#community-help
1 messages · Page 4 of 1
The only method to change your password as of now (according to official documentation) is by using the 'forgot password' option!
Do you need step by step assistance? if so let me know! ill be happy to help 🙂
Ok I'll try that
If you have a UK postcode, don't input any spaces whatsoever in the card prompt's postcode entry.
Yes this is a code generated from a small handheld device you have. Do you have one?
similar to two factor authentication on google through the google authenticator app.
CVV 😉
I don't but this is a company card I have been given at work
so someone else might
thank you!
Yep! Its called Chip Authentication Program
its a very small device!
Is that a card issued by HSBC?
not sure
CVC 😉
Why would i want to watermark a pfp i already have?
Ok thanks. I guess the other question about giving credit that a dalle image was used as a visual reference is my main concern. Want to follow the policies, but nothing is on the website 😅 as an artist, reference are huge part and dalle seems awesome for that
Card verification value or Card Verification code 😂
Assuming you set it before you found the functionality maybe? idk you'll have to ask that from the creator of the bot! 😅
Dude thats cap
I'm so funny
Yes, basically
Are you having trouble finding the policy? If thats what you meant, you can find it here :https://labs.openai.com/policies/content-policy
Nope, you can also use it to generate a curved watermark along the edge of your profile picture...
Oh ok why would i want that?
CHILL
It's your choice, can be a cute little feature imo...
Ok
Oh, ive read it. I do t think it answers the question on if one needs to credit if used it as a reference (like how artists use multiple images/photos as references to create art) 🙂
oh you meant specifically this case, gotcha!
Q:- If Dall E2 will be available for all people and if its based on subscription then will beta users will got free access ? or their access will be removed ?
would requesting depictions of historical battles flag the content policy? does the level of detail make a difference? for example "oil painting of the battle of gettysburg"
Once again, nothing regarding the post beta days of DE2 have been announced as of now!
@rugged wing
ok
Right, can you send a screenshot?
yes
Send a screenshot of your DALL∙E page please!
ho, i got it.
i clicked on an original of a variation
that does not bring the option to save.
But i got it now
Yes, as it's the original. If you want to save the original, head back the previous generation and save it from there
Got it, thx!. another question, is there a way to delete from the recent list
No, but there's a way to clear it. Go to the top of the recent list and press the New button.
This will however delete everything :)
oop
I saw that, but i only want to delete failed creations
There's no way to remove individual entries in the history sidebar at this time
That's not possible right now
would be a nice if that can be added
#suggestions :)
Ok thx for now. Just bought 115 extra, playtime 🙂
Play time indeed. Check the guides at #gpt-models, they're a lot of help!
will do
Doesn't seem to be working for me? Am I not doing something right?
What do you mean?
Just got a warning for this image, is it considered sexual?
You got a message trying to upload it or generate it?
Uploaded it as a prompt for variations
Could be because it's a copyrighted image?
How would it know that it's copyrighted tho?
Hi there. Ive read your terms of use agreement but I'm wondering if you can clarify something. If I were to make images into NFTS but not for commercial purposes, would that be allowed? I.e will not sell them for money.
Also, if I were to put an image on a t-shirt with full attribution to Open AI, would I be able to sell it for commercial purposes?
Same way youtube does? or shazam knows what song is playing?
You are comparing it to audio 🤷♂️
no, youtube does it also visually 🤷♂️
Yes you can, as long as you don't deceive anyone into believing it was not made by AI...
I advise you to not speak on behalf of OpenAI. Instead, please redirect other users to the terms.
- For NFT purposes — informally, the product manager of DALL∙E has mentioned you are allowed to mint it but no formal writing specifically about NFTs in the terms as of now. (#community-help message)
- For commercial purposes; as per the Terms, point 6, as long as the generations follows the Content Policy, then you'll be keeping the exclusive rights.
- Do note that the terms may change anytime as we're still in beta.
