#community-help
1 messages · Page 2 of 1
Question: Is there a list of words that are off limits?
For example: Say I want to do a prompt of "Star Wars". I don't know if "Wars" will be flagged.
can we invite a team member and let them use the dall-e credits if they don't have an invite yet?
Unfortunately not - guide them to the waitlist. Hopefully DALL·E will be open to all who have requested access soon!
The filter is often contextual, so words might work in some instances and not others.
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.
Do free credits expire at the end of the calendar month or some other date? Is there a way to tell when they will expire?
What is inpainting and how do I do it
Hey I joined the waitlist for Dall E 2, seems i got into Dall E 1? Is there a way i can access Dall-E 2 and download it?
i got my invite for Dall-E 2 but i don't understand where and how to create images
You got into Dall-E 2, not 1
okay thank you
Do i always have to pay for credits to generate images? I tought we were meant to test
Ik it seems kind of worse/disappointing
Do i have to generate images on the web and not download an app and do it?
You get 50 free credits at the start, and 15 free credits every months
Web
what page
But you have to specify a lot to get good results
Is it possible to use the image edit feature to add a background to a transparent image? I notice you have to erase part of it, however the interface already recognises the transparency
Is the erasure required? To perhaps influence the model to edit that "area" instead of the whole image?
You got an email for the invite.
Go to that email and press the "Get Started" button. Then, create an account
Then simply go to https://labs.openai.com/ and type in a prompt ^^ You should see a text entry box after logging in
how to get dalle2?
that's not Dall E but Dall E 2?
I don't think you need to erase if there's nothing in the background
That is Dall-E 2
how do you know
It still requires me to click the image once with the eraser. Just wasn't sure if it mattered where I clicked
It's 2, just doesn't say it
okay
Because 1 is way worse
how do u tell em apart
How do variations work ? does it keep the original prompt in mind or does it just make a similar picture ?
If you've used 1, you'll know lol
To me it looks like theres no AI it just puts together images from google
I think the beta testers got a worse version than the youtubers who revieved it from what im guessing for safety reasons?
Because they don't let you access Dall-E 1
Similar picture
How do i get access to the more sofisticated software the one that marques brownlee?
This is that one. The fact you're in this discord proves it
You just have to refine prompts, he doesn't show that process
How do i make sure my prompts are what dall e is trained on
like how do i get those commands and write them to ensure dall e understands me
also the paid part really disappointed me.
Make your own ai then lol
Its not even an AI mate.
The models are computationally expensive to train and run. Who pays for the thousands of hours of server time spent training the models, as well as the countless human hours refining it?
Only developers should be here not the people who talk like they know everything.
Sure thing man, whatever you say
so you definitely shouldn't be here
Most people here are computer science students, people in the field or artists anyways
Very few who are not
People who are developers know Dall E is not an AI after seeing the results
Can someone explain the workflow for inpainting? I saw the one image guide but I don't really understand how I am suppose to move the image over
Speak for yourself
are you looking for a person to debate with and fight? I am not. So stop talking to me.
the moocher is just mad ignore him lol
Not how that works
Good results or bad results don't make something an A.I
It's the Artificial part and the Intelligence part that makes the A.I
Its clearly using images from the internet not creating its own thing.
Yeah this is getting off topic and clogging up the questions channel anyhow, burying valid questions
❤️ Check out Lambda here and sign up for their GPU Cloud: https://lambdalabs.com/papers
📝 The paper "Hierarchical Text-Conditional Image Generation with CLIP Latents" is available here:
https://openai.com/dall-e-2/
https://www.instagram.com/openaidalle/
❤️ Watch these videos in early access on our Patreon page or join us here on YouTube:
- ht...
It uses the CLIP method and a diffusion based model to create these images
bingo
There is NO image on the internet of a cat wearing Napoleon outfit
so you admit i am right
well you did admit it
How so
Has anyone found any interesting "style modifiers"? I've found appending digital art to the description creates some cool looking results
digital art, trending on Artstation, extremely detailed, hyperrealistic, high quality, sharp, 4K
I have a question about the content policy. As I understand it, I can't use DALL-E 2 to create any images that depict self-harm. What about using images (that respects the guidelines) in a book where the content might include self-harm?
I have been writing a short story for a while now, and thought about using DALL-E 2 to include some illustrations, mainly to showcase the design of the main characters. In no way will I use it to make an unsafe image, but if I have a part of the book where a character, say, commits suicide, am I still allowed to use DALL-E 2 images in the book?
i dont see any problem w that
How does "by (insert artist here)" work? I've been trying manga artists but haven't been getting any results
I think it replicates the style of famous artists, typically historical ones
I was thinking the same but saw some in architecture using modern architects and fashion designers
Yeah if there's enough pictures out there of a modern person too it'll likely pick up the style
Try HD Digital Art
marques brownlee same prompt look how different it is.
Feels like we are getting a worse version?
or am i missing to put something in the prompt to get an image like that=?
Well, it is non-deterministic, so you can't really hope to get the same output
And this might have something to do with it, too. There are some style modifiers that you can append that really changes how good the output is
DALL-E 2 is an AI that can draw anything you ask it for. It's terrifying and amazing at the same time.
Tim vs DALL-E: https://youtu.be/MwAAH9tBoMg
MKBHD Merch: http://shop.MKBHD.com
Tech I'm using right now: https://www.amazon.com/shop/MKBHD
Playlist of MKBHD Intro music: https://goo.gl/B3AWV5
he doesnt say he wrote a prompt
also he says it spits out 10 different versions
why does our split out 4?
So more people can use it
https://dallery.gallery/dall-e-ai-guide-faq/ all your questions are answered here in the FAQ
For "a bowl of soup that is a portal to another dimension", I got this (out of 4). Sometimes I get something fantastic, sometimes, it's meh. Personally I like this one but YMMV.
Does anybody knows how can I change my gmail account of my dall-e account ??
I don't think you can
If there a reason DALL-E has so much trouble with text. Any time I try to add text or things like say a milk carton to an image it gives me a very blurry interpretation of text.
Nice cherry picking of his results. Though as someone shared, you can get results like that
From Google's research with larger models, it seems you need training with several billion images before even explicitly-prompted text becomes possible. DALL·E can get certain commonly-seen phrases correct.

is it ok if i use one of my Dall-e images as album cover art for my music?
Yes!
thanks for answering
dall-e 2 is still in beta isn't it?
and if so does that mean prices could change?
thank god , tbh it should be more leniant with the amount of prompts , maybe 50-100 prompts a month . Or one or a few prompts a day .
Its not a fact of wanting to limit users, each prompt takes considerable compute power to run. If they gave everyone 100 for free every day that would cost them massively in compute costs. They are undoubtedly working on ways to make it affordable and usable by the masses and will come up with techniques that will bring down the costs :)
I've noticed a bit of a sour vibe around pricing that people in other forums (maybe here too) are describing as affecting their mood, desire to use dall-e, and ability to get into a flow state. The current pricing model has the ability to pull you out of a productive state of mind as you prompt engineer, to a purely micro-transactional one.
I hope we can quickly get to the next experimental pricing structure!! Any thoughts? How soon can we expect some different pricing options to test?
oh ok , hope a way is found soon
I missed out on free credits because my phone number was set up on another account by mistake (didn't realize I have to register with the same Email I requested access with). Any way to get these credit points?
@zenith python its a brand new technology that people didn't even think COULD exist 3-4 years ago, the fact that we have early access to it and that it only costs 13 cents to run is already insanely cool. We are on the absolute bleeding edge of this field and many other competitors will come up in this space which will drive down costs and offer whole different models and results. Patience is key :)
Probably when user numbers have been stable for a couple of weeks, at a guess
in retrospect, i think the user survey (which I filled out gladly) could have included more about how different price structures might make you FEEL, which overall affects usage. if i feel i'm being charged per cornflake i eat, i'm going to eat less because it simply feels unpleasant. or if i'm being charged per key stroke on a piano rental. it's a mental gremlin that many artists are absolutely being affected by.
to unlock dalle's potential, we need pricing models that don't interfere with the creative state. it's counterintuitive on the surface, but charging per image causes a valuation judgement, of $0.13, for each and every iteration, and it's hard if not impossible to turn off. it's like a big bell going "DING DING DING DING" with each prompt 🙂
it's the schoolyard marm saying "you have 12 minutes left of recess. you have 11 minutes left of recess. you have...." 🙂
I agree in general. As this conversation is spanning several help and feedback threads already, I will play a different schoolyard marm and remind everyone that #ai-discussions was brought back for this kind of conversation, and we won't even mind if you're hotly critical - as long as it's fair and non-hostile
Q:- I want to know, what if any person creates nft of image generated by DallE-2 ? What are policies for that ?
