#openai-chatter
1314 messages · Page 18 of 2
Anime key visual of a white-haired magical girl with bird wings, cinematic advanced lighting, 4k konachan wallpaper, sharp image, highest quality image, trending on pixiv, trending on safebooru, 1px lines, [incredibly detailed seamless artwork without flaws] [extremely consistent] professional anime fanart
I can't get it any more detailed :(
I got this out of mj
Seems to have the right number of sails for a sloop
seems AI can't comprehend sail on bottom lol
and yea thats a funky lil' sloop
I actually got a MJ invite but I didn't check it so it expired xd
To ai: my goals are beyond your compression
They are open beta now so you can definitely get another one
@violet wigeon I feel like you’d enjoy these
Looks like indie animations which makes it look so good
Those mushrooms tho they are so realistic
Dalle2 is so weird like some results look almost flawless while some look like crap when you zoom in
My brain isn't smart enough to think of a way of tricking dalle2 into generating the right image
Dalle has consistently had amazing results for anything involving mushrooms, not sure why or if they just have lots of good illustrations of them
Humans evolved to draw lots of mushrooms I guess lmao
Lots of cool things you can do with mushrooms too lol
I am having issues with this one. I just want the number 40 edited in, and apparently there is no way for that to happen, wasted 3 generations with no luck. anybody can help please ? it's a birthday card gift for somebody turning 40 :(( I had to report all of them
Dalle2 simply can't do it (:< 」∠)
You will have to find another website for it I can try and find some for you
2560 x 1440 resolution
Someone try adding this instead of 4k
This is really good 😂
You could try creating variations of this? But may not look very 3D or good lol https://image.shutterstock.com/image-illustration/digits-soap-bubbles-illustration-number-260nw-1496425547.jpg
Looking at this image it doesn’t even look like you’ve erased anything for dalle to edit into the frame
right ??? but I erased with the brush everything that was pink
So are you wanting to keep those 2 random shapes and just have the number 40 on there somewhere?
exactly ! like ideally the 4 above the one on the left, and the 0 on top the right one
it did give a ZERO at some point but that;s it
Quotes sometimes work well, for example if you erase just above the pink one and then put something like “the number “4” floating above two colorful geometric shapes” it may do something, but no promises 😅
i'll give it a go, kill another credit
Right? That’s a discouraging part with the credit system, people were happy to test and try things for others but everyone is clutching their credits now 😂
I am going to try and start a fiverr since I am now a prompt engineer
Best of luck to you 🥰
literally f this !!!! I swear I brushed the whole area above
what the actual fck :(((((
5 credits down and not a single number 4 in sight :))))
I am confused as hell
Guess it just depends on what’s in the image 😅
oh nooooo, u wasted a credit !!!!!!! this look so flat. like Is it just me or the quality and complexity of stuff dropped a lot since 2 days ago ?
Lol I’ll probably buy a few hundred credits at some point anyway, it’s not all that expensive for the entertainment it provides
But yea the bias update really has mucked a few things up, but I’m sure they will iron it out :)
yes agreed, but I'm gonna go ahead and cry in eastern european average sallary which is 400$/month
oh noooo, you wasted a credit too, y'all are gonna make cme ry
but I love you
Just send her all of these
at this point I might aswell just do that :)))))
I’m sure your friend will find it funny you got the whole Discord community riled up with 40
SWEAR TO GO :))))))) you folks slay !!!!
okay okay, let me photoshop all these stufff . you are all something else lol.
gonna come back here in 15 min max with the final thing !
Looking forward to it!
but is for prompts right?
you have 50 prompts (well 150)
one prompt consumes 1 credit, not four
Are we really only getting 15 generations per month?