- Save the links below for your reference:
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/
Perfect appreciate the prompt reply sir
Hi, can credits also be bought via paypal or alternate payment methods other than credit card somehow? If, not will more payment methods be added in the near future? cheers
The only current method is credit cards (maype prepaids as well?) tho I cannot speak for their plans for the future
thanks, hope well see different ones soon
Can you direct me to a direct link with directions, I have been using another AI here on discord and there I have to add / imagine and prompt and then write my prompt
Is horror,terror and making scary/odd content allowed my completions dont generate pop ups or anything but i want to be sure pre sharing
/imagine prompt:
I did not find any directions on that site. Can you please direct me to directions on how to use prompts with this AI
I was able to use it the first day I was on so I don't understand why it's not working now
That’s where you issue the prompts, not in discord.
Is it giving you an error?
I think you were trying the method related to a different AI there... just go to https://labs.openai.com and enter your prompt in the big bar at the center, that's pretty much it.
If you were logged out and it's redirecting to you another page, use https://labs.openai.com/auth/login to log back in
Ok, I was, thank you I got it to work there
it is allowed, but stir away from those mentioned first here
In the content policy there is a line, telling us not to create celebrities. But even in main page there is a Marilyn Monroe's prompt and picture. What gives? What is the line here
Hi I have an account for my org, however, a member who signed up using Microsoft federated login, cannot access Dall-e. Can you please advise how members from my org can access Dall-e? Thanks
Marilyn Monroe was before the no-celebrity rule.
Ah just checked, it's a Bauhaus-style painting of her. 🤔
Policy's pretty clear though: Answer here: #community-help message
Why it is not possible to buy credits in South Korea? Is there any plan for allowing it?
Hi ameer, I'm afraid we do not have the answers for you here in Discord. Please contact support@openai.com for this.
I'm afraid I also have to direct you to support@openai.com for this. Or alternatively, go to this link: https://labs.openai.com/account and there should be a "Help" option at the bottom.
When does one's credits refill
Once a month, you can view the exact date on your profile
Click on your profile picture in the top right corner of labs.openai.com
And click on your name
Hi Nasazh! Please disregard my previous reply. I stand corrected: #community-help message
Hi everyone, are you limited in history? in terms of how far back you can go into the images you have generated?
The sidebar keeps your last 50 prompts. Beyond that you'll need to save the /e/ addresses or have them in your browser history. At the moment OpenAI are not deleting old images from their servers.
Good to know, thank you !
Are you limited in terms of how many images you can save to your collection?
10,000
can i buy more credits via Paypal?
Paypal is currently not supported.
only via credit card?
Credit or Debit card supported by the payment provider Stripe.
I work as a social media manager, the policies are a bit fuzzy to me. Am I able to use Dall-e to generate content for my clients (charities) pages/posts, so long as they are marked as AI generated?
Yes! You have full commercial rights over your creations as long as you do not deceive anyone into believing it was not created by AI. Furthermore, you are allowed to remove the watermark as per the unofficial 'Green light' given by Joanne!
Thanks
Thanks! Unfortunately many of the runs weren't on my machine but on another (but logged in to the same account). Is it possible to get a list of all URLs associated with my account?
Not right now, unless you have them yourself :/
(as far as I'm aware) :)
When generating variations of an existing image, is there any way to enter a prompt for the image? When I try to generate variations of an already-generated image it keeps losing a lot of the styling stuff from the prompt that I worked hard on 😦
You guys should just have a bot that sends “no PayPal” every 30 mins
I don't see my previous generations in my collection. "Your saved images will appear here", but it's not there. Where is it?
Well using ‘edit’ would but you’d have to keep some of the original
@viscid pine You have to actually save the image to your collection for it to show up there, by clicking the Save button when looking at an image*
Oh so the first verification pic doesn't count?
For the verification I believe it has you publish it to the public gallery (using the "Share" button) rather than save it to your own collection.
got it, didn't realize I use 5 credits
If you can still see your previous generations in the "Recent" tab on the right side of the page you can still go back to them and save/share/download
It should have the last ~50 or so prompts you entered
Thanks, I guess that works, but it would be cool if I could give a prompt while doing Generate Variations too 😄
When creating a variation from an uploaded image, is there a way to control the strength in difference? I would like to create very subtle variations.