It's fine
oh i thought it was going away. i'll bring my ball over there.
It was, it's coming back for this kinda chat 🤘
heya! I was wondering if the content policy for violence would prohibit making fantasy style characters wielding medieval weapons as I'm looking to generate images for a character of mine to use as mental reference
@past geyser, sorry to bother, but could you give me your opinion on this matter?
Hmm. As a writer, I'd be inclined to illustrate these scenes with inference rather than explicit imagery. The content policy is G-rated for a reason; it's meant to be globally accessible tech. I don't think your circumstance, albeit noble, permits exceptions to a clearly worded condition of use.
Rereading your post, I see the complexity - even if the images are benign the context is potentially not. I can't answer with certainty, sorry!
I see, thanks for the input. I might have to contact them to be sure, then
I think you're probably ok. If it were me I'd do it and worry again closer to publication...
Autobots arguing the Decepticons in deep space, Cyberpunk mixed with Surrealism, neon, leds, smoke, 15mm wide-angle lens, sharp, 8K high resolution
why does this trigger unsafe warning
Decepticons is somehow banned, most likely because of the word deception
another question why did it costed prompts
even if it didnt did them?
It does not.
clearly saw 114 drop to 113 should i talk to support if this happened?
It shouldn't have dropped, and yes, if that seems to be the case, email support@openai.com
Are we permitted to sell the content we produce? Are we permitted to edit the outputs using other editors and then include those in say articles or posts or videos on YouTube including for profit?
Btw, I tried your prompt and found Decepticon blocked. I then tried the following:
a still from the movie "Transformers", of Optimus Prime arguing with Megatron in deep space, Cyberpunk mixed with Surrealism, neon, leds, smoke, 15mm wide-angle lens, sharp
No flags.
And then replaced "Megatron" to Decepticon
a still from the movie "Transformers", of Optimus Prime arguing with Decepticon in deep space, Cyberpunk mixed with Surrealism, neon, leds, smoke, 15mm wide-angle lens, sharp
No flags. Hmm.
DALL∙E uses not only a list for blocking words, it also uses context. The first phrase a still from the movie "Transformers" might have cleared it from getting flagged.
Yep!
But please read the content policy & terms if you're unsure!
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
Thank you @verbal crag . Now if only. I can produce something worthy!
Experimentation is key! There are helpful resources over at #faq-and-resources-old 🙂
Context probably matters. Here is "fisher-price my first battleaxe, brightly colored plastic toy, large, gray metallic plastic blades, made in china, natural indoor lighting. product photography."
Why is gun a flagged word?
Hello, I tried to use the prompt “Elon Musk as a character in Kingdom Hearts” and I was told it was in violation. Can anyone confirm why?
Is there a way to search your generation history? I have a prompt that I know I generated (because I have a screenshot of the results) but I noticed that it is missing from my saved collection despite me saving every result I get? I searched my browser history and its not there either strangely. Its making me wonder what else hasnt saved?
Elon Musk is a real person, that's why
So is Hulk Hogan but it let me create “Hulk Hogan in Super Mario Odyssey”
Please read the Content Policy
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
At the moment, there's no other way to view your previous results if it does not show in your browser history. I recommend you saving them to your collection as it supports up to 10,000 images.
I do save ever single result though I just noticed that sometimes some of them 'fail' and I guess I missed these
same some just disappear lol
Will there be a way to search in the coming future? I noticed that if you find them in your history they are still technically there
@patent rain Yeah it just worries me because like, im using these for a project as a moodboard and ive been generating stuff for 2 days basically, and now im wondering, what % of all those amazing images that I saved, are gone
Because not all of them are super memorable unique images, I generate a lot of similar prompts and so Im not really going to remember that easily when 4 are missing
So yeah getting some clarification from the developers on that would be quite important, these are paid credits afterall
For example how long are unsaved images stored on the servers before they are deleted permanently, id definitely like to go back and check if any are missing
But with google chrome search history, that dosent appear to be so reliable/easy
They're definitely stored forever, but sometimes Google search history combines items or the link is just not saved for some reason
Put forever in quotes though; openai have said they don't guarantee they'll be stored for the rest of time, but as of right now outputs from months ago are still available
Yeah google search history does not appear to be so reliable, you can also search an exact term and It dosent show any results, yet you can manually find results that it dosent find..
Ok thats good to know, I mean Its early days so fair enough, its not going to be perfect, I just wanted to make sure that my images werent going to be gone next week. Is there any chance we can get some clarification on future changes to this sort of thing, a simple all time prompt history would be amazing
If there was a "download all" button, or you could use a checkbox to select multiple images (both in the main generator and your collection) and bulk save/download, I think that'd be a lot better. Even just a small heart icon to immediately save to collection.
Actually I'll pop this in #suggestions
How can one change one’s password?
How many languages can i use in?
How do I save images to my collection. Also how can i avoid the square size cutting off the edges of my generation
Do I have to add commas?
Can I invite someone?
No
Hi new here, my question is how can I download image without the Dalle logo on lower right corner
If you use Developer Tools in Chrome, you can right click on the image and select "Inspect Element" there will be a URL in the code that does not include the watermark. Also, if you use the mobile website, tap and hold on an image and that downloaded image does not include the watermark.
It's a known issue to some extent, based on how the training data was cropped. Use padding features in #dall-e-bot or learn some advanced prompting techniques.
Have to? No! DALL·E is however punctuation sensitive
@past geyser it just doesn't work great when I pad I've tried quite a few times. How would you prompt it?
When you've tried them all, let us know 😅
How do you access the uncrop feature?
Upload a zoomed out version of your photo and press "Edit" in the website! There's a bot here in #dall-e-bot that resizes your photo
How are credits spent when using DALL·E 2?
It is 1 credit per prompt or similar?
Correct. 1 credit per prompt/generation.
More info here: https://help.openai.com/en/articles/6399305-how-dall-e-credits-work
Ah, thanks!
So, one thing i can't seem to find and answer for: If i generate an image to be used as a logo, can i actually use it? (to replace the one I'm using as a profile picture and Game dev logo)
Yes, commercial use is now allowed. The Terms of Service can be found in the #old-server-rules channel under Rule 0
Thanks
@boreal oyster As of now changing the style of an image is not possible. Maybe in the future!
Prompt should be made by describing the whole image you want to see.
ah ok that's a bummer, so i just erase the part that i want dall-e to add info to?
is it possible to get an estimated prompt of an uploaded image? i understand it's not available in the frontend, but is it technically feasible?
Is there a list of all the drawing styles dall-e knows?
@drowsy coral
example if you have a low resolution image upscaled with a noisy alpha channel, could it fill in the blanks and make it higher resolution
DALL∙E is not an upscaler, it rather, "fills" in the rest for you. There's a thread here that you may want to read. Let me look it up!
Possible as of now? Nope. Technically feasible? I'm not sure. You'd have to ask a OpenAI staff when one's online.
Check #faq-and-resources-old
from my understanding i don't see the difference between upscaling and filling. but i can try some experiments 🙂
https://discord.com/channels/974519864045756446/988918555288408064
Re: varying transparency
if i ask "A dog looking at the camera" will it understand that the image needs to have a dog looking straight on?
i wanted to ask:
"An anime drawing of a Shiba Inu looking at the camera with a butterfly on its nose" but i don't know if "looking at the camera" might be taken literally as a dog looking at a camera
I think you need to be specific. In my experience DALL•E understand but its not like i really want...
didn't catch it that well xD
Nice
I've seen people do "zoom outs" presenting existing images. How is this done? Is this simply resizing the image and converting the outsides into edits, or is there an on-site feature to do this?
I've also seen this, although it lacks a tooltip so I'm unsure of what it does
Is this simply resizing the image and converting the outsides into edits
Exactly!
The tool is for resizing the "brush" when Editing.
Aaah - Thanks!
How to formulate the phrase for editing an image, e.g I am uploading a picture of a dog and I want it to put a hat on it. Do I just say
Put a black hat on the dog's head
?
Hello, unfortunately no, it won't be refunded. The manual verification is available for those who do not want to spend a credit to get verified.
Describe the whole you want to see:
a dog with a black hat on its head
a dog wearing a black hat
thanks
i had to use up one credit to verify the discord code? thats weird. how do i get that back?
