]

Hi can soemone guide me on how to create an AI image here/

[fair crescent]

To generate an image with OpenAI's DALL·E 2 AI model, please proceed to this link.
Alternatively you can go to https://www.bing.com/images/create/
This one is a free version of DALL·E and often regarded as better.

You then just describe in natural language what you want. Fir example: Salar de uyuni. A stunning picture of Salar de Uyuni, a desert that afyer rain becomes the worlds largest mirror. A couple is looking at the sky, where a few fluffy white clouds.

[pliant frigate]

Warrior angel with wings and spear full body falling stabbing spear engraving style.

[fair crescent]

#images-discussions message

[split gate]

May depend on location but I’ve never noticed it slow down when running out of credits, says it takes 15 seconds but really it’s 5-10. Don’t worry about using them up :)

[torpid sigil]

is it just me or anime illustration with woman is unusable now with dall-e 3

typing anime illustration would give every woman big... personalities and bing detect it as sexual and block it. they need to tone down the personalities or just fixed their sexual detection its annoying since dall-e 3 is great at making anime art style more than ever

u see this? this isnt even anime illustration its just digital illustration but once i added anime illustration i would surely get banned here in discord lol

[wispy storm]

I can't try dalle 3. But you can try giving a description like long hair, x color of hair, wizard robe and end with anime art style ?

That's what I did for the beta

[twin crypt]

There is no option. To create multiple custom instructions

[foggy dune]

So many thanks for hinting about Bing mobile, I just was able to test it and am really mind blown of the quality especially for Ui design inspiration 😀 At least some relief even for eu

[clever jungle]

Did you get it through chatgpt or bing?

[rigid snow]

I have 2 questions. How can I generate variants of a image but with a prompt to for example change the pose of a character and can I somehow extend a image if I generated one abs I want to generate the rest if a character if it got cut off by the edge of the image.
I use the API

[sleek owl]

hello i'm new in the server

[mortal summit]

Welcome 🙂

[torpid sigil]

Its on dall-e 2 website inpainting

[rigid snow]

As stated I use the api but do you mean with inoainting the create_edit endpoint and I need to create a mask around the object I want to repose or a mask around the images bounds to extend

[torpid sigil]

I dont think u can edit the pose at best u can move the arms around with inpainting and u can extend your images but like i said its on Dall-e 2 website but some API websites that has dall-e 2 can do inpainting in they're website like OpenArt or PlaygroundAI but idk how

[rigid snow]

I know that I can't really control the exact pose of a subject but with some descriptions you have some control over the pose. For exact posing I need other ai models.

[torpid sigil]

My english understanding is bad but do u mean that u want an ai to ask the exact pose u want? Like the subject is jumping jacks?

[rigid snow]

That's a more extreme example a more realistic example is like that the subjects looks at the camera or other object in the scene described by the prompt

[fair crescent]

Could you post the question in #❌┃prompt-help ? It is better to separate it from discussion 🙂

[torpid sigil]

Can u give me an example of what u want to do?

[rigid snow]

It's nor really about the prompts it's more about the api. For prompts I use gpt 3.5 turbo

[rigid snow]

I just want to ai generate characters in different poses but in a consistent manner

[fair crescent]

Extending an image is only possible with inpainting/outpainting on the labs website. Best to start with #resources to find out more about it before you try. I would also recommend to wait until DALL·E 3 fomes to the labs website and hopefully is added to inpainting/outpainting.

For the rest the closest you can get is making an accurate description. I would use the metaphor that an artist (you or the new ChatGPT feature to look at pictures) describes the picture as well as possible. The description with your adaptation like waving is then given to 4 artists who never seen the original picture, resulting in 4 similar bit different pictures.

[torpid sigil]

Owww I really dont think u can. Dall-e can't save any characters or train any characters so its always randomized unless u told dall-e the description of your character. Another alternative i recommend is train your own Image/character using sdxl u would get consistent character to do with any pose

[fair crescent]

Consistent or persistent characters aren't possible (yet).

[rigid snow]

Then I need to wait until dall-e 3

[rigid snow]

GPT 4 is kind of expensive and I don't have chat gpt plus becise is too expensive.

[torpid sigil]

Bing offers crappy sassy version of gpt-4

And they're also have dall-e 3 (in some users)

[rigid snow]

Then I can try out GPT 4 becise I never tryd it and maybe ist good enough that the model is werth ist higher api price

[torpid sigil]

GPT-4 is great but on bing... its not... microsoft lobotomized GPT-4 on bing and hallucinate alot and gives alot of wrong answers. Don't really recommend bing GPT-4 unless u want some taste of GPT-4 and as far i know Only bing offers 32K context which is pretty useless since Max conversation is cap at 30

[rigid snow]

Then I know to test it in creative writing and character imitation

[torpid sigil]

U can't roleplay in Bing GPT-4. Microsoft recently banned it

[ionic ore]

Bing!

[acoustic crater]

DALL-E 3 is incredible

[graceful crest]

Very

[quartz vale]

[tulip patio]

How can you tell it’s Dalle3? My improved Bing outputs still don’t seem to match the demos I’ve seen from users at OpenAI that have shared prompts.

[ebon tiger]

just generate a sign with text

if the text is wrong it's not dalle 3

[mortal summit]

We do not have any official statement regarding the Bing outputs, neither from Microsoft nor OpenAI. What we see is that some users have images that seem to be much better in quality than previous Bing outputs.

What kind of model is involved there, we do not know.

[clever jungle]

I'm using 2 exp. And it generates text pretty good.
That's not a good way of knowing if you got dalle 3 or not

[urban locust]

Yess! Finally got access to DALL-E 3 and the images are so good! This is amazing

[clever jungle]

Howwww

I'm dying to try it out, it looks amazing

[urban locust]

Got it on Bing Image Creator, you should try it out and see if you got it

[clever jungle]

How did you know that you got it?
Was there a notification?

[urban locust]

Not really, I just tried generating an image with text and it worked :)

[clever jungle]

That also works with dalle 2

[urban locust]

Not really, DALL-E 2 doesn't generate text nearly as well

[clever jungle]

Here you go, generated using dalle 2.

[urban locust]

This is actually DALL-E 3 mate haha

[clever jungle]

Not true.
Since the first beta of dalle 2 I have been getting pretty good text and titles in images.

It was mostly dependent on how you ask for the text and where you implement it

Also dalle 2 :)

[urban locust]

I don't think that is true tbh. I've been trying DALL-E 2 ever since it came out and it never generated text as well as DALL-E 3 does. It always came out pretty garbled

[clever jungle]

It did have some misses but for me it usually worked

[clever jungle]

And I am speaking from good results that I got, not just random images

[urban locust]

Try generating this prompt then: An illustration of a cute dog holding a teddy bear and a sign that says 'DALL-E 3 is here!'

Yea, but DALL-E 3 gets it right most of the time. That dinosaur image is too good to be generated by DALL-E 2

[clever jungle]

I also still get hands very wrong

[urban locust]

Yea definitely DALL-E 3

Congrats mate :)

[clever jungle]

How

That's not even close to dalle 3

He completely missed it on 3 of them and on the last one he missed a letter

This is very underwhelming to what was presented on twitter and on their page. I very highly doubt it is 3

[urban locust]

Yea, but DALL-E 3 doesn't get text right sometimes. For me it's still very hit or miss

But sometimes it gets it perfectly. Still very good compared to the last version tho

[fair crescent]

There are not too many people who have seen more models than me I wager. I have even bought a new 4TB SSD to store all the pictures of DALL·E. There has only been one DALL·E 2 model that has been posted here that was adequate at text. It was not public and Bing is not it.

Currently Bing is making DALL·E 3 available to some people. Why there is no communication about it I do not know. Probably they are still arranging their computing power for DALL·E 3. I meself have irregular access to a DALL·E 3 version on Bing. See the pictures below.