That’s definitely an immense downgrade compared to before if that’s true.
was this your Discord Code verification number? =P
I didnt have to verify haha i joined a while ago
You have this style nailed down it’s great
Downloading w/o watermark
Easy and oft requested
final result
Nay
so did the quality actually go down or are we all experiencing a mass placebo effect because we're mad about the new token system
because lots of people keep telling me they think the quality has gone down some but it seems so subjective
i believe it's because of the 4 images per prompt thing
and how it seems like its because of the new tokens in our mind right after the 4 images
wait
that was worded horribly
For the past three or four days, I've been actively experimenting with various graphic design elements, and I can genuinely say that this is the most normal version of the logo I've created with Dalle so far. The request was for a modern, simple logo with the theme "sun and bird." (Of course, this cannot be the final version; the final version must be vectorized and edited.) It is not cost-effective for anyone with an interest in graphic design to purchase Dalle. You may get a fully vector logo created to your specifications on Fiver and similar sites for a much less money. It is fact. There are essentially no testers in graphic design, therefore I'm not sure what Dalle's potential is in this field. Again, I'm really disappointed with the Dalle team's decision because it essentially crushed the potential that was being exposed daily with the help of testers. I have spent all my credits at the moment, see you next month 🏹
yeah right now dalle mostly just has hobbyist applications
so the pricing scheme seems silly
im lucky in that i have money but as soon as something cheaper comes a long im jumping ship
it should have been like $20 a month for a set amount of prompts per day or something
improved ability to generate sensible text.
I wonder if adding to text to the database would constitute as a dalle 2 add-on, or rather an entirely new “Dalle 3”. This would be similar to the differences between Imagen and Parti I assume
I want dalle-3 to do much better on complex prompts, and also generate images up to 3840x2160
why that resolution
it should get things 'right' the first time more often, and not require as much iteration
make me some cool 4k wallpapers. i think higher resolution images than 1k x 1k are also needed for most commercial usage
dalle2 is 1024x1024...?
yea
there's plenty of models that can upscale from 1k x 1k to higher res, maybe they can add some of the training data for that into dalle as well
Have you seen the alphabet sets? Dalle can generate nice looking letters. It just cant order or use them in a coherent way
ikr
I was severely disappointed with my last few generations with Dall e. I do thinking something has changed, I may just wait a while and see if any additional changes get made to it. I also find the move to go down to four images per generation right before adding a paywall a really interesting decision.
dalle2 makes super cool sans serif fonts
😁
although
look at this one result i made before
That being said I did a few landscapes and they came out great. But it seems like human subject centric images are struggling.
this was with the prompt "Sans-serif" LOL
Unfortunately, what is $15 for someone else is $45 for me. Unless you're just an NFT seller, paying $45 for something that doesn't really help you with certain jobs isn't really worth it.
I don’t think people are wrong when they’ve been hypothesizing it’s due to the changes to improve diversity. I’m all for diversity, but maybe they should roll it back and wait until the results generated are more up to par with everything else.
Rant over, sorry everyone ❤️
Reminds me of microsoft word art
theyve dug themselves into a hole with this one
its gonna look bad to roll back diversity
True
It’s really interesting the difference in results from before. I asked for an image of a Druid making a concoction for a group of children nearby, and not one of the images included any semblance of a kid.
Tutorial on making a wallpaper with dalle
https://www.tiktok.com/@auwimo/video/7122894520819141930?_t=8UAwY3OPUK6&_r=1
Although it can get it right sometimes.
i just did this one:
"A cackling crow laughing at the moon under a rainbow in a top hat, oil painting, acrylic painting, academic art, volumetric lighting, watercolor"
Is the moon also a laughing crow
Does anyone know what’s happening when you give it multiple art styles?
Yep! A hybrid art style
Discord verification consumed a token 
@silver hazel I feel like the examples in there are different than what I’m talking about. Maybe not. But how does it reconcile multiple mediums? Oil and watercolor and acrylic
Is it using all of them in different areas? Is it blurring them all together somehow?
Don’t have examples of these yet. Come up with a prompt of 3 styles including “A robot reading a book” and I’ll run it for you 👍
Probably want three different mediums as different as they possible can be to see what happens.
“A robot reading a book, oil pastel, watercolor, charcoal” maybe?
I’ll add “A painting of…” in front and I’ll share the results
You’re the best @silver hazel
First one hasn’t loaded. Second one looks like it prioritized the first medium given.