There isn't 
where can i change my password?
Hi! Am I allowed to use the images for marketing/ branding purposes?
Yes! Commercial use is allowed!
The only way to change the password as of now is by using the forgot password option. Provided you used the direct sign in method and ofc not a sign in partner.
Thanks! 
How do I upload an image to use the bot to pad?
By using the small plus sign to the left of the message bar or by pasting it to the message bar itself from your clipboard!
Thanks. I'll give that a shot.
Is there a way to share an image link without revealing my full name?
Also, I still can't figure out how to download a watermark-free image, from which screen can I fetch the .webp?
you can't dl without watermark, but you can always crop it
Right click -> save as the image then upload it to imgur or something and share that
plus the shared one will have a watermark too
I just remove it in Photoshop if I wanna edit it further
I just mean to get around revealing your name
There's no way to give anyone in your organization on openai access to DALLE, correct?
Does Dall E allow selling of the art? Like as NFTs?
Because I've no idea what I'm doing, I have to ask: How do I know I'm using Dalle 1 or 2? My top left corner of the screen just says DALL E. I though I signed up for 2 but I just am not sure.
You're using 2 if you're in here
afaik openai doesn't offer dall-e 1 access directly on their site anymore, some 3rd party tools still use the dall-e 1 api but if your image got you verified to join this discord it was made in dall-e 2 🙂
Legit, thank you. There's nothing that says "2" on the labs so I started second guessing myself.
Take a look at this :)
#community-help message
was just reading it haha thanks
You may want to check out the current status of the nft ecosystem before putting your time and resources into minting any, it's deader than dead:
https://cointelegraph.com/news/nft-sales-down-90-since-market-peak
Where does generating images of live-action characters fall with regard to the rule Do not attempt to create images of public figures (including celebrities).?
Can I upload photos of myself if my face is blurred?
Do you mean generating live-action style images of characters that don't have an existing live-action portrayal?
That should be fine because it wouldn't give you the likeness of any specific real-world person/actor
I mean like generating an image of Luke Skywalker
Yes but you need to be pretty zealous with the blurring
Ah gotcha, you can generate images of Luke Skywalker but it's not going to be Mark Hamill as Luke Skywalker if that makes sense
(but also there's a good chance it's not trained on any starwars stuff, most existing franchises/IPs seem to be missing from its training data, which makes sense)
Hi, I'm confused about the policy that I can't upload faces - does that mean I can't even upload an image containing a full figure, if a face is present, no matter its size? So any full figure uploads need to be head facing away from us?
And another one - before I read the policy, I tried to generate Alexander Fleming, the discoverer of penicillin (sorry about that). And I was actually going to try a bunch of historical figures. Is that also not allowed, no matter how long the person has been dead?
Any kind of realistic face is currently disallowed without any exceptions, no matter who it is, if it's a real person it can't be generated in their likeness. (This is likely to be more relaxed in the future)
In the full-figure example, you can blur the face. Some people are doing that, then after finishing their generation with dall-e they use the tencent face restoration ai (https://arc.tencent.com/en/ai-demos/faceRestoration?) to generate a realistic face over the blurred face.
However keep in mind that even if you do this you are still subject to the content guidelines regarding your use of the final image
Any possibility for textual prompts to go along with image variations? I've been experimenting with it in Disco Diffusion, and it's very powerful. You can give it an image and prompt it with "Painting by Vincent van Gogh" and it will generate variations in the style of van Gogh. Of course, DD is iffy in quality with certain images, especially with the setup I made for generating variations, but that's a problem of it being an older model.
Is this something that will be considered for DALL-E?
ive reworded this one a few times but this is the original, i looked through the content policy but i dont really see much on how this is wrong, will this affect my account?
"Abduction" is definitely the problem here. In the context it's harmless but in most other contexts it represents a violent crime.
oh ok thanks! is this like a warning on my account now?
Not at the moment but it should be at least theoretically possible as far as I can tell; we're awaiting an update.
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.
great, thanks!