These were made within a minute from each other with the same prompt

[clever jungle]

Very weird

[fair crescent]

Any quality difference can be thanks to how Bing and DALL·E Generate pictures. OpenAI seems to give 'full' pictures, around 1-1,5 MB in size for 1024×1024. Bing does this in the 50-150KB range

Probably they compress the pictures in some way to save of computing power

Also note that the announcement pictures are often of a larger size because of the aspect ratios. 1024×1792 tall or 1792×1024 wide can make a big deal in how you feel a pictures quality.

[clever jungle]

What about hands

Did you see all the hands in their announcement posts?

[fair crescent]

If you truly want to experience DALL·E 2, try the labs website! If you can get accurate text of over 5 letters I'll salute you and crown you king of prompts 😄

[dim cradle]

geopolitical map of the world according to dall e 3 lmao

[clever jungle]

I did many times, I really loves dalle labs, still waiting for it to come back

[teal timber]

Dalle 2 did simple common words well. Like Love.

[fair crescent]

It is still available!

[fair crescent]

They looked good to me!

[fair crescent]

That is why I said more than 5 letters 😉

[teal timber]

It could so happy birthday

[clever jungle]

My bing doesn't even get close to these results

[fair crescent]

True, but it was more replicating the thousand pictures it saw than text. Like a stop sign. It always sees a stop sign with that text, so it creates it like that

[dim cradle]

does it get something like this?

[fair crescent]

Just wait until it gets to ChatGPT and Labs...

[teal timber]

Hands and text and just overall quality with DE3 are better

[clever jungle]

Dalle 2

Old one from labs

[fair crescent]

From Bing or labs?

Really!?.... be right back!

[clever jungle]

I gotta go

But dalle love stays forever

Cya

[fair crescent]

Image is 1080×1062. That is bigger than standard labs at 1024×1024. As the bottom is also the larger one I would expect the labs logo in the bottom right, or at least partly

[clever jungle]

Yea it's an old screenshot from it that I sent to a friend from this event.
I removed the dalle logo using inspect.

[clever jungle]

Now this is a beast photo

[teal timber]

This is size of de3 results for me

[fair crescent]

I tired. Dragon says no 😦

[clever jungle]

Whats the prompt?

[dim cradle]

gpt-4 describing itself as a supervillain

[teal timber]

Watercolor painting of a scene with a cloaked wizard, illuminated by moonlight, pointing assertively at a silhouette of a dragon against the moon, with 'Command the Dragon' inscribed below.

[urban locust]

This looks amazing

[fair crescent]

Just because he's got red eyes and a menacing glance does not mean he's a supervillain! Justice for misunderstood evil contraptions!

[clever jungle]

How did you get the image to be so clear though

It looks amazing

[teal timber]

Bing is different maybe.

Who knows what model or settings it has

[dim cradle]

I think saitama got pretty angry over genos' death but not this angry

Is it possible to make dall e generate good, imaginary national flags?

[urban locust]

this is awesome

[subtle pendant]

He did get this angry

He was so mad he did not care that his Serious Punch would have destroyed the planet

but cool

it works

[dim cradle]

Probably lost to king too many times

[subtle pendant]

True

but yeah

I have never seen an image generator actually generate Saitama like that without finetuning it first

Good job DALL-E

[dim cradle]

the accuracy of most of these borders is kinda scary, especially considering it's ai generated

[subtle pendant]

Midjourney and Stable Diffusion are gonna need to step up their game

And that's great

[lone mantle]

But they also have their own advantages over DALL-E currently

[subtle pendant]

I guess but it is something

[verbal sequoia]

<:book_icon:1046291503158399107> `` Rule 8 `` Other AI Discussions.

Discussions of non-OpenAI products / models should only be posted and discussed in the #ai-discussions channel.

[quartz vale]

could you two move this conversation to #ai-discussions ? 😄

[lone mantle]

If subbed more for the updates to ChatGPT, especially live interaction with images

Oh, sorry

[placid orbit]

bing chat has dalle 3 and is rolling out to everyone by 8pm pst

so enjoy guys

[quartz vale]

Source?

[placid orbit]

just trust me

i cant send link in here

i can dm u

also it was pst sorry

[quartz vale]

"just trust me" isn't a great source

[placid orbit]

I dmed you source

[obsidian rose]

big if true

[open wind]

does anyone know if the credit structure will change for dalle 2 once dalle 3 becomes widely available? dalle 3 is amazing but i have a world ive been creating on dalle 2 that i want to continue to produce

[ionic ore]

DALL-E 3 is great with character recognition!

[solar stump]

i am trying to get dall e to create images and scenes with a similar "vibe" does anyone know how I could get a decriptive prompt?

[subtle pendant]

That is one of the first things I noticed with DALL-E 3

I love the character recognition

[ionic ore]

Me too

[minor sand]

Did a prompt-to-prompt quick comparion between 2 and 3 of my favorites images from a year ago. And for me personally Dall e2 is still better in quality and mood of light, understanding of films stock, nuances of depth of field, faces subtleties and expressions, normal people faces variations. Mood of photography overall. Did the training of 3 was of supermodels or part of it was synthetic??? because c'mon the differences are huge. 2Exp could still retain something of 2 in the images with a bit better resolve in the images. But looking at this comparison in this specefic type of images I think better "resolution" doest not make an image better in this specific style. I loved so much some of my dalle2 images from the initial beta that some of them I've been making them better in SD and upscaling. I'll try to explore how to get "natural" faces in 3 until my credits are done xd ...

I mean, there is something going on here with the training data maybe? The "lost" of natural faces is very notorious.

[minor sand]

well.. no more credits left for the day xd ....the prompts should be as natural as possible for good photographic results. 3 its really really good, amazing high frequency detail

[acoustic crater]

@fair crescent I was right!!!

[obsidian rose]

so will there be a difference in settings? Is bing dalle as good as chatgpt dalle in resolution and quality?

[fair crescent]

There was a X post for Bing that no changes were planned for their credit system. So kind of unofficial answer. For ChatGPT or Labs no further details that I know.

[fair crescent]

Very nice

[acoustic crater]

Thanks
Very excited right now 🙂 Have you tried it on Bing Image Creator yet?

[fair crescent]

It is my sleep time so I should put my phone away 😆
Tomorrow!

[acoustic crater]

ah cya then! gn

[placid orbit]

credits dont matter, image gen on bing is infinite

[dim cradle]

it's kind of sluggish rn

which is unfortunate but understandable

nvm edge just sort of broke with the image display I think

[dim cradle]

damn

[sleek cave]

some outstanding possibilities for those of us whom like mario

[outer juniper]

why the hell is dall e paid 😭

[mortal summit]

Because servers cost money to run. 🙂

[outer juniper]

sadge

i thought it was free

:((

[mortal summit]

Bing image creator is free! You might have a look there.

[outer juniper]

can you redirect?

[mortal summit]

https://www.bing.com/create

[outer juniper]

ty

[wanton hatch]

At the end @placid orbit was right.
DALL-E 3 is now on Bing and free.

How come they already have it even before OpenAI subscribers?

[fair crescent]

Do note that sharing pictures is for other channels, like #🌄┃general-art , #landscapes or #daily-theme

[fair crescent]

It was a guess. Eventually someone needs to be proven right.

[exotic stratus]

Man this is a beast dalle 3

[vague kelp]

I love that I can have it make memes of the news headlines now

[acoustic crater]

Anyone noticed Bing Image Creator is extremely slow now?

It was 10-15 seconds yesterday for a prompt (without boosts), now it's 1-2 minutes

[plush hare]

I think that DALL-E 3 uses a technique where it generates parts of the image and assemble them together a the end, because on this image you can see that it pretty much failed

[snow thorn]

Hey guys, how long is it taking you to get a generation after finishing the credits?

[obsidian rose]

Bing is not bad but I prefer ChatGPT by a lot. Can't wait until there is DALLE inside it

[primal heart]

i think this is neat

why does this keep appearing?

[fading spoke]

can you also dm me it?

[primal heart]

so, dall-e ai keeps getting the skylanders franchise or some of the franchises wrong, not sure why (it should gotten the characters by now)

[fading spoke]

Is there a way to change aspect ratio?