Would you agree @silver hazel
Now that we have rights to our dalle2 creations this might take away at companies like Getty images
I’m perhaps not as knowledgable as you are! But it seems it prefers a crayon-ish art style
Lol I’m by no means an expert. Far from it.
stock images wise, very that.
they are big on celebrity photography, however, so that’s not dying anytime soon
there's an opportunity for openai to generate a million stock art images themselves using prompts they think would be popular and start their own stock photo site too
But I’m not seeing a lot of evidence of charcoal in the images.
https://labs.openai.com/s/uzMS2lEgfA3B9FkKPSTcYPL0 it’s evident here
Looks more like wax to me
Yeah that looks like layered pastels.
Well, we are trying to combine 3 art styles so I presume it won’t be as easy 👼
I am not judging by any means. I was simply curious to see how the software would reconcile multiple art styles.
Hope these results were of use for you 👍 Will add to DD
Prioritizing the first given seems more than reasonable.
Dalle2 made orange rhyme with banana
If any of you are low on prompts and have a specific artist you want to see experimented with DALL-E, do let me know as I’m trying to expand the DALL-E Dictionary to be helpful for all of you.
Sincerely, thanks for humoring me.
I was genuinely interested.
Hm. @silver hazel one way we could test the theory is to do one more reordering the art styles?
I’d rather experiment with specific art styles and artists that are more likely to guarantee a good result 👌
Ok.
Any suggestions welcome, I’ll do them later ⭐️❤️

Fantastic!!
The keywords “chibi” and “kawaii” typically make a fairly consistent art style that may be worth adding to the dictionary
“Chibi robot” etc
Anyone else doing movies with DALL-E / MJ? I'm working on a 70s sci-fi collab storytelling film that so far was 100% MJ, but have incorporated DALL-E for the latest "story seed": https://twitter.com/SALT_VERSE/status/1550195893042823172
Before we continue! On a thousand narrative paths...
A STORY SEED.
Is this...a faction? An🥚for the 🕵️♀️.
🧂 𝚒𝚜 𝚊 𝟽𝟶𝚜 𝚜𝚌𝚒-𝚏𝚒 𝚠𝚘𝚛𝚕𝚍 𝚙𝚛𝚘𝚍𝚞𝚌𝚎𝚍 𝚠𝚒𝚝𝚑 𝙼𝚒𝚍𝚓𝚘𝚞𝚛𝚗𝚎𝚢 & 𝙳𝙰𝙻𝙻-𝙴 𝟸. 𝙲𝚘-𝚠𝚛𝚒𝚝𝚝𝚎𝚗 𝚋𝚢 𝚈𝙾𝚄.
Mya
I'm making a few short films - even shorter now since the updates 😂
Please give me your best generation to be in the front cover of my special project
this is what internet explorer does to people
rude
Im sorry for sharing that
I don't like you
It's okay ❤️ Yes, the paid model is now live!
lol are you seriously offended? What a soft world.
Tough up, learn to take a joke.
be nice
Where was I not nice?
both of u
Internet is full of offenses, I don't like to offend people, it's vague and cheap
You never know what some people may be going through.
Yeah, your joke wasn't rude, but if you see someone took offence it costs nothing to say sorry 😦
Also. It was not a good joke, come on, since 2008 we have internet explorer being slow jokes, you can do better :)
So... are you liking the paid model?
Yes
Limitations somehow make a barrier for artistic choices, you work only with what you have, I'm using dall-e for some inspiration on architectural projects
Conceptualization more than anything else
I find your comments rude, and I wish a sorry from you too please. ^^
I'm not sure how anyone can pay given what we had before
but maybe I'm just poor lol
If attention is what you want, here’s some, but don’t pick fights with other users, thanks ✌️
I don't think they want us to use it as we did before tbh. I think they're banking on a shift in how it's used (and probably need a shift if they want to open it to the public)
It's too expensive for work environments to get in a good work flow. No way you spending 15-30 dollars a day on it.
I think they just started the price high instead of low so they can lower it without upsetting people
$30 a day in some commercial contexts is small change
$30 is what lunch costs in silicon valley
And the tool 100% has commercial value
Hey there, tretoko.
I just wanted to say thanks for being such a big help on the discord server. It's really appreciated.