The benefit of being in discord with us is not worth 1 credit to you? 😄
do i need to buy new credits once they are all used or will they refill each month?
ah sweet
👀
Apparently I used a different email address for my GPT-3 and DALL-E 2 signups. Can I get my invite moved so I use the same account for both?
Hi! We have no access to OpenAI systems here. You may send an email to support@openai.com and ask if that's possible 🙂
Thanks, will do!
is it possible to change the aspect ratio of the images?
Welp, apparently uploading parts of a face are also prohibited. I uploaded my beard and wanted to see it on a random dude hahaha. Got warned. Hopefully I won't be banned now. 😭 Anyone else been warned before?
Q: Why can’t I upload my own face?
• To prevent deepfakes, we are currently prohibiting uploads of all realistic faces, even when the face belongs to you or if you have consent.
• This is because we don't yet have a scalable way to verify whether you have consent, or whether a face belongs to you. So we prohibited all face uploads.
Click here to learn more about the Content Policy.
It would seem that it generates faces though - I did a request asking for people working on Laptops and it gave me a lot of real looking faces, so it seems to only be an issue with uploading. The warning will probably look at the image and request, although with the amount of new users it's unlikely that a mistake like that will cause much of an issue.
Been warned for a variation of the image that dall-e gave. Seems pretty random
Yes, it will detect a realistic face with its own generations.
Wasn't even a face in my case. My collection is now empty, funny.
is it possible to change the aspect ratio of the images?
Nope, it will only output 1024x1024 for now.
Head over to #inpaint-outpaint and there are tricks on getting higher resolutions.
the resolution is fine ig but i want differen ratios.. 1:1 aka square gets kinda boring
Yeah, you can extend it. Check the bot at #dall-e-bot for the "!composite" command.
Sorry, I cannot find this answered, but how long are the "Recent" saved?
The history sidebar saves 50 sets of images. Others can only be accessed through the saved collection or your browser's history.
Oh okay, so are those saved forever?
in your collection they're saved forever (limited currently to 10,000 images). at current i don't believe OpenAI have deleted sets from the past that you can access through your history, but they haven't committed to keeping them forever.
thank you!
I tried to get Dall-E 2 to generate an image containing an Anglo concertina but everything it generated had a picture of an accordion - close but not quite right. Is there a way to teach Dall-E what an Anglo concertina looks like?
Q: DALL·E doesn't know about [X]. Can it learn them?
• While model updates aren't frequent, we can definitely consider adding licensable data for it in the next training round.
• Please tell us more about it through this form.
am i stupid or what
i go to the website and there is now no button to log in and get to dall e
thank you
Can you specify an aspect ratio or pixel dimension? Is there a pixel size limit? Why are all the examples square? I didn’t see anything in the prompt book
ya everything is 1024x1024
Quick question -
Do the initial 50 credits last a month from the day they were received, or until the end of the month?
Free credits expire one month after they were granted
Ah, alright, thank you!
If I would purchase credits (say 115) and a month passes, do I keep the unused credits?
Yep, they last 12 months
Hmm - Is it just me, or are several fictional characters excluded? It doesn't recognize "dobby the house elf" as the associated character. Is this simply a case of the characters being too niche, or "house elf" being too broad?
What in this prompt is triggering content policy? I’ve been deleting words trying to find it: A film still of a man wearing a dark grey quilted medieval gambeson, belts with tattered rope and cables around his waist. Many coins are woven into the ropes and cables. He’s wearing tattered layered linen scarves around his shoulders. full body wide shot
Does the "Do not attempt to create images of public figures" rule include those related to characters portrayed by actors? (Eg: A prompt depicting a scene with the Terminator)
Hello! I just activated my beta account and it is working great, but will be changing phone numbers soon and would like to know if there is a way to change what number the system uses for 2-factor authentication.
https://labs.openai.com/
-> Click this icon at the top right
I've only tested it on a PC so am unsure about the mobile version
It isn't available on the mobile site
But images you generate on the mobile site will still appear on the desktop history panel
I recommend saving your prompt page urls so you can view previous generations even after they leave your history
The history panel keeps the last 50 prompts
Multi-Question about commercial sale
Preface:
- I read we need to mention that the image was created with the aid of OpenAI
- We're to keep the watermark in the image
- I see the OpenAI metadata in the file itself
===============================================
Q1) Can we modify the images?
Q2) Do I need to re-add the metadata back to the file, if my editor stripped it out?
Q3) If I resize the image, do I need to re-add the watermark at scale?
will we be able to run dalle locally on our own machines and not have to use credits?
I apologize that this is probably a stupid question but I asked to be on the waitlist for Dall-E 2 and it brought me back to the original Dall-E page, which I already have access to.
And the link above to log in just says Dall-E also, but when you join this discord it says it is for Dall-E 2.
Are people here using the Dall-E 2??
Hello there!
Today i gained access to Dall-E but i thought they would give access to Dall-E 2 😭
either way, i am having a lot of fun with it
does anybody know if credits are limited? i mean, once i run out of it, the only way to get more credits is by buying them?
is it possible to save a whole session, or does it just never get deleted?
thanks!
same doubt here
This is DALL•E 2 🙂
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
@heavy merlin ^^
DALL·E mini was not affiliated with OpenAI?
Nope, and OpenAI asked them to change their name to avoid further confusion
How many blocked prompts do you need to hit until you are automatically banned?
I'm not sure if this question fits here, is there another place to ask this?
what is the current license structure regarding the generated images?
Can we publish? Do we need to state that it is generated by DallE?
yeah I was about to ask if the generated art is free to use
i think as long as you keep the dalle-e watermark at the bottom, it's ok
ok
I'm getting blocked prompts for innocent things. I uploaded a photo of myself to see how it could edit it. It was a close up of my face at the beach, but since you could see the top of my torso it said it was a content violation probably assuming it was nudity. Another was a superhero fighting a super villain. and another one was trying to re-create that sketch of elon musk with a baby operating his body from the inside
At what point might I actually get suspended? Can I appeal any related suspension?
it's You'll wanna read this https://openai.com/blog/dall-e-now-available-in-beta/ @floral peak @crude mango @rose laurel
Using DALL·E for commercial projects
Starting today, users get full usage rights to commercialize the images they create with DALL·E, including the right to reprint, sell, and merchandise. This includes images they generated during the research preview.Users have told us that they are planning to use DALL·E images for commercial projects, like illustrations for children’s books, art for newsletters, concept art and characters for games, moodboards for design consulting, and storyboards for movies.
so even with removed watermark?
@vast warren read this doc - https://github.com/openai/dalle-2-preview/blob/main/system-card.md
That's what I'm trying to find out
in this message
Non-users / third parties who find that their images have been used without their consent or that violate other areas of the content policies can report the suspected violation to the OpenAI Support team (support@openai.com) as outlined in the content policy, which is publicly available and discoverable by users and non-users both. A limitation of this reporting mechanism is that it assumes an individual would know that the image was generated by DALL·E 2, and would therefore know to contact OpenAI about their concerns. We are continuing to explore watermarks and other image provenance techniques to aid this.
judging by this I'd say you gotta keep the result as it is
I don't see the part that answers my question?
Hmm, I thought it was in there... 🤔 We can't upload images containing photo realistic people and there are certain terms that are just out right blocked for now
I thought uploading photos of yourself was allowed. It just says to not do it of other people without their consent: "Do not upload images of people without their consent, including public figures."
are we going to have unlimited access in the near feature? i mean, will we be able to enter prompts without credit limit?
Yes, but a bit further up, it also says "To prevent deepfakes, we are currently prohibiting uploads of all realistic faces, even when the face belongs to you or if you have consent." - https://labs.openai.com/policies/content-policy
Ahhhhhh oops. Thanks for that.
No problem! :)
I knew it I saw it somewhere!! 😄
Only if you pay, they gotta pay the bills
or, join the company 😉
happy to join 🙂
How to Apply to OpenAI?
I have a question about the filter. I just wanted to get a pistol in my scene. just for style reasons and got a warning.. Another time I just mentioned the word body and got a warning as well
How can I tell which isn't allowed?
just wondering if i can stream dall-E content.
You can
Can I edit a stock photo of a person, blank out their head and face, then upload that to Dalle to generate a new image of a different person? I want to change the person in the stock photo to someone with a different description. The Dalle rules appear to allow this (because the blanked out stock photo uploaded to Dalle has no human faces), but I want to verify.