[obsidian rose]

so will ChatGPT's DALLE be as restrictive as Bing? I'm finding it a bit too restrictive imo

[cedar onyx]

hey guys how can i get access to dall-e 3

[primal heart]

wait a minute it came out?

[obsidian rose]

you can use in Bing, I think ChatGPT only october

[primal heart]

so it will release on october?
(Wish we can share images so they can get the reference)

[mortal summit]

Take a look here

https://www.bing.com/create

[cedar onyx]

ty

[obsidian rose]

yeah, the optimistic in me wants to believe we get access tomorrow on chatgpt

[primal heart]

hoping we can share art or images

[obsidian rose]

I don't have the images to share right now, didn't save

[mortal summit]

Keep in mind this channel is not ment for sharing art. Head over to #🌄┃general-art for that 🙂

[primal heart]

i mean as in sharing the art to chatgpt so it can understand the characters design

[obsidian rose]

yeah we do have GPT vision

[primal heart]

as in app? Cause i don’t have the inage share tjing

and dall.e dosen’t recognize the skylanders yet

[obsidian rose]

vision is the ability of ChatGPT to understand images

we will get it soon too

[fading spoke]

Is there a way to change aspect ratio for the pictures?

[tacit mortar]

I read that for some users Bing is already on dalle3 , but for me is still on dalle2.5

Do you guys confirm the same?

[snow thorn]

I seem to have Dall-e 3 on my bing

[primal heart]

what should i do about that?

like really, i really hope there’s a way

[fading spoke]

In Bing, where can I see all of my creations? Is there a page for that?

[minor sand]

i've never seen a place where to look at all the creations made. But you can save a creation in a collection (just next to download) an then go to bing search - images - collections ...

[fair crescent]

There is no way to know if you're in DALL·E 3. The quality is simply higher at one point. It is not completely sure if currently people have access to DALL·E 3, but it seems to be the case.

Regardless of the naming, you can test if you have the new quality pictures by asking for text. Where the original heavily struggled to get a word, this one can do several accurately and a sentence with some small hiccups.

[quartz vale]

I'm not sure. I don't know the details behind Microsoft's rollout

[candid leaf]

I destroyed dall 3. Been waiting for the picture for 45 minutes 😅

[gilded osprey]

I guess this is DE3

I also got

[fair crescent]

Yes, but you already knew 😆

[gilded osprey]

Now I no longer have good excuses for not being able to create good images 😅

[obsidian rose]

so I can generate anything very well, but it seems dalle has a bias to block anime style images. Or is it just me?

[mortal summit]

It is possible that anime images lead to an increased number of NSFW outputs activiting the filters.

[obsidian rose]

yeah I think that might be it

but also lower tolerance, things that nijijourney would be fine with aren't accepted

I really hope the chatgpt release relaxes a bit on that

[fair crescent]

I have seen people asking prompts that should lead them to jail. MJ isn't the example of what should be ok imo

[obsidian rose]

I mean the same img generates with a different art style but not with anime

[mortal summit]

Keep in mind that this is cutting edge technology coming straight out of research. Making it safe is an ongoing effort and filters come down on the side of caution 🙂

[rigid bough]

the safety features exist to keep everyone safe, I doubt they will be relaxed

[obsidian rose]

I don't see how a normal anime pic anyone can find on Google not being created will harm anyone but sure

[gilded osprey]

Oo.. this is magic.

[gilded osprey]

I tried posting it and the prompt to anime-manga, but it got removed for being in a foreign language. 😅

[gilded osprey]

(Not really true)

[wanton hatch]

Yeah i've noticed too, and if you add beautiful as a keyword (e.g. beautiful female blablabla) it's even worst

[earnest flame]

Even the clock is quite good!

[obsidian rose]

If you look at Bing chat, it used to refuse even basic request e.g "Write me some code". So I'm not surprised DALLE on bing is like that

AGI won't be as useful if the word 'beautiful' is prohibited

[teal comet]

the vibe, you mean liminal spaces ig? and then you just gotta add what you wanna see

so here's my prompt: a row garages of small suburban houses next to each other fading into darkness, driveways, lawn. ground level view. the suburbs. liminal spaces. night. vivid stars. dim. nostalgic. strange. weird. greenish cyanish ambient. filters.

would be cool if someone used premium version and made full res image

[solar stump]

really cool, thank you

[wind gull]

DE3 minimalism

[verbal goblet]

unfortunatly for me my DE3 is not working

[subtle pendant]

how so

[verbal goblet]

when I create an image it forever generates

[nocturne granite]

me too.

[verbal goblet]

Is it a server side issue?

[wanton hatch]

Yep it's broken. I've finished the tokens and i'm stuck on "Creating..." on the button. Been like this for 3+ hours

[mortal summit]

It is possible that there is huge demand, be patient 🙂

[midnight mango]

is there any way to access your entire history on Bing Image Creator?

[placid orbit]

Sorry, I don't think its possible, but they arent deleted (links to old creations still work), hope in the future bing adds a better history browser

Best to save generations you like

[fallen scaffold]

Been generating three images for over four/five hours now

Seems like everyone is, i think we'll just have to wait

Anyway here's some comics I made with the wonderful DE3

[quartz vale]

I love them!

[fallen scaffold]

For comics I've found using "one panel comic" or however many panels you want gives a great style, but you have to be very specific about what characters are speaking and where, otherwise you might be getting talking objects or walls

[quartz vale]

😆

[tacit mortar]

guys just a question, I have seen a lots of users on "daily them" channel using DE3 send images with like 16:9 ar

[rocky acorn]

any way to make minecraft images that dont have shaders?

[tacit mortar]

How does it work? they have access to lab openai instead of bing?

[midnight mango]

yeah they have it ouside of Bing, probably API access

[tacit mortar]

Ahh ok that's why thanks, I think we have to wait for the same version

Hope will come soon!

[midnight mango]

It should be early october

but I'm happy we get to play around with it in Bing first, even tho it is relatively low res

Dall-E 3 is absolutely mindblowing

[subtle pendant]

The perks of having a 49% stake in OpenAI

👍

[vague kelp]

Maybe say alpha or vanilla Minecraft

[tacit mortar]

yea true

[midnight mango]

I love the absolutely massive increase in quality in the daily theme channel

[placid orbit]

I hope they release it to playground as well, 50 prompts every 3 hour will be annoying

[fair crescent]

Playground?

[placid orbit]

like for dalle 2

[fair crescent]

Doesn't ring a bell 😆

[placid orbit]

just on labs

and hope it will be cheaper

[peak trellis]

when will dall e 3 come out?

and when it does will be free for the general public?

[subtle pendant]

Technically already out through Bing Image Creator

[peak trellis]

is Bing Image Creator free for the general public?

[subtle pendant]

Making a Microsoft account is free so yes

[peak trellis]

is there a credit system just like dall e?

[quartz vale]

there are credits, and you get 100 a day for free (I think)

[dim cradle]

Yes, it's as good as infinite

It's 100 credits a week, but when you run out it's supposedly slower

couldn't feel the difference tho

[peak trellis]

so how many credits are needed to generate an image?

[lone mantle]

1

[peak trellis]

nice thank you all for the information

I'm trying it out right now but it's taking very long to generate an image, is it supposed to take over 4 minutes to create an image?

[wanton hatch]

There are problems at the moment apparently, probably because now it's open to the entire world 😅

[peak trellis]

Makes sense, is it reasonable to expect that this wait time is to be decreased in the future?

[wanton hatch]

This morning i was rendering images in a bunch of seconds (with the tokens). It should be not so slow anyway when you reach 0 tokens.

[peak trellis]

okay thank you for all the information

[fallen scaffold]

Yeah I was generating images in 20 seconds last night. They're just having issues right now.

[wise sleet]

Hello! Apologies if this topic has been discussed elsewhere, but I am wondering if anybody else tried using DALL-E to generate assets with a chroma green or alpha background? I am trying to create some assets for a livestream project and was hoping to use DALL-E to help create some images I can use for visual effects.

[vague kelp]

It’s not slow, something is wrong with the app. Even if you are out of tokens and stuck, you should still be able to generate, just go into the bing chat and ask it to generate the image for you. Just have to wait til they fix the main app interface for the generator

[graceful hill]

Does Bing use the DALL-3 API? it got stuck generating.