However, I have to say that I was pretty disappointed when you told me to apologize to someone for making a harmless joke towards him. I don't think I was doing anything wrong and I certainly wasn't looking for attention.
I hope you can understand where I'm coming from and maybe reconsider your decision in the future. Thanks again for all your help on the server.
Idk any other program that is that insanely expensive to use
but alright
just limits it's use case a lot
I agree with tretoko your comment wasn’t rude but the follow up was needlessly brash
We should all just be nice 🙂
Remember it's not a"program", it's a shift of intense, costly compute power you're paying for (and a recoup of the training costs). So it's unlike most analogous software tools
Irrelevant people want to use it like a normal program and it's what it is to them.
Friend, I laughed at your comment. But I'm a big preacher of mental health and I'd rather be safe than sorry. That's all! Lets move on in peace ❤️ I have not made any decisions. Ty for the kind answer
also we are definitely over paying for what it costs them lol
It's probably more about what they can open up to a million people and anticipate a use rate that doesn't break the system. The price is a throttle to protect their resources
Exactly
then they shouldn't open it to more people till they can properly account for that instead of insane prices
My conspiracy is that 1 million users will be more thhan enough for an AI to analyse and come up with the sweet spot of pricing, and I think Joanne let us know that they are doing exactly that.
I believe this pricing model is just a phase, but I may be wrong ❤️
yeah it's a phase for sure
maybe there will be so much demand that they raise price higher 😛
We’ll have to see how they change it
Kittys
This is correct, this is likely the highest and most restrictive the pricing will ever be
it's best to start prices high instead of low after all. You don't anger your userbase by starting high and lowering. But if you start too low and raise the price you will piss so many people off
They're probably not in a stage to build infrastructure if they've taken a loss for so long on the generous free access
The free cooldown changing to a month just feels like way too long
Maybe if it was per week
Idk
They are doing it because people will be trying to make NFTs out of it
It's not fair to us but it's how the market goes nowadays
NFTs are so harmful
lol I don't think thats why
NFTs are all hot air at this point. They don't affect anything and I doubt that's the logic
Has anyone tried generating their own city in Dalle2?
I did a birds eye view of mine and they got some features right but you know how it goes (looks like Google earth)
Please make it 1 credit per download, not per prompt running
Thats unfortunately not how compute time works. They are charging for the cost to run the servers and calucation behind the diffusion models. Those are fees that cannot be avoided
Try wiff ur city or town
also people would just screenshot the results even if it was lower quality
heres with just "neighborhood" instead of skyscrapers being mentioned in the prompt
Add watermark like stock sites
again, charging per download doesnt solve the problem of it costing money to run each prompt
When you run a prompt it costs openai money to actually use the compute time, they are now just passing those costs onto you instead of letting you run it for free like they have been
Add a maximum per month
When I do it it doesn't look like that
They've been operating at a loss for months
They should operate on a loss currently if they are smart to grow the audience since they are front runner
Back then:
Just gotta be creative with the promp keywords :) you've been used to characters, theres tons of different useful architechture keywords too!
Wow, it was the same prompt?
No but very very similar
This. One thing I suggest you guys advocate for is 64x64 previews of your prompt. Maybe that's a compromise OAI will be willing to make and I see this feature benefiting literally everyone using DALL-E
They are THE frontrunner, there is no need to grow an audience when millions are already on the waitlist and dying to use it. They have no need to operate at a loss.
Operating at a loss is for companies who are trying to get more users, DALLE has millions of users waiting
I agree, this is definitely the best solution.
I miss the old dalle2 generations
Dalle's audience is not that large currently compared to what it could be
like not even close to the potential
Why not adding ads? To reduce the losses?
What are you not seeing now that you were seeing before? Just like with childhood, your first generations probably just felt more magical, now you are getting used to it and may just be less impressed with the results.
.. You saw it with your own eyes
I will show more examples
Nah, they definitely changed something
We have many examples now from now and back then results
256 votes and 59 comments so far on Reddit
@rocky burrow
Many people advocate for ads as the solution to many marketing problems.
This is not really how the marketing world works. OpenAI does not track much user data and know nearly nothing about their customers aside from the art they are trying to generate. They have no way to target ads to specific users, thus its difficult to find marketers who are willing to pay to advertise to a specific person.