Hey so I just spent 30$ on credits and my card was charged but I have received none of the credits I should’ve
Does it take them awhile to be added to your account or do I need to contact support
i bought some today and it was pretty instant
at first it never charged me tho and i never got them even though i clicked purchase
Alr if I don't get them soon I'll contact support
Hello. Is this a good place to ask some questions about what's allowed? I've gotten a couple prompt warnings and just want to make sure I'm making safe prompts. Is it okay to use brand names, names of movies, videogames, or toys?
I know real people are off limits, but what about Garfield?
Hey, how do I expand an image?
Is there some way to generate 2000x2000 pixel image
From the prompt book it say 400 characters is the limit
Ah, I guess it's just truncating the rest?
On this i have no idea
i think it's against tos
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.
What are the rules regarding sharing images with the internet? Do i have to mark the image with Dall E in some or all cases?
please @ me if you answer
That's a statement directly from OpenAI
Oh ok thanks!
Yeah, Joanne provided that statement a few weeks ago.
what happens if you try uploading an image and entering a prompt without erasing any part of the image?
is DALL-E available on mobile, is there an app or just browser on phone?
Can we use the images for commercial use?
It would ask you to erase a part
Are you able to change the phone number and how?
Are there any actual dall e staff here that I could talk to about an issue I had buying credits so I could either get a refund or get the credits
Just browser
Yep, with the most recent update, you have all commercial rights to your image
Full details are on the most recent blog post
Should be okay, so long as the person youre trying to generate (in your prompt) isnt a real person
Please read these pages before asking questions and try searching this channel, a ton of these are duplicates
Please read the Content Policy
@short salmon maybe you could review this set from earlier? 🙏
Can't use the word "poop" in prompts?
Perfect From now on I am gonna create bunch of nfts from Dall e phew no need to draw myself anymore XD
Make sure you follow the rules, like the watermark
It's ok to remove the watermark 🙂
Hellllooo, how do I pick the image to perform DALL-E bot actions on (in my case !pad 50)?
Where can I read that to avoid legalities?
The images you generate are yours to do with as you please :)
And then you type !pad 50 in the before you press enter
@slow gull
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
i just did a prompt but i dont see it in my recents anyone know why that might be
@patent rain do u have any ideas
no like it generated perfectly fine and everything
but i just looked at them and went back to the homescreen
and theyre not in the recents
do the 50 credits you get when you were invited expire after a month as well?
all the free credits you receive have a monthly timer, including the 50. paid credits have a 12 month timer i think
Yes! They do! Proceed to the credits page and you will be shown the exact day they will expire on.
does dall-e use CLIP?
Is there a faster way to publish a multitude of images?
Might be a bit of a dumb question but how do you change the public name displayed with your OpenAI account? 🤔 Might be because I'm in an organization currently... hmm
What are the different channels? Most of them don't have a description so idk what's in them
Thank you for the feedback!
Is there a particular channel you don't understand the purpose of?
if i'm trying to uncrop an image, do i have to provide a description of it?
and how detailed should that description be if so
Hello! Do you guys need to hire any mods?
Check the latest #announcements 🙂
ngl, i am half asleep, so i'll probably understand them all after i get some rest. But just having a description to focus on is nice. I think I figured them all out tho
Wow grreat timing lol
Is there a way to see all the images we have published?
and where do published works go?
You need to Save them after you generate them.
Click save on your fav images. Then head to My collection.
Ah yes, I've saved all my images but some of them are published
I can individually check them and see if they are published or not
I suppose there is no other way other to check published status other than individually checking each of your saved images? @hoary wren
Yes I believe that is correct at this time.
Thank you for your help :)
when will dall-e 2's model become open source?
or training dataset
Am I the only one who feels like the AI generations suddenly got worse after yesterday's downtime?
The model is static, and can't be changed easily. There have been rare instances where an input-side glitch has caused noticably different results in the past but they were resolved in about an hour and literally everyone knew 😅
20%
For contrast, the six images were generated July 14. The four images were generated just now. The model is static 🙂
Thanks for the info, guess I just got a few unlucky generations 👍
Anyone know when the text diff feature is expected to be released?
Does Dall-E have a subscription option, or is it only the credit purchase option available?
Where can I read about copyright regards to using the AI in commercial works etc?
For now, it's only the credit-based purchase that is available. For the terms, refer to Rule 0.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Are you allowed to make horror suff as long as it doesn't have violence or gore?
As long as it's not breaking anything in the Content Policy then I don't see why not 🙂
Can I use these pics on social media platform or as NFT please Guide Me
Yes but please read the terms before you do so.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Thank You
Thanks
@trail sundial – hi, when uploading, instead of clicking "Variations", click on "Edit"! 🙂
But than I am asked to erase parts of the image, what if I want to edit the entire image?
Currently, that is not directly possible, but one user experimented and found these results, I recommend to start reading from the top: https://discord.com/channels/974519864045756446/988918555288408064
Does there exist any collections of wisdom about how to best use Dalle?
Ok, thank you!
Yes! Do check #gpt-models
Does somebody explain to me how to generate a 16:9 image instead of squares only, or isn't this possible?
*can somebody explain to me
Can someone explain how this is done? https://www.tiktok.com/@omarxrc/video/7122605900686298373?is_copy_url=1&is_from_webapp=v1&q=dall-e&t=1659167895317
I'd guess they padded the image with some blank background, and then they uploaded the image, chose edit, erased the padding, and asked it to fill it with whatever they want
New to DALL-E 2. from my initial prompt I clicked variations and got exactly the same image 4 times, also tried a modification to the text prompt re-generated and got exactly the same image 4 times. What am I doing wrong, please advise.
hi, i can ask about any question relating to openai, not just dall e right?
Is it possible to define the resultion of the image or is it always set size?
always set size
probably
Does support@openai.com reply on weekends?
@torpid meadow ah okay
They're currently still a small team, expect delays! But who knows, maybe they'll get to you soon 🙂
understandable, I appreciate the reply :)
@untold python – you can change it only if you're using DALL∙E with your Google account.
hello, please tell me what's wrong in the following sentence, I didn't find anything that breaks the rules in it:
elon musk racing with mr. bean for a penny, digital art
As per the Content Policy, celebrities are not allowed to be generaed.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
but it's not stated in the policy
shouldn't be? it's quite confusing when you're getting warned due to policy and policy says nothing about celebrities
That's a grey area and I can't comment on that for now.
Can I share my account with my closest family and let them use my credits (i.e. mom, dad, wife)?
Is there a list of all possible styles you can add to your prompts? for example:
- 3d rendering
- synthwave style
- photograph
- cartoon
- pencil drawing
- abstract oil painting
- and so on...
Take a look at this https://dalledictionary.notion.site/dalledictionary/DALL-E-DICTIONARY-2ec3dd6dc2184a45a9b6b4b8a021f0a6, and also at #top-tips-and-tricks! :)
Some have been sharing the technology with their family! But just to be sure, do read the terms found at #old-server-rules 🙂
Plus the resources at #gpt-models!
exactly what I was looking for, thanks a lot! 😉
Yw :)
hi, i can ask about any question relating to openai, not just dall e right?
Yep! We'll try to answer it
@mighty sierra – check the resources found in #gpt-models + #top-tips-and-tricks
Right, so this question is about app review
do i have to implement all of these doesnt matter if my application falls under the category where you have to make minimal/standard safety precautions?
Right, this is awkward. That is out of our reach here 😬 Have you tried support@openai.com?
Is the chat feature in the site sufficient?
That could work! But I think they're off for the weekends. Try again on Monday perhaps
they are
can i send a message today and wait for an answer on monday?
How can i exceed the number of accounts i can add into my phone number when creating an openai acc?
They're a still a small team so please expect delays. Won't hurt if you email now though!
I understand and I can wait until anytime, so can I do that?
I believe the phone number to account ratio is 1:1 – I may be wrong though.
Yep! Drop an email now and wait for their reply.
Email.. or.. chat?
It's up to you, whatever you prefer 🙂
How do I expand the generated image?
There's a bot at #dall-e-bot that resizes your image. Resize then upload it to DALL∙E, press 'Edit' and generate! More examples in #inpaint-outpaint #—post-dalle-edits
thanks you!
Have a look at this too https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y, the Pages 59-68 should be of interest to you :)
Alrighty, will take a look at it!
Is it possible to send invites ourselves to friends? Or is everything done through the waiting list.
The Discord is solely for people that have access to the early access version of Dall-E 2. :)
Oh I mean invites for access to DallE 2 itself
You can absolutely invite them to the waitlist 😊
^ that :) Mb for misunderstanding your question Alex!
Ah ok, no way around the waitlist then. No nepotism 😛
how do users in #animation make the animation video?