For my profile pic, I scaled the circular image so that the whale is positioned a little better, but it does slightly modify the appearance of the watermark. It's still clearly visible, just much flatter. Is this okay?
Very cool, and my experiments I've been doing on Disco Diffusion and Latent Diffusion show that it should be possible on most models that support CLIP's image and text prompts
Re: abducting/abduction, it seems that you can generate aliens abducting inanimate objects no problem. It's just when people are involved in the abductions that it gets filtered. Not sure about animals, haven't tried. This prompt included "a UFO abducting the eiffel tower"
(Might also have worked because I specified an inanimate object ('a UFO') as the thing performing the abduction)
interesting, nice observation
ive done like 300 generations on aliens im like obsessed lol
Have you been able to get it to generate stereotypical big-headed, big-eyed Greys? I'm assuming not because of potential conspiracy ties/deception but I really want to make goofy pictures of Greys wearing silly outfits :p
Honestly a lot of the time I have a hard time getting it to create other kinds of aliens 😅 I think "Grey" is the wrong route though because of the stronger visual connotations with the colour and the fact DALL·E is not case sensitive.
Ahhh "Classic Alien" is a great way to do it, cheers! When I was just using "Alien" or "Grey Alien" I was getting stuff like this
The "diner" setting might help as well because you're tapping in to Americana
will i get banned if i let my friend from different country use his credit card to get credits? i dont have credit card
I have had a few dreams about dalle 2 before I got access lol
And I mean actual dreams
Can I use licensed property like Snoopy or sports logos in my prompts? Also, will it work for sports arenas or ballparks?
You would be responsible for following proper copyright procedures for anything you publish featuring licensed property, I don't think dall-e will stop you from uploading images of them though
Hi, I was wondering if there is any risk of generated images to be too similar to the source images or if they are always changed a lot?
Like if I generate an image, what are the chances it's actually too similar to existing artwork
(as in directly taken from it, not coincidental)
In what way because if it’s a copyright issue there’s no way of concretely defining it as fair use is only a defence so it would be up to a court to decide if it constitutes fair use or infringement.
Nice pfp
Yeah I guess I mean could it infringe copyright without me knowing it? idk how to explain better i guess
Thanks haha
and yeah like I don't mean just a coincidence cause that can happen with any art, I mean can dalle fail to transform it enough
Eyes immediately flew towards pan flag xD
It depends it is quite hard to say, if you are from the US look at fair use laws and obviously other countries will have different copyright laws.
Hahaha
DallE is creating the art by itself, so if you get in trouble for it then DallE would also get in trouble. So I'm pretty sure they thought this through
huh, okay
Not a lawyer, just personal opinion
I mean more like.. lets say you generate a prompt and it just straight up takes an existing picture and fails to edit it. I know it's probably not possible for that extreme of an example though
At worst you will get a C&D
It can’t do that ever, it’s not collating or editing existing images it has learnt from a data set and is creating entirely new images. It is over a 99.999999% chance that that will never happen
Hello DALL - E team can you kindly confirm this. I know the images can be sold or used commercially . . But do we still need to attribute ? For example stating that it is generated by dall e. Thank you.
and if it somehow did I’m sure 99.99% of copyright holders would be completely understanding once you explained and would likely just ask you to take down the image if you had posted it
Not DALL•E team but yes you do, it says here: https://openai.com/api/policies/sharing-publication/
im planning to use this for art/product reference and inspiration so thats why i ask
Dude we said this in the other channel
Actually I don't think I've seen the DallE team here?
Oh. This is not like official question group? Much apologies !
I was looking through member list and it’s buried in their but some are here, their role just isn’t shown separately, but they have a different name colour
No it's fine. I came off as snarky
Especially if your using the generated images for inspiration and reference you will be 100% fine
anyways thanks for your answers everyone, that helped :)
Long day at work lol. Sorry
Is there any way to change the resolution of the images it generates? So rather than creating square images, you could create something A4 size?
I’d say the best work around to achieve this is to use https://editor.p5js.org/Pi_p5/full/qAqoieAhx and then inpainting to uncrop the image then crop the result to be A4. I can explain how to uncrop in more detail if you don’t know how to do that
A web editor for p5.js, a JavaScript library with the goal of making coding accessible to artists, designers, educators, and beginners.