Simply throwing ads on the platform does not get users to click on those ads, which is what those advertisers are paying for. They want to find customers for THEIR products.
ads will come eventually
Dalle2 is garbage
Well, it's the same user base for creative and tech interested people
And you could even get some Infos through the prompts
Like many things, time cures all. This technology didnt even exist a year ago and now here we are telling them how they should make it work. Let the market decide and the solutions will come, patience is the key.
That's true
Be happy we are in the select few who even get early access to use it, don't shut it out so early in its process. No product was perfect in its beta, the engineers know what quality generations look like and have every reason to continue to refine it to look better and better as time passes
They would not intentionally make a product WORSE, they are testing, that is what betas are for.
Also CPS is 3 orders of magnitude below their current pricing
So for each 2 min generation you would need to watch 2000 minutes of ads to break even
AFAIK the 64x64 thumbnail generation already represents the majority of compute energy and the upscale is relatively painless, but I may be wrong on that
DALLE2 IS RUINED
I don't think any user should be happy with the pricing model currently. But it is what it is
This is the best it can generate
Take a deep breath! This comes up often and it's always either 1) a prompt issue 2) a psychological issue or 3) a replicable bug that is quickly acted upon
It generated shit like this back then. this is gorgeous
Probably 2nd
What are y'alls thoughts on the passive-aggressive narrative going on Reddit that DALL-E is "cancelled" / "ruined" and that there are plenty of open-source competitors that will blow OAI out of the water in the following months?
Do you know of any competitors?
You are still speculation and this is what you think is happening
I generated this yesterday
Until they actually say something about it, let’s calm down
Nowadays it cannot generate anything
I don’t think it makes sense lmao. Different products, pick whichever one you want. No need to act like a little babie
Or no its because im using it
Ai gets better and better fast 2 years from now there could be competitors for sure with the same quality as dalle. But going to be awhile
Reddit's gonna reddit. Once a dominant narrative takes shape there's going to be very little in the way of nuanced discussion
Most of the tools in the space are diffusion models. Beautiful results but i view as different things. I am using many different tools right now
Keep in mind most Reddit users are only able to communicate in binary terms - text good, text bad
Also reddit complaining is probably good, the pricing is insane after all. Can't see how it negatively effects us
I'm very excited to see "Stable diffusion" mentioned many places on twitter
(I say this as a longtime Redditor 😅)
Have you all experienced the "nerf" people are speculating? That DALL-E is generating worse results?
I personally generated one of my most stunning pieces yesterday... 🤔 Hysteria + Delusion or there's a bit of truth behind this narrative?
I do a lot of 3d so you had my brain going a different direction when you said nerf people 😏 (neural radiance fields)
Naur
I think people just like to assume something’s wrong
Like an “AHA! I knew it” type of thing
The Memo: https://lifearchitect.ai/memo/
====
Demo: https://nuwa-infinity.microsoft.com/
Paper: https://arxiv.org/abs/2207.09814
Read more: https://lifearchitect.ai/
Dr Alan D. Thompson is a world expert in artificial intelligence (AI), specialising in the augmentation of human intelligence, and advancing the evolution of ‘integrated AI’...
They already beat dalle2
The only thing i have solidly seen is the diversity add-ins adding people into prompts with professions
Trying to fix the biases by messing with our prompts will in some way negatively effect some prompts and generation. I mean it's pretty much impossible for it to not happen ever. But hopefully it's very rare and not a big problem
Very impressive! Thanks for sharing, havent seen this one yet
As AI scales exponentially, don't you think DALL-E has an edge over most competitors? How hard would it be for OAI to release a sudden update that blows y'alls minds?