[vague kelp]

Go to the chat and ask it to generate for you, just a bug in the app right now

[graceful hill]

Where is the chat?

[vague kelp]

The blue bing chat button at the bottom of the app.

Turn on Gpt4 when you get in

[graceful hill]

You mean Feedback? I'm using the Web.

[vague kelp]

I guess just search for bing chat then

[floral bolt]

What are the usage rights if I generate imagery using Dalle3 via Bing? Good for commercial or?

[fallen scaffold]

but it's still generating through the same app, so it's experiencing the same issues. We're just gonna have to wait until whatever's wrong is resolved. I thought for a moment that it was, because the images I ordered eight hours ago just finished, but it's still experiencing difficulties.

[gilded osprey]

“AI-generated art cannot receive copyrights, US court says” - Reuters

[turbid sleet]

so you're allowed to sell dalle images through the form of nfts or merchandise? i read conflicting answers when i searched this up

[subtle pendant]

you have bigger problems to worry about if you think NFTs are a good thing to invest in

[turbid sleet]

fair enough

[subtle pendant]

at least you took it like a champ lmao

[gilded osprey]

You shouldn’t search for legal advice in an internet forum

[turbid sleet]

yeah shouldve known

[midnight mango]

Oof, I have to wait 5 minutes between gens now

[verbal goblet]

mine been stuck for 2 hours

[quartz vale]

RIP

[hazy frigate]

I have a question. How long does an image generation usually take for the Bing version of Dall-E? I've been waiting for almost an hour. Also my prompt is pretty detailed. Should I make less detailed prompts?

[quartz vale]

You might want to refresh your browser 👀

[hazy frigate]

I just did but it says that it is still creating.

Also this is my first time using Dall-E so I apologize if I'm using it wrong.

My prompt is basically an eldritch abomination Incredible Hulk attacking Superman on the streets of New York City but a little bit more detailed. Is that prompt allowed?

[fading spoke]

how do boosts regenerate? I dont really understand it. earlier i used up like 40, and then I refreshed my browser and was back to 100. then I used 3, closed the tab an 10 mins later I was back to 100. Now I have used around 40 boosts like 2h ago, and still am at 66, the number I stopped at

[candid leaf]

I am trying to test the limits of D3. At the bottom, a printer, upon a coffee machine. on top of that a cat. On this cat, a monitor and on the monitor, a cup of ice cream

It's not perfect but it's pretty impressive

[dapper plank]

lol

[placid orbit]

bro hit the limit 😭

they give us like $10 worth of credits a day

[runic granite]

hello everybody. Maybe asked several times and a dumb question but will GPT 4 gen images too and edit them on the fly?

still gen...

[tepid adder]

As long as I don't have to deal with Bing's servers, I'll be happy

[placid orbit]

yeah but im not sure what you mean by edit

theyll let us generate in different aspects and stuff

but its still text to img

[subtle pendant]

Not exactly sure what you mean by edit but ChatGPT could iterate upon the image generated

OpenAI used the sunflower hedgehog for example

[runic granite]

@subtle pendant @placid orbit First of all thank you for the answer. I mean.... forexample i want to gen a drow but with "vantablack" (as dark black as possible) but GPT gen a grey skin drow. I like the composition. Can I write that same image but with darker skin? or will it gen a totally different (not the same with darker skin) image? So... technically edit it...

[oak fossil]

I'm so glad I'm not the only one stuck on this issue. The times has been fluxuating all day, ranging from 30 seconds to 5 hours.... lol

[midnight mango]

I am definitely a Dall-E 3 addict now

[quartz vale]

aren't we all 🥲

[midnight mango]

Just to answer my own question here for anyone who is curious: I found out you can access the whole history through your browsing history

[raw jolt]

guys when is dall-e 3 going to get released for plus users

[stark stream]

I can’t believe I’m using the bing app for it now

[dim cradle]

yeah it has extremely high traffic

probably why

[fair crescent]

It is even 'comprehensive', as you can search on the prompt via the history. It isn't as good as visually searching, but it's something.

[oak fossil]

it's good they're acknowledging it

[cloud narwhal]

What prompt did you use to get dalle 3?

[tardy kernel]

been getting these for weeks now, can't generate anything!? anyone else get these? "Error: Additional properties are not allowed ('sampling_params_config' was unexpected)"

[fair crescent]

These can be caused by anything that modifies your internet experience. Think private or cookieless browsers, Ad blockers, VPN, etc. I would check what has changed after the last time you could use it.

[raw jolt]

Is bing really using dall-e 3? It doesn't seem as impressive as MJ

[vagrant zodiac]

Some users seem to have access to Dall-E 3 via bing, but not sure everyone does yet.

[obsidian rose]

We are finally in October, I can't wait to get DALLE on ChatGPT

has anyone gotten it yet?

[wanton hatch]

At the moment yes, Bing is using DALL-E 3 model (should be rolled out to majority of users)

PRO:
now it's better then MJ, as it can do texts, bows, crossbows, better pixel art and it's more coherent with the prompt.

CONS:
Bing model is restricted a ton by Microsoft obviously, e.g. you can't style by artist name (or they removed references from the model, we don't know) and NSFW content filters are easily triggered, e.g. by simply using beautiful <subject> keywords or by using anime style

[obsidian rose]

yeah, but tbf it's completely free so I'm still very satisfied with Bing despite the heavy filters

[vagrant zodiac]

The release date hasn't been announced yet. The blog post stated 'in october'.

[vagrant zodiac]

As per the announcement blog, Dall-E 3 (at least on chatgpt) is trained to reject artists from the last 100 years

[prime badger]

Gimme a less-moderated version, where I can set my own custom aspect ratio with in and out painting and i'm GLADLY switching from MJ

[obsidian rose]

I think even simple requests like two girls hugging was blocked when asked in anime style, so I do hope ChatGPT has less filters tbh, otherwise I'll stick to other software

[cloud narwhal]

Question once DALL-E 3 is officially released, will we still be able to generate images with celebrities?

[prime badger]

To be honest, all I want is simply a tool that can let me tell my stories without in total freedom and with continuity.
I'll gladly pay a premium for such a service and i'll glady stich my IRL name to those images

[wanton hatch]

Understood

[lone mantle]

That might be the case with DALL-E 3

Except for the moderation part

[prime badger]

For as long as all this services are going to be so heavily moderated, I don't think they will be useful for storytelling purposes..

[cloud narwhal]

I don't think dall-e will ever have the freedom like midjourney imo

[prime badger]

There's no much freedom in Mj either.. This means that certain type of stories cannot be effectively told

[obsidian rose]

exactly, actual creators will not be able to reliably use it

[cloud narwhal]

Ig we will see when V6 comes out for midjourney

[prime badger]

Anyone who can draw it's at a great advantage and effectively don't need to bother with the moderation part.

That will mean, that sotries like "Kill la Kill" or "Berserk" won't be able to be fully generated using those services

[prime badger]

LOL yeah, my hopes are not up. xD

[lone mantle]

Well yeah, for those that dont go for the further editing and drawing/painting part anyway its a hard limitation, especially stuff that involves NSFW, violence and more

Those people are then more likely to use SD

Or even UD

[obsidian rose]

If only nsfw and violence is blocked, a lot of us will already be happy. The problem imo is that currently Bing is blocking even basic stuff

[lone mantle]

Because they might have other tools that are main tools and DALL-E is just a supplement tool

[lone mantle]

I didnt get many of those tbh but i was a bit annoyed that devil is censored. For Jensen Huang i could understand it, but why the devil

[obsidian rose]

I assume you didn't try anime style then

[lone mantle]

I didnt indeed

i actually use two AI image generators

And i see a bunch of people will do it too

[fair crescent]

I agree that art/stories often play with the line of acceptable and not. This way they can transfer powerful messages that must be told.

However, this technology is not ready for that. I have seen enough prompts in public that were clearly only aimed at their own satisfaction. The anonymity and easy availability makes it important to set up these boundaries.