You'd have to ask them I'm afraid 😬
Is this indeed DallE 2? Just says Dall E so wanted to be sure
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
Thank you Bot!
Why are some of the images cut off at the top and bottom? It seems like the whole image is generated but then cropped?
how strict are openai with uploading public figures
even if ur prompt is really harmless
is it immediate ban if ur caught doing it?
does lowercase/uppercase matter?
How descriptive can you be with DALL-E? Like, how well would it try to discern everything in your prompt if you dumped a paragraph of descriptive information (or rather, used up the character limit)
no DALL-E is not case sensitive
is there a limit in time or size for the history ?
I seem to be getting some missing images. I can see them if I click on an image, then publish it, but refreshing the page resets this. Downloading works fine. Am I doing something wrong on my end?
How to use Dall-e on PC?
happens to me too, I thought my phone was glitching
seems to be a Safari (15.3) issue, works fine on Chrome
Hello All! A total newbie here so I have a lot of questions. First and foremost, how do I get high res image? I downloaded the image, which is only 1024 x 1024, and I can't find an option to render it at a higher resolution. Thanks in advance for your time and help!
Hey there! Unfortunately the maximum export resolution at the moment is 1024 x 1024. Feel free to ask any more questions you have here...
Disclaimer: I am not a volunteer or an OpenAI employee, purely another user helping out.
to answer the latter part of your question, quoting the server FAQ.
While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
Disclaimer: I am not a volunteer or an OpenAI employee, purely another user helping out.
thank you
Does anyone know the link to unofficial search engine for Dalle2 in the prompt book v1.01?
I could not open it.
Yes the said search engine at https://dalle2.app/?search=style seems to be down for the moment. I was in contact with the creator a short while ago regarding another project, will try to get them to take a look at this too. In the meantime you can try:https://t.co/k4ltdIFq57
Disclaimer: These are unofficial tools which are not affiliated with OpenAI.
Thank you, Avin!
Happy to help buddy!
Thanks for all your help
Is there a way to make a wallpaper, without photshop?
Oh, that's a little disappointing. Thank you, Avin.
maybe a dumb question, just curious, why those colors for the squares? Yellow/Blue/Green/Orange/Blue
You'll need a photo editor
But theres good free ones, like GIMP
I'll try out GIMP, thanks
Still wondering how one can change a password
No problem!
did you use a sign in partner or direct email sign in?
is there a way to change my phone number in open ai?
Yo! Wanted to ask how far the "public figure/celebrity" policy goes, like does it include movie characters like yoda and sucks?
Can I still invite people?
Nope unlike a traditional refferal system, DE2 works on a waitlist through which OpenAI solely accepts users.
Just got a ban warning for " by Pierre-Auguste Renoir", I thought this is just an artist/style, so I figured this must be a mistake it blocked me from using it? Anyone have any idea why it would block this?
While I dont have the exact answer to your question I remember OpenAi documentation clearly stating that changing your email or method of authentication is not supported. So chances are it's not supported.
Quoting server FAQ • Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
as to why it happened, is a question that someone from OpenAI will be able to answer and unfortunately not us.
got ya thanks, I feel better now - ha.
great!
If I already have 40 prompt credits and I buy the $15 pack does it stack?
yes
yes it does and the free credits last for 30 days while the purchased once will last for one year.
did you ever find out why? I've been wondering the same thing 🤔
You generated variations right? Your input image was cut off at the head so it isn't too surprising that the results are too
That original was generated cut-off, though
Why I cannot buy the credits, it gives me an error “unable to add a payment method”
is it ok if i put these images as my profile pic for gmail
Are you sure your card is working?
is that allowed
I am 100% sure
is it ok if i put these images as my profile pic fro my gmail i mean the one i generated
Yes!! You have full commercial rights to the pic, if you wanna add the watermark too, type !watermark in #dall-e-bot
It's !watermark 😉
Theb the best thing would be to contact help in this case...
Was about to correct haha sorry!
Probably an issue with your card, maybe it's getting flagged by your bank's systems, I would give them a ring just to make sure what's happening
it’s problem because I have Ukrainian bank? Because I’m sure that my card is working, I even payed yesterday
can i use harry potter in my prompty
cause hes technically a celebrity
or a public figure
Can we change the aspect ratio of images?
Negative, not as of now!
oh alrighty!
In my experiments I like to prompt Dall-e with queries in which I tend to explore the subjectivity of the algorithm. Is there anyone else who is exploring this, and what are the findings you have gotten so far?
what's the line for "public figures" with the content policy? do characters from tv shows count? what about characters that don't have their face shown and aren't really tied to a specific actor/actress?
If it’s a real person, or a real face, it’s against the content policy.
How far does that policy go? Does yoda's face count as a real face?
is it possible to buy credits / gift-cards convertible into credits for someone else, who also has access?
Just paypal them some money
Yoda is not a real person, so no.
hi, i'm getting the error "It looks like this request may not follow our content policy." when I give it an image of a car and enter "cut the front of the car off" I think it's triggering on the word cut? maybe I have not tried chop yet, any suggestions?
Ok
I want to do that by staying double-anonymous.
...
Cut off is probably the issue.
yar
Cut probably triggers the self harm policy
Hi, dalle won't let me buy more credits because my account is part of an organization. Is there a way to buy more credits for my own personal use without having to go through my lab?
Hi, i can't seem to find the answer in the faq: is it still necessary to include dalle's signature in the pictures i use? The content policy only mention that i should disclose the role of the ai and give credit to dalle.
The reason I ask is that I had dalle generate a favicon I'd like to use (also I'd like to cut off part of the image)
Can we sell the pictures that we generated on DALL-E? Like on shutter stock?
Hi! For your case, you'll need to contact support@openai.com as we don't have access to OpenAI's systems here.
Do you keep your access to Dalle once you get it?
Here's a statement from DALL•E's Product Manager: #community-help message
Do read the terms if you're still unsure! Found below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Pls refer to my answer on top of this reply.
Yep! Make sure to read the content policy to avoid account termination 🙂
Alright thank you!
Thanks a lot for your thorough answers!
Do the monthly points add up? ie if I didn't create for two months, would I have 30 points, or does it reset
It resets
Hey everyone! Just started with Dall-e this morning, and have a couple of questions if anyone can answer them. First off, can I assume there's no way to change the aspect ration of the images it produces? They will always be square (unless I edit them later after the fact)?
That’s correct
cool
second question, when you create a variation of an existing image, is there any control on how close or how divergent the variations are?
do the 100 gifted tokens and any bought also reset or just the monthly 15
The paid tokens (including the 100 gifted) reset in 12 months
ok ty
third question, other than the edit feature, is there any way to combine a variatins image with a text prompt. Like "Make me a robot made of triangles that looks similar to my uploaded image"?
How exactly does the composite feature in #dall-e-bot work? I don’t feel like I understand how to use it correctly
Not currently
thanks sushi
No, but I’ve found that oftentimes, they are arranged in order of how different they are from the original.
would be fantastic to get a control, depending on whether I want to go more experimental or more slight variations
I'll add it to the suggestions channel
thanks for the help!
if we use a de2 generated image as a pfp, do we have to disclose in our bio that it was de generated? or when asked
I believe there’s nothing against it in the T&C.
is payment possible with paypal?
how can I appeal a policy violation warning?
are you allowed to mention guns or weapons at all when generating designs?
dont think so
damn
I've seen examples of people saying dalle can extend the crop/image like the cosmopolitan magazine cover but I can't seem to find the tool on the website, can anyone point me to what I'm missing?
can i use darth vaar in my prompt
or harry potter any other fictional charactersr
wow you are typing for a very long time\
I tried my best at reading the rules and the licensing structure. Still have some questions
- Can we further modify/photoshop/alter the generated result images?
- Are we allowed to crop/remove/alter the metadata rainbow bar at the bottom right (I ask because I have seen others posting/selling works stating it was created here but there is no bar.
- Sometimes Commercial Rights doesn't mean broadcast rights i forget the official term, but on Fiverr there were a few types of rights and to my understanding don't overlap in the US you need to get both at times.
- Separate from those types of questions, is there a guidebook or prompt faq for modification of input? For Example on VQGAN+CLIP you would do the following, is this similar? the same with DALL E 2 ? ( a red hot air balloon:80 | neon glow:10 | illustration: 10 )
Thank you again for the privledge to be here, I waited four months to tinker around with it and will be a paying customer. Previously used VQGAN+CLIP on Google Colab.