Depends how they want to keep updating the dalle2 product i guess
Very excited to see what Google and NVIDIA come up with, this seems like a space they will thrive in
we were just lucky to get to test the first one to market
but I was speaking more about just reaching the current point we are at now for others
Yes, I understood. My point was, by the time competitors reach DALL-E, won't DALL-E be 2x ahead?
hopefully, but not sure how much support it will get
Openai is working on a lot of stuff after all and more important stuff
The effect is predictable when you understand it
Google was definitely working on the same problem at the same time, Parti and Imagen were underway for years, OpenAI just announced theirs first
sometimes its best to be the first to market simply for the fact that everyone thinks you did it first
now everyone will say "Google just copied them"
I think if imagen or another model can do text better, even if the model is less general many people will prefer it
Agreed, Googles 'Parti' does text excellently
Should be interesting to see how this competition will make the product better and better!
google doesn't plan to monetize imagen though from what I heard
They will likely add it to google images at some point in the future (I would imagine), but likely wont be free
Nuwa infinity isn't pre-trained
yeah that does sound like a good idea from them, no idea when that would ever happen though
No doubt Google uses its own cache of images for parti and imagen. The sources are wild, too much copyright risk.
I think that's 100% of the reticence
Well
Maybe also the insane demand for this tech with a Google brand
I think Google unknowingly trained Imagen on a dataset that included "pornographic and violent material" (LAION 400M), which is something they disclosed in their piece on dangers and limitations. I'd expect them not to release Imagen any time soon
I also think they'll be closely watching how OpenAI is dealing with the bias problem
I think Google saying they won't release IMAGEN to the public is a lie
But I'm also curious what a difference does it make to have an AI trained on NSFW images and then teaching the AI not to produce said NSFW images, and a model that doesn't know of anything NSFW
easier to train the ai on most images then trying to exclude nsfw ones I guess
Obviously, the difference is one generates NSFW and one not.
I mean, people are all expecting restrictions from DALL-E to lift off (in other words, they are training DALL-E to care with the NSFW info it has)
Chile, all of this tech talk confuses me, let me get back to generating 101 "a robot reading a book" stuff 💀
what's that
so the employees can generate porn
That would mean the model would have had to been trained with NSFW content, which it was not.
which model are we talking about
Are there any NSFW AIs in the works y'all know of?
Not that I'm looking for one 💀 Just got curious... if there are people wanting to create a safe AI, there for sure are people who want to create a not-safe AI 🤔
"Generator" of sorts, but not one I would reccomend people actually use
Actually going to delete it for safety reasons
Some creepy logos they have
Agreed, but Im sure you can imagine what the team behind a technology like that may look like
Stable Diffusion definitely has my interest once they open up, their results look very good
so you can generate celebrities with this one
cool
but it seems more like midjourney than dalle doesnt it ?
@violet wigeon will like to experiment with this model for sure lol https://twitter.com/ArtIsLight_/status/1548086649493475333?s=20&t=Idci0iladdzNomJ3J5AV5w
So how is a new project more advanced than DD / MJ
AFAIK this Stable Diffusion model is an amateur project, no?
Yay I’m back
Seems to be a side project from an AI company
Dont quote me on that one
This made me laugh lol need to find the original
I'm using it to actually save money... I can give artists references or alter their art and images to give them an idea what I'm looking for. Then they do the things that Dall-E cannot. Plus there's other areas I can't afford artists on to begin with, so Dall-E fills those gaps too. I find it incredibly cost effective compared to other visual options. Even stock graphics are incredibly expensive. There's more limits to how you can use these images, but I guess you can't have everything.
I'm incredibly happy with what it's generating. Today I generated the most ever. Couldn't be more pleased with it.
I think some people forget Photoshop mockups are a minimum of $15 A PIECE 😭 not a month, not per 115 mockups. A single mockup
You bring a valid point 👍
I find myself wasting my credits today generating images of bunnies eating donuts.
don't know why heh
Yeah, $100 doesn't go very far with Stock Photography (or many other image sources) but there's so much that $100 can do with DALL-E. a
Worth every penny, so cute!
Once there's a some cost associated with each image in a way it makes the images more precious... more like shooting on film than the infinite digital camera style that i've become accustomed to with having big hard drives and free access to ad supported internet media. its interesting
Himwa
My generation if I did that:
Mine look like stablediffusion
I think you're just too scared to use your credits now lmao
Its from Stability.ai a soon to be Opensource model Cough actually opensource
How long people need to wait until its available? Soon™
I do have access to the model, So feel free to share some promts, in #off-topic
How did you go about getting access?