Maybe one day some artists could get access to a less restrictive version on a per project basis or something, but right now it's in development. The restrictions need to stay to prevent abuse.

Can you remove suggestions to circumvent any restrictions?

[lone mantle]

I agree. But as said there are major differences between artists (especially skilled ones) and prompters

[lone mantle]

Hmm okay, but that is a bit different imo but i will do it

Done

[fair crescent]

thanks!

The B option as well if you can 🙂

[lone mantle]

What do you mean by B option? 0_0

Oh

I know

[fair crescent]

This one!

It isn't that anyone thinking about circumventing it will not think about it, it is that people who hadn't yet might get ideas.

[lone mantle]

Had to delete the whole message xD

[fair crescent]

But you can edit! The start was very valid.

[lone mantle]

I would still like to point out that people shouldnt expect the amount of control within the generative AI as they would have doing a lot of work in Photoshop itself and that gen AI is usually censored and moderated for a reason

I personally dont get bothered by moderation for obvious reasons but i get why that is

Especially with a company like OpenAI or Adobe with their Firefly

[obsidian rose]

I guess the technology is new and companies are cautious not to harm their image by relaxing restrictions, but I think if DALLE-3 inside ChatGPT can filter only the stuff that actually needs filtering and not harmless stuff, I think it can completely defeat Midjourney. A lot of people would migrate just for the increased freedom imo

like I'm sure some G rated anime will be blocked by Bing currently if we try to replicate their scenes

[mellow dome]

debatable

[obsidian rose]

Maybe I should give an example. I asked ChatGPT to make prompts for a Halloween-themed contest for a server logo featuring a cute anime girl. Nothing more, pasted the prompts on Bing and 90% of them were blocked. I think that level of filtering will make the tool not useful at all for most creators that make games or stories

[tacit mortar]

Bing is too much restrictive true

[mellow dome]

It's not about how powerful or accessible DALL E 3 will be on ChatGPT, MidJourney is more of the Discord community rather than just an image generator. You can't simply kill off the community with a powerful tool

[tacit mortar]

But who knows if dalle3 will be unlimited for ChatGPTplus subscribers or anyway u need to get tokens at a certain point

[mellow dome]

MidJourney has the community server where people posted their generated images every second publicly and you can engage these generated pictures which is cool, I don't see that community engagement else where. Here and other image generator, we're only seeing pictures

[tacit mortar]

Also I don’t understand what’s the point (until the release of de3) of buy tokens on labs if bing does it better and unlimited

[mellow dome]

dunno but can't wait to use it on GPT 4

Leonardo AI is good for creating photorealistic-like, MidJourney is good for community engagement and I don't have much experience with DALL E 2. Looking forward to DALL E 3 on GPT 4

[obsidian rose]

I think both Midjourney and DALLE-3 have products that are good enough. The one who will win is the less restricted platform, that will be the major difference. Creators want to create and don't need excessive limitations. A community will form itself if DALLE-3 gets a massive lead and defeats Midjourney

[mellow dome]

Maybe if DALL 3 can introduce image generator bot to Discord, it'll kill MidJourney

But it doesn't seem that way, it's not on MidJourney's territory

[obsidian rose]

that shouldn't be very hard. I mean if DALLE-3 really wants to, it can kill Midjourney

[mellow dome]

If they want to, they would've announced it that it'll be release on Discord

Dall E had their bot on other social media like Telegram

Wouldn't be their first time

It was before Dall E 2

[obsidian rose]

yeah, I guess it would be convenient to have it in Discord as well

[gilded osprey]

OAI has no business reason to go after MJ.

[mellow dome]

In that case, DALL E 3 won't kill MidJourney despite how powerful it will be

[tacit mortar]

It’s bad the system of subscription they have (or had until now let’s see dalle3)

Mid journey has a huge community because you subscribe and u can generate unlimited

The token system doesn’t work

[obsidian rose]

Midjourney got 250 million last year IIRC, I think OAI does have a business reason to get that market if they can

[mellow dome]

The unlimited relax works great. I generated 7000 pictures with only $60 while 7000 tokens of Dall E would've cost me around $500

[obsidian rose]

not to mention the market is growing very fast

[tacit mortar]

Exactly

That’s why mid journey nowadays is winning

[subtle ferry]

Couldn't literally anyone do that with the API, besides idk why OpenAI would want to.

[tacit mortar]

I generated the same with 30$

The 30$ plan have anyway unlimited relaxed image generations

[gilded osprey]

This is getting out of scope for this channel, but 250 million is ‘peanuts’.

[subtle ferry]

Can always move to #ai-discussions ig.

[tacit mortar]

I hope openai starts a new good system for subscriptions

[mellow dome]

It's related to Dall E discussion

[obsidian rose]

yeah we are literally brainstorming how to make DALLE-3 more successful

[tacit mortar]

Creating images with ChatGPT plus subscription is a perfect way to take advantage on mid journey

No tokens anymore please 😂

[gilded osprey]

Moved there..

[obsidian rose]

So yeah, if ChatGPT+ offers discord bot generations, surpasses Midjourney in relaxed generation pricing, somehow incentivizes a Midjourney-like community to form, makes it less restricted than Midjourney. Not only will most Midjourney users migrate to ChatGPT+, but the market itself will expand

[tacit mortar]

Yes

[wooden turtle]

Hi

[tacit mortar]

It’s ok even without the discord bot, creating on ChatGPT web is fine, but at least unlimited creations for subscribers (maybe 5$ more to add this feature)

Something like that

[young umbra]

Me too

[verbal sequoia]

<:book_icon:1046291503158399107> `` Rule 7 `` No self-promotion, soliciting, or advertising.

Do not post or direct message any members of this server to promote non-OpenAI services, products, or projects.

Exemption: API Channels

[acoustic crater]

Bing Image Creator is slow now

[exotic bane]

When can we access dalle

[waxen geode]

Use Bing image creator, it uses DALLE 3

But it’s incredibly slow

[exotic bane]

What’s the thing shown in the video

[subtle pendant]

Suffering from success

[waxen geode]

I think that is the ChatGPT integration of DALLE 3

[exotic bane]

When is that coming out

[waxen geode]

Should be rolling out in October to all plus users

[subtle pendant]

Did you read the announcement

[exotic bane]

Link plz

[subtle pendant]

You heard about DALL-E 3 but not the announcement?

[waxen geode]

https://openai.com/dall-e-3

[fair crescent]

At best generation with you. Asking to have other people generate for you is not allowed, and I'm in favour of you learning.

What do you want to make?

[raw jolt]

I still don't have dall-e 3 on my bing

[clever jungle]

is there any way to cancel bing generating an image?
It is generating this one for over 15 minutes now (With boost)

big oof

[dapper plank]

Damn, they weren't kidding last night...

[placid orbit]

wehy is that not allowed? that is a ridiculous rule

[placid orbit]

everyone does

[jolly parcel]

is there a cap in image creations using bing?

[fair crescent]

With a level playing field it seems ridiculous, but earlier it wasn't.

[fading spoke]

How long does it generally take to generate an image without a boost? i have waited around 1 1/2 hours now

[quartz vale]

the best way to learn how to make images you enjoy is to try and find your "style" 😄

[placid orbit]

even with boost its taking >5 minutes! I think the servers are super overloaded

[fading spoke]

is there a way to cancel the generation process?

[placid orbit]

no, I dont think so

if you have boosts I think you can overright it

[lime pumice]

25 mins lol

[acoustic crater]

You do. Everyone does

[night oak]

Wait, so it is realeased already but for now need to use it through bing?

[lime pumice]

Uh

45 mins

The heck

Why dalle in openai labs looks so bruh

[nimble rose]

I have a question, just ask Bing to generate the image and it will do it with dall-e 3? Or do I need some configuration?

[earnest flame]

Bing should currently use dall-e 3 for image generation. Per announcement on twitter.

https://www.bing.com/create

[cloud narwhal]

You sure? Because my bing doesn't have dall-e 3 yet

[earnest flame]

Allow me to modify that statement: there was an announcement that they are. Many users are able to. Unfortunately I am not sure what to do or who to contact if you are unable to.