Can people in my organization access DALL-E 2 and use my credits if they don't have an invite themselves?
hi, is there any way to not share your full name when sharing images? I prefer to keep my anonymity but would still like to use the share links 😅
same
you have to pay for more credits if you want more than 50 per month
Are we going to be able to pay with PayPal? I live in the Netherlands and credit cards aren't really a thing here. At the moment I have no way to buy extra credits.
Cash/debit-only? That's something I didn't know 🙂
Yes, we really don't use credit cards here. For online purchases we use PayPal or IDEAL.
WISE! Credit debt is ALL the rage in NA 😛
i saw the website uses stripe so it shouldn't be that hard to just accept paypal
unless they are keeping other information that credit cards provides and paypal doesn't
haven't worked with the stripe api much so im not sure
nvm stripe doesn't allow for paypal
that's huge bummer
are we allowed to share dalle2 access with others?
Some people have been sharing their access with their family members! :) Be sure to read the terms related to that at #old-server-rules though
How many policy violations before you get automatically banned? I sometimes forget to check my saved prompts for filtered words before generating images and i get policy violations, it happened twice already
Also is there a way to reset the total number of policy violations?
Q: I just got a content warning! What does that mean? Will I get banned?
• Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
• We chose to err towards our classifiers/block list being aggressive to start, but have been constantly iterating on them as we learn more. Please share prompts you don't think should've been banned in suggestions so that we can improve.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
I can only informally suggest, going by the experience of others, that Revolut apparently works well. I can't vouch for the company myself, but it apparently works, at least.
So it rather kinda judges rather than keeping an exact track of the warnings you get?
Probably by "inspecting element."
A lot of flags repeatedly may be an autoban, but human moderation is definitely involved.
Alright, thanks 👍
Informally the answer to the first three questions is yes. OpenAI are really happy for you to take the images and run with them. Keep in mind much legality around AI art is untested, and the ToS does still say you must disclose the role of AI in producing the art somehow.
As for 4, I don't believe DALL·E can parse weightings like this. Check out effective promts in our gallery channels to get a sense of how it works.
thank you, yes i also found that awesome like 80+ page guide and saved the PDF to help with prompting, it seems it just uses commas.
and yes I will still state it is made by DallE2 Just didnt know how it would look on a storefront page to have a bunch of those colored bars etc.
Are there any rules as to what celebrities are ok to use in prompts? I’ve noticed that some are fine and others will automatically generate the ”breaking guidelines” pop up. Seems pretty random to me at this point (pop singer A is ok but pop singer B isn’t)
The filter will not prevent everything, and you're still accountable for what you try to create. The Content Policy is clear, and breaching it can lead to your account being revoked.
Yeah ok. I’m not looking to make any creepy deepfake stuff. I use Dall-E for art references and it’s just easier to put celebrity x than to specify every feature I want.
I understand, but OpenAI can't see your intent if your account ever comes under scrutiny. The rules are unambiguous on this one.
Does the 15 credits per month add to your total or does it only refill to 15 if you have less than 15 credits?
There's no accumulation, you have 15 each month whether you use them or not.
Ah alrighty, but I started off with more credits than 15. Does that mean my total gets set to 15 when the next month starts or does it just not change my total?
All beta testers were given 50 extra for participation I think 😊
You can check your credits and when they expire in your labs.openai.com profile.
Can you change the resolution and aspect ratio?
You can't, however, you can expand on images. To get a better grasp of the whole topic have a look at this https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y, starting page 55 should be of interest to you :)
I really recommend to have a look at the whole promp-book itself though, and also at #top-tips-and-tricks, you'll find a lot of helpful information there
Everyone gets 50 credits during their first month. Those 50 will reset to 15 next month, even if you never use them, so it's best to use them all
How do people do uncrops of realistic faces that were generated by Dall-E? If you try uploading a padded photo Dall-E generated, does it let you since it knows it was a Dall-E image or something?
is there a way to use paypal as a payment method?
I don't believe at the moment, though it is a highly requested feature
can i use DALLE for commercial use?
and can i take the watermark off for certain uses for example an album cover
Same country, same issue bud 🍻
What makes Midjourney bit different than DALLE? Midjourney outputs look so much better but DALLE understands the context way better
Wait that makes little sense. Why would they delete your paid credits?
or is it just those 50 (which still don't make sense imo)
Refer to this #community-help message and this #openai-chatter message :)
100% yes, have a look at https://labs.openai.com/policies/terms and https://labs.openai.com/policies/content-policy though, simply to ensure that you stay on the legal side of things.
thanks so much
Are there plans to amend the pricing model as the beta progresses?
Currently, the price is far too high given the amount of tries it can sometimes take to get the prompt just right. Maybe if there were some way of getting partial credits refunded for bad/inaccurate results?
Or credits are only spent whenever the full res is saved/downloaded? I'd gladly pay more than the current price if it meant I could try out 3 or 4 prompts before settling on an image to download.
Maybe that's too much of a "suggestion" but. the question is mostly the first part
Hello! I would love to save images generated by Dall-E to "my collection" however I cant seem to find the button which allows me to do so (when clicking the 3 dots it only gives me the option to download, generate variations, edit image and report). Does anyone know how to save an image to "my collection"? Thank you!
Would the AI be able to understand "in the style of Studio Ghibli" or imitate the style of a specific contemporary artist?
Is there any way to set a nickname or otherwise hide my account name when sharing? I wouldn't like my full name being shared with my outputs.
studio ghibli? probably. A specific contemporary artist might be more difficult depending on how well-known they are. If it doesn't get it close enough after a try or two, try describing the style in detail, rather than just the person. What about the style do you want it to replicate? put those qualities after the prompt with a bracket like this |
not sure if that does anything, but it's common and seems to work
you can get pretty specific it seems
I tried it with George Caleb Bingham but didn't seem to get anything specifically resembling his paintings - is there a way I can insert a painting of his and have ONLY THE STYLE (not the content) carry over to a prompt?
as far as I'm aware, not yet. there is a trick to it in the "top tips and tricks" channel that might be worth a try, it's the most recent post in there as of right now
other AIs can do that though, it's called style transfer
it doesn't always work very well but it can
How about "In the style of Fur Traders Descending the Missouri"?
the specific painting
apologies for asking, the DALL E version we have acces to during testing, is it DALL E or DALL E 2?
I lost half a day of images, they are not in Recent
can I pick multiple continent roles
Are all images automatically saved to my account so i can access them later? I'm on mobile browser Andi can't find any images I've been generating
Unfortunately not. But if you access labs.openai.com on a desktop browser you should be able to see your previous 50 sets in the history sidebar, regardless of where they were created.
So i can view and save them from desktop? (They're not lost?)
Prior 50 sets, should be all good. Just don't accidentally hit the 'new' button 💀
How I upload a picture and add words like style to it? I must always erase a part.
Per example, I upload a realistic photo, I wand DALL-E to give oil painted results back
Not a function of this version. It has been mentioned in #suggestions but you can contribute your opinion also if it's important to you
When I logged in for the very first time DALL-E told me cuz my Number was used for another account I wouldnt get free Credits. Does that mean the monthly refill or like initial free credits ontop...
Can you say more about the types of subjectivity that interest you? There has been quite a lot of bias testing discussion.
Hey, just got acceptet and wanted to ask if the credits stack, lets say if I only use 30 credits this month and I get 15 free credits next month do I have then 35?
No, the free credits do not carry over so you might as well use them. The ones you pay for carry over though.
Oh alright thanks
Does anyone know if there's a way to get Dall-E to draw a photo that you take in a specific style (ik that it can create variations of a photo but you don't get to choose the style there), also the edit mode seems to only be for erasing parts of photos
https://help.openai.com/en/articles/6399305-how-dall-e-credits-work
How many free credits do I get?
You get 50 free credits your first month.
15 free credits will replenish every month after that, on the same day of the month.
For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
Free credits don’t roll over to the next month, so they expire a month after they were granted – but you’ll get 15 new free credits.
This would be a really cool feature
oh ok, I thought it possibly already existed, maybe I should move it to suggestions then
I'm new here too, if it exists I haven't come across it.
I can answer your question, need a little more detail
Are you talking about taking an existing photo you took irl and changing the style with DALL E?
Or generating an image of a photo in a specific style?
The first you cant do, the second you can
yeah, like in theory if I took a landscape photo, could I get it in the style of starry night or something
Well, hypothetically you could erase the background in edit mode, and add "the sky is in the style of starry night" and run the prompt
But you cant take an entire photo and change the style
you can do that with wombo
oh ok, thanks
Mhm
@atomic sable You might want to take a look at #gpt-models specifically the DALL E Prompt Book
oh that looks pretty cool, thanks
Did they reduce the number of generations in recents even further? I didn't download a bunch today and now I guess they are lost to the ravages of time.