It was a related project to stability to create a human rated generated image dataset
I can't really share a server invite, But still, things are active
You can check the Generated stuff using the model
Not the model by the way, This is a dataset
Weeeeee
Prompt plsy
"Digital art of an adorable innocent puppy joyfully eating a tasty strawberry donut during a beautiful sunset."
Digital art of an adorable innocent skunk happily hugging a cute white kitten during the aurora borealis
no skunk :(
scarie world thistle
dont know why it generatied that
dalle2 is racist to skunks they removed skunk from keyword because of weed huh why did they do that I am going to bomb the openai HQ for not allowing skunks
F
Professional wildlife photo of an adorable innocent skunk happily hugging a cute white kitten during the aurora borealis | pin on wildlife photography
is looks like a albino skunk hugging a spotted skunk (smollest skunk specifes)
or no es ferret
ferret or mink
mephitis mephitis
could try the latin name for a specific skunk species
okay lemme do that
Thoughts on best places to sell creations?
idk
WTF 😭
DALLE 2 IS RACIST AGAINST SKUNKS IN THEIR BID TO MAKE IT MORE "DIVERSE" THEY REMOVED SKUNKS FROM THE DATASET BECAUSE OF PEPE LE PEW I HATE OPENAI MORE LIKE OPENCCP
openai is down bad
😳 yikes homie chill out with the terroristic threats n stuff
skunks are one of the chosen animals by the kingdom of God which i dont know if that exists or not but skunks are very special yes
yayayayayayaa
Lets not talk like this in the chat please
not even coherent
this is bullshit fuck openai
where is my skunky skunk
we are all going to die this kind of stuff makes the entities not happy they will distort reality
Bad vibes man
Bruh, even this got affected by sneaking in "black"
Top right canary
Top right sign
black black
How do you enable dark mode?
settings
or something
I have a dalle2 chrome extension
it makes mine dark blue
lol yea i was gonna say they don't have a settings thing, only thing you can do is go to your payment options
i'll have to find the chrome extension
Was the default on phone for me
Probably in my browser settings, not by OpenAI
You can use the darkreader chrome extension
turns every white site into a dark mode site
this time you've gone too far
@violet wigeon it's hard to read tone online and even harder to read intent. But remember that OpenAI employees are members of this community, facilitate this, and don't deserve to feel threatened. I suggest you delete your posts.
I feel like I don't want to risk using my credits... I was doing max creations/day, now I feel aware of 50/month
anyone else feeling like this?
Omg this one is awesome
I’ll dm
leave them alone :(
they’re a small team trying to deal with the pressure of a million people
levitating god skunk for you
hello besties.
Hey, was it here on the server or on the FB page ppl were complaining not being able to generate pentagrams?
haven't heard about that!
Yep! It was in #ai-discussions
I figured it out
This server has grown so much it took me a bit to find again
Ill post it there
I believe they were trying to generate pentagons though.. not pentagram 🤔
sorry for the silly question but how do you set an avatar/pfp on dall•e?
I've already burnt through my $30 of prompts xvx
if you mean this, i believe you can only do it when signing up through your google acc
they go over really quick
but i'll check about that!
also yeah
i just burnt through 20 taking stuff from my friends in like 10 min 
L
I find that every generation requires at least 2 or 3 regens because of the often very bad generations that seem to come out of my first generation, and with the reduction to 4 images/generation it is abysmal. I really hope they increase that to 6 or even back to 10 by some miracle, but I don't know if they ever well.
honestly i have good luck sometimes but i agree with you on having to retry some...
all I can think of is this meme lmao
i loved these ones i just did 😭
Like, having four generations, it feels like just ONE more image spat out would've been the one
I miss having 6 gens lmao
god having 10 must've been heaven
With the backlash I wonder if they'll consider new options, i.e, increasing images per generation, changing price models, etc:
I almost hope that they pulled back the 6/10 image generation here to free up the servers a little for a rapid increase in new members, and they may increase it again once they're confident the servers can handle the capacity.
id be surprised if letting people pay to generate any number was not on the roadmap
like you refine a prompt, and then you make 100 concepts for the character
lol I'd be terrified that it wouldn't come out how I wanted
I'm not on the eng team but I can tell you it's almost 100% on the roadmap, they made the UI scaling and an easy variable change to request more prompts, very likely it will be an option at some point in the future similar to GTP3 return length
I wonder if alternative aspect ratios are also on the roadmap, or if the model is just not trained to produce outside of 1024x1024.