[lime pumice]

100 minutes of waiting for image

[placid orbit]

prove it?

great

[somber zenith]

I get the message "10 min waiting time", but I have been waiting for over an hour.

[waxen geode]

Ya my generation took like 10 hours 😭😭

[subtle pendant]

Mikhail profile picture when

[earnest flame]

If you generate them a nice one maybe they'll add more gpus

[waxen geode]

Just reset your cookies if you don’t have it. This is what got lots of people access (no promises though)

[ionic ore]

2 hour wait 😦
My boosts refreshed but my other request is still going

[warm stirrup]

No promptober this year?

[peak trellis]

Will it not take this long to generate an image?

[quartz vale]

nothing about an event has been announced

[fading spoke]

alright they are STILL NOT DONE, i have logged out of my account and logged back in... nothin

[acoustic crater]

Thousands of new GPUs will be added to make DALL'E 3 faster

THOUSANDS

I love Microsoft

[placid orbit]

only thousands?

well im glad, hope it gets better soon

[midnight mango]

keep in mind one H100 is a beast and costs about $30,000

thousands is bonkers

[placid orbit]

its only 30% faster than a 4090

it just has crazy memory bandwidth

[quartz vale]

and more VRAM

[quartz vale]

cloud computing goes brrrrr

[acoustic crater]

Omg

I freakin love microsoft

[placid orbit]

ok but the censoring is

like I get being microsoft and all you dont want any controversy but if youre really pushing AI to be the future of the internet and creation you shouldnt really be censoring it

like its really not gonna hurt their brand, bing sucks anyway

[unborn remnant]

How do I know if im using DallE3?

[waxen geode]

Do a prompt with text in it (a sign saying “Hello World”

[unborn remnant]

Whatchu think

[waxen geode]

Definitely DALLE 2

Do you have Bing image creator?

[unborn remnant]

Hah damn, i pressed Sign up for DallE3 and it just brings me to the dalle2 generator

[unborn remnant]

I have an account done use it tho. Is dalle3 part of bing?

[waxen geode]

Yes it is

Try there to see if you have access

[waxen geode]

I get wanting more freedom in the AI, but this is new technology, and incredibly dangerous. I love creative freedom, but with it comes things that you don’t want in peoples hands, so there is a balance, and more often than not, it’s better to er on the side of safety than on possibility. Seeing history and all major life changing technologies, I hope we learned our lesson as a species.

[placid orbit]

there will be open source models with no censoring on the level of dalle 3 soon, its just delaying the inevitable

[velvet prawn]

It's the start, until the model is very stable you need to add some restriction. To avoid that the model give wrong information. After you can fine-tune with you own data for plus member.

For chatgpt, so I think we will have the possibility to fine tune for dall

when the two will fusion

[waxen geode]

That’s true. I wish there was a “universal law”, but I’ve even made my own models on my MacBook and let them train overnight. This stuff isn’t hard to do.

All issues we will have to learn to solve.

[placid orbit]

whats exactly the issue though

[waxen geode]

Research ethical ai issues

[fair crescent]

That is like saying theft and worse will always happen, so why bother with police. A fallacy.

Doubling down on restrictions makes it harder to abuse. Harder to abuse means more difficult for people who want to use it badly, and easier tracking of them as well.

[waxen geode]

It’s nothing G rated at all

[placid orbit]

i think microsoft is less concerned about bad actors using it for evil and more just blocking nsfw content

[waxen geode]

Very true, good point.

[waxen geode]

Im not so sure about that

[placid orbit]

i just dont like them being the authority on censorship just because they have the best model

[waxen geode]

I do agree, the censorship could be handled better.

[hexed umbra]

By then a DALLE4 model will be out, or something else that can generate good images another way

And none of this “censorship” has affected me at all, so I’m fine with it lol

[hexed umbra]

In what ways?

[hexed umbra]

Honestly I’m completely fine with them blocking nsfw content.

[placid orbit]

to each their own

[hexed umbra]

True but they will never satisfy everyone

[placid orbit]

people have rly strong opinions about AI. i just hope it doesnt get completely shut down

[hexed umbra]

And these restrictions will help mitigate that possibility

[devout sun]

Has anyone made a cool robot design with the Dalle stuff yet? I wanna get inspired

Like so I can do something with it that goes from the AI to making it into real art stuff. Kinda wanna get some more inspiration for this enemy robot race I wanna make for my Shellshocked story thing

[earnest flame]

Give it a shot! Just ask for a robot!

[waxen geode]

I work at a church, and for a while, it censored religious things or topics, like chatgpt refusing to talk, or the image generators censoring things like a cross or symbol or figure. ChatGPT is better now, I think they have fixed most of it.

[waxen geode]

I am as well, I don’t think AI should be used there

[unborn remnant]

Does this look like dalle3?

[waxen geode]

That looks closer, but the text should be accurate

[earnest flame]

I don't know of another model that can spell so well

[boreal tartan]

Probably Dalle 3. It sometimes makes mistakes

[subtle ferry]

DeepFloyd is also pretty good.

[boreal tartan]

These were Dall e 3, according to copilot anyway

[devout sun]

This is scary at how good these can be

I unironicly love the Roman armor esthetic

[cloud narwhal]

Is this Dall-E 3??

[devout sun]

I believe so

[vast yarrow]

What is the commercial use policy for images generated with dalle3?

[devout sun]

Don't even try it. You can't copyright AI generated stuff even if you wanted to

Basically you can't

[cloud narwhal]

What will be the difference between Dall-E 3 on ChatGPT and Dall-E 3 on Bing?

[devout sun]

I don't support it being used anywhere other than for maybe a joke or two, and for maybe getting some ideas

[vast yarrow]

This is the commercial use policy for Dalle2, just wondering if it applies to Dalle3 as well

Subject to the Content Policy and Terms, you own the images you create with DALL·E, including the right to reprint, sell, and merchandise – regardless of whether an image was generated through a free or paid credit.

[fading spoke]

Do images work for you guys again? my single image has been generating for almost 7h

[vast yarrow]

Working here, had delays earlier today

[fading spoke]

its been 7h now, and I can not even cancel the process

I am getting very annoyed

[waxen portal]

I am still waiting for over 20 hours, still not working for me so it will be tomorrow 😄

[cloud narwhal]

Is this Dall-E 3 or Dall-E 2?

[fading spoke]

wow 20h is crazy

looks Dall-E 3 to me. Text is a good indicator

[waxen portal]

It happens, I understand their case so some patience it's fine 🙂

[devout sun]

What the hell

[cloud narwhal]

What? LMAOOO

[devout sun]

Is that actually AI made or not?

[cloud narwhal]

Yeah lol.

Nahh this is wild. I'm definitely switching over to dall-e 3 when it officially comes out 😭😭

[waxen geode]

How are you getting these so fast 😭

Mine take upwards of a day to generate

[quartz vale]

would you please keep images like this in #scary-and-spooky as some users may find them disturbing 🙂

[cloud narwhal]

My bad

[lone mantle]

From what? 😄

[subtle pendant]

Midjourney or Stable Diffusion I assume

[lone mantle]

I had a lot of fun with a combo of D3 and Firefly but now its impossible to generate xD

[smoky barn]

is dall-e available within chatgpt already?

[lone mantle]

For some

[cloud narwhal]

Midjourney. Dall-E 3 so far is WAYYY better imo. The only thing dall-e 3 doesn't have is all the customization that MJ has.

[lone mantle]

True

Thats why i use it in combo with Firefly

Someone might combo it with SD

[smoky barn]

So there is no way to know if I will get it prior to subscribing to plus? Thank you for your reply!

[lone mantle]

Nope since apparently its basically rolled out over the coming weeks

[cloud narwhal]

How would dall-e 3 work with ChatGPT tho?

[smoky barn]

Thank you. Do you know if there is still a message limit?