Try your browser history
I’m interested in using Dall-e 2 as a teaching tool for art students to inform them of stylistic differences between artists, composition rules., etc. Are there any resources currently available about using Dall-e 2 as an educational tool?
How do I include actual text in an image, the output I'm getting doesn't match
My input is
Boat with the word "fishgal" on the back
No resources afaik but I'd love to chat more in #ai-discussions when I get a chance. I've used it in the classroom already!
DALL·E sucks at text, way too complex for an aesthetic model. Future / research-only models with exponentially more data have started to get better (e.g. Google's Parti and Imagen)
Why is the word " fursona" against content policy?
It's been discussed quite publicly already. Without too much retreading of worn ground, it's a combination of both the data that tended to be tagged with that word in the training phase, and the context with which early alpha users applied it. It was largely sexual in both cases afaik
Similar words like "anthropomorphic" or "humanoid" are fine and effective.
have a link to some of the public discussion?
Not that I care to share
not very helpful, thank you
I've personally been vilified for contributing to the discussion, in the interst of helping others, so forgive me for not wanting to keep stoking the fire
This is still something they might find a workaround for in the future. Anytime I see questions like this I think about Microsoft’s Tay bot. Have you heard about it?
Yes, I have heard about what happened to Tay
How do we own what we create?
Is it possible to generate/download images with higher than 1024*1024 resolution?
Is the model case-sensitive? There doesn't seem to be a way to a/b test it since every generation appears to be a unique result.
The model is not case sensitive as confirmed before!
Not at the moment no...
You could use another tool to upscale though

Single-Image Super-Resolution for Anime-Style Art using Deep Convolutional Neural Networks. And it supports photo.
Nero AI is pretty good to
This doesn't seem like it belong in this channel, If you wanna discuss about it I suggest you use #openai-chatter!
Disclaimer: I am not a volunteer or an openAi employee, just helping out.
actually it does! Nero AI is another upscaler they recommend 🙂
Are you allowed to crop out the few colored pixels at the bottom right of the generated image when using it somewhere, provided that Dall-E/OpenAI is still credited appropriately?
is there a way to upload an image, and ask dalle to remake that image in a different style
?
As per DALL∙E's product manager: #community-help message
Thanks!
Hello! Not yet, hopefully soon!
kk thanks
Technical, and social subjectivity. I like to look explore how the AI interpret queries. Currently I am exploring how it creates various shadows of people to which I attain a single characteristic that might say something about the person's identity. But not so much about its shadow. At least... So I thought
Is it possible to generate images with transparent background?
Not at this time
Is there an announced ETA on API/programmatic access to Dalle?
Not yet. It should be in announcements when they have one
can't wait for that
My prompt is stuff at 99% loading any way to fix?
haha me too, seems like there might be a bit of traffic since a lot of people are getting it?
Rip my one credit
This might be it! OpenAI's goal is to invite 1 million users. Heavy traffic is expected. 🙂
AFAIK it will still generate. Check your sidebar in a few moments
Is it against TOS to have two Dalle accounts? Also what does AFAIK mean
Sorry, I can't answer your question, I advise you to read the Terms found below
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
It just said this request was canceled but still took my credit rip lol
tuff, rip to your credit 🙏
That shouldn't happen 😦
18 left lol
I’m not worried about it but mildly annoyed lol. Plus I forget what I was trying to make haha
Quick question, if OpenAI ever releases any form of programmatic access to Dall-E, will they keep the models running on their servers, meaning that we will have to endure the traffic? or will we be able to run the model (like on dalle-mini) ourselves?
No heads up from OpenAI at the moment, stay tuned!
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.
Let’s not do this please, thanks.
How do I know if my account has been suspended? My prompts are coming up cancelled
DallE is down now
rip
Hello everyone. My first day on dall-e. Somebody know how to change format to photos? eg 3:3 16:9 and so on? Also if there's some promt to get better resolution for realistic photos?
2048*2048; when
Sad to hear it's down, glad my request for Kanye as a table was not the reason it wasn't working 
Good time for server maintenance I guess. 🙂
is there a specific channel for reporting openai errors in code?
What do you mean by "in code"?
im running to some problems using openai in python
Question: can you accidentally "overload" the AI with a prompt?
It's currently down for everyone - see #announcements
Ooo
thank you
sorry x.x
Hi, do we have banned word list? I wonder if I can just search it before making a prompt
Anyone else having generations not generate or get hung and then cancel? I tried signing out and resigning in, but still not generating... It looks like its trying to but it just keeps on the loading screen.
See #announcements
oh yes I see
any ideas as to when dalle will be back up?
now
hey is it possible to share your entire collection?
currently no but we'd love to build this in the future! can you tell me more about your current use case / reason for wanting to do so?
thanks!
I just want to share it with a friend who also has dalle2 access
and is it that you want to share a bunch of different generations at once, or you'd like your friend to be able to edit / make variations out of htem?
I want to share a bunch of generations at once
Can I sell the image as NFT?
You mean this policy? https://labs.openai.com/policies/content-policy
Hi All,
Is DALLE currently at a level where abstract human concepts can be visualised ?
I tried various thematic drawings - like 'cosmic serenity masking serenity' and other more elaborate concepts which tie human level abstraction on concepts, and each time the image creation failed.
What is going on ? Could somebody throw the light here.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
^ These policies.
I believe I might have missed out on the answer for this but for those who were gifted 100 credits for being in an earlier Beta, will those credits expire after a month? Or would it be just the additional 50 starting credits
With DALL∙E, you'll need to be more specific with what you prompt. DALL∙E will only try to generate what is given in the prompt. There are resources available in #gpt-models.
The 100 credits I believe are counted as paid credits and will expire in 12 months.
Got it, are credits specifically queued in any way where the ones that expire sooner are used up first? Such as the extra 15 we get in the next month or so
Thanks new here. Just got Dalle access today. Will read through the prompt guides.
The free monthly credits will always be used before the paid credits
Sweet
Welcome!
Is the real face ban going to be permanent? It’s frustrating because it takes a lot of possibility out.
Yes you're spot on! The 100 extra credits for early testers are count as paid, it has been confirmed several times by now :)
Yes, have a look at both https://labs.openai.com/policies/terms & https://labs.openai.com/policies/content-policy :)
If you're wondering whether you can get rid of the watermark, take a look at this. #community-help message
How to get a 4K image resolution from DALLE generation?
Upscale it externally or stitch multiple generations together to create a composite image
The resolution will always be the same with Dall•E, you can however "combine" them. Have a look at #inpaint-outpaint and
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y starting page 55 :)
Any tool suggestions to make it possible?
Okay i ll check
Please don’t flood this channel with images, makes it hard to answer questions.
How many credits do you get for free after a month?
15
I think this is just because DALL-E was struggling earlier with load (quite rare)! It should at least attempt to create anything you throw at it - whether it's a good representation or not is up to you to decide!
Don't do it. One immediate ramification is you can lose free credits for both accounts.
USD everywhere
i just got access this morning, is there still a way to do the zoom out?
any servers i can join to get help using openai's api?
'Losing free credits' means that you will lose your free credits.
So imagine the 15 free credits in your account, those will be gone ||poof||
and you dont get any more?
duh
h
That I'm not sure of, however, simply follow the https://labs.openai.com/policies/content-policy and https://labs.openai.com/policies/terms and you're all good :)
You may try support@openai.com
Also, simply don't do anything that is against the policies even for "experimental evidence". DALL·E is a great tool and should be used as such
any servers i can join to get help using openai's text generating api?
Let's keep the channel for questions only. To discuss further, kindly transfer to #openai-chatter or #off-topic.
Anyone figured out how to have Dall-e convert an existing photo into another style. like from a regular photo into stained glass?
Not sure about that, sorry!
Currently, that is not possible.
im having some trouble with it
how do i get help?
There's a live chat option in beta.openai.com but it's current offline as it's the weekend. I suggest contacting support@openai.com or wait for the live chat.
Do paid for credits rollover to the next month?
Paid for credits will last for 12 months :)
<3 thanks
Do the initial 50 credits expire?