Man do I wish it was unlimited, but it's more like money limited
how they gonna keep the computies running 😔
won't someone think of the computers smh
It's just so insanely expensive atm. I'll have to wait for a subscription model or something when it comes out hopefully it's under 100 a month or something for unlimited
i get it, hopefully something more affordable for you comes your way!
I wish there was a way to give out prompts
like for giveaways n stuff
kinda like how we can gift discord nitro
would be fun to host a contest, then winner wins some dalle credits
Would probably take more credits then it's worth for the winnings lol
some people probably pretty good at prompting and editing now
Sure but if you want to win you would need to more effort
yeah, i also wanna do stuff w the bot when theres an api like let people generate through it with my credits (obv v limited but like a few a day idk)
omg i love this idea!
drop in #suggestions bestie
I’m actually surprised that they didn’t have a subscription plan for unlimited credits.
if they did that then 1 person would pay for it and do everyone else's as requests
Lol 1 person couldn't do everyones prompts, nor would most want to
Either way it would be very expensive to begin with so it would be fine
It also might have a soft cap
incase some users do like 300+ a day
😵

i'm reaching the last of mine, spent $30 and I only have about 30 left lmaoo
Those who can wish you the best creations
That's right, I'm happy with the images that have come out
I spent $30 on credits. Gonna make them count.
Isnt that a good doggo, is that a good doggo?
Good Doggo, swim doggo
swim doggo, swim, good boy
Love the water
Who would have thought that the beta would have been the good old days
tin foil hat on. first attempt gen being iffy on purpose so you go through more prompts quicker, so you are encouraged to buy the prompts 😄 tin foil hat off
I have 113 credits left, gonna hold onto them for dalle3 lmao
will the bonus credits transfer to another month?
The free credits reset month-to-month, if you don't use them, they do not stack. Purchased credits do stay in your account for several months. Not sure of the exact number off hand.
so the 100 free credits are gonna disappear next month?
noooo lol
~~uhm.... yes, I think so? ~~
Edit: No, the 100 credits count as paid credits and will last several months.
oh.. 😦 good to know, thanks
look at your account, click your icon and click on tokens
it will tell you how many "free tokens" you have available until a certain date, and below that, it shows your purchased tokens.
"We’ll invite 1 million people from our waitlist over the coming weeks. Users can create with DALL·E using free credits that refill every month, and buy additional credits in 115-generation increments for $15."
ah, okay.. so 100 are considered paid, neat!
"You have _ free credits to use until __________,
and ____ paid credits remaining."
yeah that lol
thanks 😄
Oh, cool... That's cool then. I'll edit my answer.
"You have 13 free credits to use until August 19,
and 100 paid credits remaining."
that's mine
hmm
ok cool
Or one day in my case 🤡
Sad reacts only pls 😭
How’s y’all’s day anyways
its 8:39pm but it was a good day
i want to watch a movie or a tv show or something but i cant decide
im very unique in this way, no other human has ever felt this
I woke up from a 3hr horror dream I had and I feel drained 😭
i had a really horrific dream a couple nights ago, you just reminded me... its actually so intense I wouldn't even repeat what happened tbh
It's funny how we go to sleep so we can replenish our energy and our brains are sometimes like "Nope, here's the most realistic awful situation you can possibly be in. Enjoy!!!"
i wonder if dalle has anything to do with it................... probably not tbh haha
im very confused. Is DALL-E available at the moment? or is what everyone is using DALL-E 2?
Here it says only DALL-E but I haven't found anywhere dalle 2
can someone clarify what im using at the moment
DALLE·2 is what you're using
DALL·E is the brand and this is just the second iteration of the model, maybe based on e.g. the size of the dataset
heyo folks, does Dall-E accept payment from paypal or other? instead of credit cards