[placid orbit]

gpt 4 makes the prompt for you

[placid orbit]

it will be on gpt 4 so assuming it uses the same 50 msg per hour

[lone mantle]

ChatGPT makes the image for you as the user above mentioned

It can even analyze images you generate

Give tips etc

[lone mantle]

As of now 50 msg/3 hours

[cloud narwhal]

Will it be more censored on ChatGPT than on bing tho? Because that's why imo MJ was better than Dall-e 2 with less censorship

[placid orbit]

probably yes but we dont know

bing currently lets you write in the style of real life artists but gpt 4 wont

[smoky barn]

Are there any advantages of using dall-e through credits rather than through ChatGPT?

[cloud narwhal]

I'm kinda curious if they will let us continue using celebrities or take that away as well

[placid orbit]

idk, they already block you from making anything political, so maybe they'll keep it in

I really dont want them to do that though, because they will probably remove ability to use copyrighted characters too

I love making manga panels for example they look really good

[smoky barn]

Also, is there any advantage to using dall-e through credits rather than through Bing? Thanks to whomever may know.

[cloud narwhal]

If they allow us to continue using celebrities and copyrighted characters then that ChatGPT plus is looking not too bad imo

Cheaper than midjourney too

[lone mantle]

Even if, i see GPT plus as kinda worth it if those new features work as expected

The only question for me is if Google will have a proper answer with their Gemini or not

[cloud narwhal]

Can't you search up real-time stuff now as well? Because the free version is still only up to 2021

[lone mantle]

In 2 weeks we might get some other exciting news regarding generative AI at the MAX event from Adobe

[lone mantle]

Yes, with integrated Bing

[placid orbit]

also gemini is rumored to be announced wednesday

in pixel 8 event

[lone mantle]

I dont know, i doubt that date tbh

Could be wrong ofc

[placid orbit]

well its weird openai been announcing dalle 3 and voice chat and vision

[subtle pendant]

Yeah

there was also that article on the Information that said OpenAI was rushing to get things out before Gemini's release

[fervent mango]

When do you think or there is a rumor of when it’s going to be released?

[winged tiger]

is bing ai image generation using dalle3 already?

[cloud narwhal]

What's Gemini???

[subtle pendant]

Google's response to GPT-4

[cloud narwhal]

Wasn't that Google Bard tho?

[subtle pendant]

Bard is a chatbot

It is powered by PaLM 2

[cloud narwhal]

So Gemini is pretty much gonna be Googles version of ChatGPT without the Ai generated images?

[subtle pendant]

Oh, Gemini will have AI image generation capabilities

I suspect that is the reason why DALL-E 3 came out recently

[cloud narwhal]

Google has their own AI Generation?

[subtle pendant]

They are working on it

Gemini is a multimodal model

[cloud narwhal]

That sounds pretty interesting. Wonder when it comes out.

[subtle pendant]

Rumors say this month actually!

Google has already given some people early access

[cloud narwhal]

Wow! The main thing that concerns me is how much restrictions there will be for the ai image generation

[subtle pendant]

I suspect that Stable Diffusion will remain the king of uncensored image generation for quite a while

[cloud narwhal]

And it really only works the best on PC

[acoustic crater]

We can't create right now
We'll notify you as soon as you can create.

[earnest flame]

Hey everyone, I'm sure you're all excited for dall-e 3. Please use this channel for dall-e discussions only. See #ai-discussions for folks discussing all other models.

[acoustic crater]

Bing is dead

I don't have access at all

[earnest flame]

Yeah seems very buggy right now. New service roll outs often come with hiccups. I suspect it will be a week or even more until things have stabilized for them. However, that's just a guess.

[acoustic crater]

Microsoft exec confirmed it should work today

[cloud narwhal]

What does thousands of H100-equivalent mean?

[acoustic crater]

H100 is a graphics card designed specifically for AI

[cloud narwhal]

Ohhh ok

[teal pier]

interesting i never heard of these

[midnight mango]

I'll just say a single H100 costs about $30,000. That should give you some idea

[acoustic crater]

I'm afraid this isn't sustainable for Microsoft

[cloud narwhal]

Danngg

[waxen geode]

How?

[jade epoch]

Microsoft definitely has the tech power to make it possible. Dall-e 3 is probably one of the best tools they have offered for free. Hope they are able to continue this cycle.

[stark stream]

use text and see if it actually shows

[lime pumice]

Still waiting for image

[jade epoch]

Dall-e 3 is down for Bing at my end. Got to use it 4 times before my luck ran out. Anyone else facing similar problems?

[midnight mango]

yeah I can't make anything either right now

[naive lake]

WHen will dall-e 3 become available in chatgpt?

[cloud narwhal]

Tbh, I'm having alot of fun with Dall-e 3 but I wish I could actually generate whenever I want and not get stuck on waiting.

[boreal coral]

where is the daily theme

[prime hinge]

Did you add the typography in dall-e or after?

[warm wyvern]

this is just straight from dalle

Shrek eldrich horror monster with dark, skinny tentacles and multiple Shrek heads and eyes everywhere. Huge number of tentacles fading into the dark background. biblically accurate angel. dark, shadows, black and red only. speech bubble "I'm sorry Donkey" speech bubble "I'm sorry Donkey". hyper detailed, fractal, psychedelic background. black and red.

that was my prompt

[prime hinge]

MJ has a difficult time with adding text, so I'm impressed!

[warm wyvern]

Yeah me too. I'm waiting for the dalle 3 api to become available

[cloud narwhal]

What would be the difference?

[balmy merlin]

Can we change dall e 3 aspect ratio on the bing image creator? or must using chatGPT for change dall e 3 aspect ratio?

[fair crescent]

At this moment it is impossible to change the aspect ratio in Bing. ChatGPT is teasing the wide and tall aspect ratios, but as they haven't mentioned it in any official information it might still be subject to change. Wouldn't worry about that though. Soon we will have certainty!

[balmy merlin]

Great! Thanks, mate..

[midnight mango]

How soon will there be an addiction rehab program for Dall-E 3 addicts

[oak fossil]

2 days without dall-e 3, realizing that the first day and a half of use was just the beginning of my addiction

[opaque gorge]

I remember last year when I got so excited to join the testing phase of Dall-e, we had to first join onboarding video sessions before given access. Amazing how fast things move in this realm

[dim cradle]

Dall-e

is amazing w t f

[hardy galleon]

Is it because I said "girls"? :/

OpenAI must hate Urusei Yatsura bruh

For hecks sake

[wanton hatch]

OpenAI? That's Microsoft Bing

[hardy galleon]

hmm

thanks Bing

I wonder when will DALL E 3 be available for free users

[wanton hatch]

You mean the official one? Thru ChatGPT?
Because at the moment is free for everyone using Microsoft Bing

You are using it

[velvet prawn]

It say november, but there will be limitation because it will be in the pay plan on chatgpt soon.

[hardy galleon]

Yeah that's what I meant

but it's sad that it's already limited with its filter

while other players literally makes images of harry Potter using drugs and stuff

[candid leaf]

So far I haven't been able to get dalle 3 to get the words right, why did I wrong.

The promt was Create an image where a savvy-looking cat stands on its hind legs, holding up a sign that reads: "Study: AI Could Replace 300 Million Jobs"

[stark stream]

Try putting “saying the text” or “has text” then your words

For example: Create an image where a savvy-looking cat stands on its hind legs, holding up a sign that says the text "Study: AI Could Replace 300 Million Jobs"

Yeah I got some better text with it.

I wish the gave us 100 boosters a day

[fair crescent]

Have you read the policy of Bing before use? I don't know the one from Bing, but it is useful to check what they allow and block before you try.

[hardy galleon]

I asked for a simple amine character and I think I'm getting flagged for constant nsfw for some reason

[mortal summit]

From my quickly googling the character you describe is a 17 year old girl whose appearance is described by " For her everyday attire she wears a yellow tiger striped bikini with matching boots."

[fair crescent]

I think that might be the fault of the source material. In my experience, anime outside AI generators can quickly get NSFW. That can mean it gets flagged easily when asked for it.

That being said, I generated a few with females. None got blocked. Double check your wording for things that could lead to NSFW in anime.

[hardy galleon]

she also wears a school uniform as an alternative, which covers her entire body.