Hello, so I got access to Dall-E 2 yesterday and as I got in I got a message saying that my phone is used on a different account, hence I would not be getting monthly 15 credits, instead I'd only have the starting 50. The second account that I have the same phone attached to is a beta.openai acc. I believe it's a developer account, which I made without knowing much about it thinking it was the way to get access to Dall-E 2). I'm wondering if there's a way to either delete the beta.openai account or to somehow change the phone number attached to it
dunno if I should write to support instead
After a year. Not that they’ll last that long though lol
Is there any possibility of getting an exception to the content policy restrictions on weapons? I'm looking to generate images for a card based rpg
Is there way to uncrop photos of dalle created people? I tried but it gave a warning, telling me photos of faces are not allowed. the photo was ai generated though
If I were you the I would simply message the support, that's the best help you can recieve.
support@openai.com, or you wait for the live chat on Monday :)
No exceptions as of now. Always keep the https://labs.openai.com/policies/content-policy and https://labs.openai.com/policies/terms in mind.
Will try my luck with support, thanks
You're welcome! :)
They last a year
Thanks!
No, once generated Dall·E will recognize people's faces as real ones, and will therefore treat them as such.
You could try creating a new prompt using words like "wide shot, full shot" etc. and simply experiment. :)
Q: How DALL·E Credits Work
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire a month after they were granted – but you’ll get 15 new free credits.
thanks
Seems a bit picky as to what it will add, Eiffle tower, citadel or St Basil’s cathedral all work, but zombies, aliens, grim reaper only fill with content that you’d get from photoshop content aware fill. Ghost of death is banned. Greying out the faces seems to work to stay within the no real faces limitation, easy enough to restore later. In any event I’d like to get something unexpected standing behind me. Burned most of my first 50 with not much success.
I've been thinking of expanding these two ArtBreeder portraits into a full character. Are these too photorealistic to upload?
Also, how touchy is the filter around military stuff?
I've had "a soldier floating on a hang glider, carrying a grenade and a makarov pistol" rejected, was the pistol or the soldier the prohibited part?
Yes, those will get flagged. The filter is very sensitive to anything related to violence. You may refer to the Content Policy found below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Will DALL E in future be less strict towards certain prompts like "sword"?
Some have generated those without mentioning swords. Sometimes it will get flagged, sometimes it wont.
Fair enough
Hopefully! Read the reason on the 2nd point below:
Q: I just got a content warning! What does that mean? Will I get banned?
• Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
• We chose to err towards our classifiers/block list being aggressive to start, but have been constantly iterating on them as we learn more. Please share prompts you don't think should've been banned in suggestions so that we can improve.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
ic so it's probably subject to change then. It's weird because sometimes it goes trough and it generated very nice sword design but now it refused to go trough
DALL∙E not only uses a list to block words, it also has contextual understanding. Thus the variance of whether it allows it or not.
thx for clarification
I searched death gods and got this prompt
has anyone from India tried to purchase credits? Which cards work?
Is it permitted to use non-contemporary real people in prompts?
Like Marilyn Monroe or David Bowie
@heavy sigil Questions go here, yes, theres an image history. But it only goes back so far
Look for this button on the right side
my bad for posting in the wrong channel!! ty for the help
np
How do you do uncropping? I only see how you can give prompts. Do you have to be on desktop to do that?
Do the 50 free credits you get at the start roll over to the next month? Cause I just got access and it's the last day of the month.
My understanding is that credits reset on the day of the month you got access. For example mine will reset August 20th
Hello guys. Im super interested in Jukebox, which i need for a project.
Is it active yet?
Thanks! Email already sent....how does the Livechat work?
Soo, when I report an image for being inadequate and they do find it wrong, will they refund the token used for it?
Because I think that'd be fair
Not that I’ve seen. You can contact openai and ask there though
Simply go to https://help.openai.com/en/, you will find a button in the bottom right corner :)
awesome sauce!
Yes! After one month...
Hello! Just wondering if we are still required to keep the Dall-E watermark when posting artworks or when using them commercially? Would disclosing the role of Dall-E in the post's caption be enough or should the watermark be left at all times?
Will the 100 credits given to the early access people expire too after a month?
Feel free to remove! But please do not deceive people that the art is not made by an AI.
hello!
how to make “uncrop” technique?
passing a picture that is unfinished and let it append/addition it
“edit” require prompt and i cant work it out to pass an unfinished picture
The 100 credits expire after 12 months.
ill let you take these Major!
You may check the #dall-e-bot for the pad command. Examples are available in #inpaint-outpaint. Also check resources at #gpt-models
Thanks for the clarification! I will still disclose that the work is made by Dall-E in the captions but I am planning on using the generated art as a base and add elements along the way.
Thanks for the assist!
thank you!
what to write in prompt if i just want dalle to go with the flow?
The same prompt as your first.
np
thank you
will hex codes for colour work in dalle? eg. #ff0000 for red https://www.color-hex.com/color/ff0000
#ff0000 color hex Red1, #F00 color chart,rgb,hsl,hsv color number values, html css color codes and html code samples.
how can i share images on instagram?
can i do it right away or is there some other policy?
Feel free to do it right away!
Post away! The picture is yours, have a look at https://labs.openai.com/policies/content-policy & https://labs.openai.com/policies/terms
thank you
Sometimes it does, sometimes it does not. Check this trick out: #❌┃tips-n-tricks message
😊😊
That is technically wrong. @crimson moon, as per the current terms, OpenAI owns the images but you have the exclusive rights to use it anywhere. Please have a look and read the terms, found below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
This is against TOS fyi
Indeed. Account termination is possible.
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.
You're right! Allthough we "may sell the rights to the Generations you create" etc. "OpenAI still owns all Generations (including Generations with Uploads but not the Uploads themselves)". :) Sorry @crimson moon!
no problems 😅😅
OpenAI will not sell the rights. Kindly re-read the terms.
OpenAI grants you the exclusive rights to reproduce and display such Generations and will not resell Generations that you have created, or assert any copyright in such Generations against you or your end users, all provided that you comply with these terms and our Content Policy.
I'm guessing uploading album covers for decropping is against the TOS?
Technically, yes.
? I never said that 😅
By "we" I'm reffering to us, the users. OpenAI simply owns all Generations etc. as stated in the policy
Or am I wrong? We, the users, have the exclusive rights to use it anywhere, also commercially (as long as we stay true to the policy).
OpenAI owns the images, will however not sell the right etc.
The statement that I re-sent via screenshot is wrong. The one you've sent now is correct-ish. I would just avoid answering directly about the terms just to be safe.
What was the prompt for your profile pic? It’s a great gem
Fair fair :)
Gen***
“32x32 pixel art of an old grumpy ship captain, white beard, white cap, black coat, yellow background”
Had to look it up!
I haven’t tried explicit dimensions like that. Do they usually work best as exponents of 2?
Is it okay to use trademarked names in prompts? Like "Hotline Miami" or "Saul Goodman"?
Yeah but you might find limited success depending on how familiar the model is with them
Tbh, I'm not quite sure about that, but leaning on yes.
As aeholbrook said, you'll be having limited success, that is because of the strict filter. But fictional characters are fine.
I have a question about the "free credits", it said if you get the credits on the end of the month you'll get the next ones on the 28th, but it also said the free ones don't roll over. I just got my 50 free credits today, are all 50 going to expire after today so I need to use them all now? Or do they expire on the 28th?
When did you get in?
today
All of your free credits will expire and refill every 28th.
Q: How DALL·E Credits Work
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire a month after they were granted – but you’ll get 15 new free credits.
thank you
Hi, I just got in. So Discord says Dall-E 2, but the url I was given goes to the Dall-E generator. Is Dall-E and Dall-E 2 used interchangeably?
Just to make sure, you can check it for yourself. Click your icon on the top right in the website and press your credits.
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
@timber quiver ^
Hey actually I have a question — if you buy credits on top of your free credits, will you still lose 50 credits at the end of the month, or are the first 50 credits you use the free ones?
Just curious
Thanks!!
The 50 credits are free so it will expire after a month. The free credits are going to get used first before the paid credits.
Got it. Thanks!
If we set our Discord pfps as a DALLE generation, how do we adhere to the Content Policy (specifically "Disclose the role of AI").
The watermark acts as a way to do that but it's cut off by Discord.
i can't pay with any bank account i have. is it possible to add paypal as a payment method?
There's a bot in #dall-e-bot with the !watermark command
Not as of this moment, sorry! Others have had successes with the app Revolut though!
That makes life so much easier + I can now use the bot's status for something else. Thanks so much
You don't need to do this btw
Enjoy!! Do share your work with us in the relative channels 🙂
It's just an optional thing that personally I thought was cute 😛
It is cute
Oh, it isn't necessary?
The terms are so confusing