[waxen portal]

Try aswell to change some words around and see if it happens

[hardy galleon]

I tried. I asked for a school uniform variation. Still blocked.

[waxen portal]

What's the prompt?

[hardy galleon]

My internet is really slow, but you can find it somewhere above the conversation.

there it is ._.

one of them

tried a photorealistic one in here

[dusk hearth]

holding hands, that'll do it

[waxen portal]

Either wholesome

[hardy galleon]

Is wholesome that bad? ._.

[waxen portal]

It can mean something else so yeah

[hardy galleon]

I had no idea

English is my second language after all and this is the first time I'm hearing this ._.

[waxen portal]

So yeah try to remove these and see if it happens

[hardy galleon]

hmm

I removed both of the words and it worked

However the program doesn't know who they are 😎

[candid leaf]

I don't know why, but your generations are better as far as the text is concerned

[somber zenith]

I have used DALL-E 3 via Bing. But in the meantime I get only poor quality pictures. Also text in the image is no longer displayed correctly. It seems like it's DALL-E 2 again via Bing. Has anyone else noticed this?

[wispy storm]

Dalle 3 with chatgpt > bing

[somber zenith]

Please excuse me, but I don't understand what you want to tell me

[wispy storm]

Imo, chatgpt plus with dalle 3 is far superior.

It can correct picture after the prompt and chatgpt4 is powerful. Understand what I need

[dusk hearth]

What are the current limitations on gpt4 with plus? Any limitiations for dalle on top of the ones for text messages?

[blazing mortar]

Im also a plus user but i can use dalle 3 yet

[wispy storm]

For nor as far as I can see it is 50 prompt /3h

[blazing mortar]

How did you get access to dalle 3 in chat gpt

[vagrant zodiac]

Some people currently have early preview access.

[waxen portal]

The point that you can even generate accessibility for deaf people is amazing. You can generate people using sign language for deaf

[wanton hatch]

They actually give 100 boosters a day

[nimble rose]

One interesting thing, here's my prompt:
" a papyrus containing the architectural plans of an ancient civilization."

In one case I simply put it in Bing's dall-e and in the other I asked Bing chat to generate it with a prompt I developed.

Firstly, the version without Bing chat.

Now with Bing chat generating the images:

What is the best result in your opinion guys?

[tacit mortar]

I suppose bing chat but should be the same model it uses anyway

[dusk hearth]

Seems like bing chat is closer to prompt. At least in these particular examples.

[nimble rose]

I've done other tests and in some cases dall-e seems to do better with a more "concise" description. But in the end, I think having an LLM understanding and expanding your idea is the best way!

act like an "art helper" you will improve my idea and will expand it, add descriptions, details and improve the quality of the prompt before generating the image.
Fallow these rules:
0. Define a theme for the art in general.

1. Search and attach the image description for each possible term to generate the most detailed and accurate image.
2. Improve image quality by adding best image improve quality prompts.
3. Add an appropriate image style relative to your search and image request.
4. Always add a context, whether of the scenery or the surroundings of the image, as appropriate
5. Always find a mood for the image.
   Make a list of bullet points describing your detailed responses to each point on the list and how they were translated into the prompt and generate the image.
   My initial idea is:

The prompt I used.

[willow isle]

I'm pretty sure Bing Chat may do a little bit of prompt engineering of its own sometimes unless you specifically tell it to use your exact prompt

[icy void]

Hey guys how long do you think bing chat is gonna offer almost unrestricted access to dalle3. It won't last will it?

Is microsoft really getting enough new edge and bing search users they can potentially advertize to in the future that they are willing to pay for the GPU costs?

[obsidian rose]

also curious if there are less restrictions

[vagrant zodiac]

Please do not ask this as per rule 7.

[viscid slate]

I wish Promptober was continued this year.

[stone mist]

hmm

what time the boost are replenished?

anyone knows?

[abstract sparrow]

Site says weekly

Could anyone with access to the GPT image generation tell me if it's faster than the Bing image generation?

[tribal relic]

how do i make those images where its diffrent when you squint?

[near vector]

Search for Illusion Diffusion

[wispy storm]

Doesn't take long to gen on chatgpt 4

[dark condor]

Guys. Is it me or have bing reversed back to dall e 2

[stone mist]

the reflenishment time is different?

user by user?

[dark condor]

Man I think it’s already over. Maybe they’re phasing it now because I’m sure I’m getting some dall e 2 photos now 😦

[buoyant sand]

Does anyone else have this issue? It says I have boosts but when I try to generate smth it shows me this

[vagrant zodiac]

A lot of people are using bing rn, possible they're under heavy load so aren't running as fast as usual.

[somber zenith]

Yes, I currently have that as well. Yesterday I also had this and the generated images I got only about 10 hours later

[winged moon]

Took forever but DALL·E 3 seems to still be working for me via Image Creator.

[somber zenith]

I now get a message that the team is already working hard to fix the problem

[dark condor]

It’s breaking my heart. Why can’t it not go viral. Look at what I was producing yesterday v today

[blazing mortar]

How did ypu get access to it

[abstract sparrow]

Is there any way to know in advance if I buy GPT-4 for the first time that I'll have access to the image gen?

[wispy storm]

#announcements

[blazing mortar]

So the only thing i can do is wait?

[wispy storm]

Yep

[thick smelt]

look at this cutie 🥹

[blazing mortar]

😔

[nimble rose]

I think all the images I generated were not with dall-e 3, but still, the result is incredible.

I mean... For me, this quality is undeniably incredible.

[somber zenith]

What do you base that on?

[stark stream]

Yeah, keep generating it and it will most likely get the text right

[nimble rose]

Bing dall-e request: "a robot made of dragon bones holding a rubber sword while appearing to be disgusted at not having a real weapon." But using some prompt engineering.

I tested it with a very abstract idea involving an emotion and something very disconnected... And to me, it seems to have great quality.

[hardy galleon]

The servers are overloaded again?

zamn

[somber zenith]

yep, seems so

[stark stream]

Has anyone tried any anime images like niji journey in dall-e 3?

[somber zenith]

Check out #anime-manga 🙂

[stark stream]

Ah ok thanks

[nimble rose]

What makes me think that it's not the dall-e 3 is this detail of the letters not being legible... I saw some images posted by other users and in them the dall-e 3 seemed to know how to generate texts.

[flat dome]

Hi! need help, I try to create images being also really specific and dalle still creating things I don't want, well with results I don't want

[nimble rose]

How are you doing? Via Bing chat or directly in the image creator?

[flat dome]

directly the image creator of OpenAI

[nimble rose]

Dall-e 2?

[mortal summit]

You might post a question in #❌┃prompt-help 🙂

[flat dome]

well, how do I check that?

I'm new

[mortal summit]

Do you generate the images on the labs website?

[flat dome]

yap

[nimble rose]

Open Bing chat and paste this along with your idea: act like an "art helper" you will improve my idea and will expand it, add descriptions, details and improve the quality of the prompt before generating the image.
Fallow these rules:
0. Define a theme for the art in general.

1. Search and attach the image description for each possible term to generate the most detailed and accurate image.
2. Improve image quality by adding best image improve quality prompts.
3. Add an appropriate image style relative to your search and image request.
4. Always add a context, whether of the scenery or the surroundings of the image, as appropriate
5. Always find a mood for the image.
6. Generate the image using the latest imaging technology you have (it needs to be the most up-to-date version).
   Make a list of bullet points describing your detailed responses to each point on the list and how they were translated into the prompt and generate the image.
   My initial idea is:

[mortal summit]

Then you are using DALL-E 2.0

[flat dome]

ooh, okey, which one use Dall-E 3.0?

[stark stream]

You gotta use bing image creator

[flat dome]

got it

[flat dome]

ty

[mortal summit]

https://www.bing.com/create

[stark stream]

Man if we could use art styles by people like mid journey that would be so good.

[flat dome]

thanks

[fading spoke]

WHY does dalle take so long rn

I am so mad

[flat dome]

that seems cool

[fading spoke]

I have waited 8h yesterday and am waiting another 3 rn