I have a question! To what extent will Dall-E do fictional characters/worlds?
As in copyright characters or creating entirely new ones?
Copyrighted ones, like Wario or Mickey Mouse
does anyone here know if the ai's performance is at all related to what machine youre using? i was using my fiancee's saturday night and shes got a gaming rig and the art i was producing was significantly higher quality than my dell laptop
I think if you are making new images just using prompts it is fine, just if your uploading images, you have to own the right to use it. (Unless if it is public domain or something)
Probably some kind of confirmation bias, the generation is done entirely by OpenAI’s own servers so the machine used will make no difference at all.
it should not have an affect! maybe your prompts just got better

seems worse, im using exact same words but i just cant get anything even remotely as close to the first one, and it was gorgeous
(Hate to be this person but it’s effect not affect)
that's interesting, i'd love to see both generation sets
i get it, thank you
the computer you're using doesn't do anything to the quality, but the website is very unoptimized so your internet speed does matter in terms of how fast pages load
you only have one example from the first computer though
yeah i only saved that one at the time, i didnt realize id lose them as i made more
In my opinion some of the ones from the second computer are better anyway
i dunno, that first one just really gets me
and the generations based on the upload of that one dont do it justic
I’d say if you want generations like the first one try changing your prompt. Can you share the prompt you used?
sure, but you might laugh
oh nm, its not that one i used lol
high definition of 2 black holes colliding with planets destroyed around it and galaxies in the background, vibrant colors, , gravity
i threw gravity in there as curveball for it
I think adding more about the style (I’m not sure what it’s called) could definitely help. Try asking in #🌄┃general-art for the style or any artists with a style similar to it that you could potentially add to the prompt
ive delved deep deep into making psychedelic mushroom images lol
oh im sorry, i forgot this ws questions and not general chat
hehe it's okay
if you want to you can continue sharing these pictures in #art-and-illustrations they're gorgeous!
my generated images are not useful, lots of crop, not in finished form, please help to fix
You can use the ! pad function in #dall-e-bot to get a transparent border to re-upload to "edit" and expand.
Hi, I just got access a few days ago
I want some help with this: What problems do you think there are with the prompt "Portrait made with oil paint of a human figure of unrecognizable gender. Suffering has torn hope from her large eyes. Thousand yard stare"? When I try to generate images with it, I get a policy violation warning and I don't quite understand why.
remove suffering
possibly gender too
Ok, thanks
can i sell my art work in NFT or commercial use?
yes, but i think u have to give credit to openai

Hey Dalle Team,
Is there any official dalle API? I was thinking of displaying the images in my collection on my website
ofc I could go to network tab and get the endpoint details, but I am looking for something official
Can I see my history without entering a new prompt?
@brisk reef you should be able to by clicking on the little clock with a backwards circle arrow near the upper right of the screen a bit below your account picture within the dall-e site.
it seems to remember "recent" and if you click it again, it makes it "clear the session.
I don't see it right now but that might be because I'm on mobile. Will try again when I get home.
how might one save their generations from their collection into their photos folder on their phone? i can just take a screenshot from the website, but i'd prefer to save it to my phone as an image. is there any way to do this?
When looking at a single pic scroll down and there's a dl button
i believe that just saves the generation as a file though. i would still have to screenshot that file in my file folder on my phone if i wanted it in my photos. i was wondering if there was a way to save them directly to my photos
Oh right, you should then be able to move it. It is a lil clunky
ok, thanks. i'll check that out
Hello there, may I know where the rules and regulations are for using DALL•E?
I see the disclaimers now, yet one more question. Is art that is affiliated to religion allowed?
Of course it being peaceful and loving for the sake of AI learning
Padded Image (50% smaller), what i have to next
can i get a discount if i get a DALL-E 2 tattoo on my upper arm, it's free real estate
Quoting FAQ.
• Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.```
Here you go!
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/term```
Is there any way I can add text to an image created by DALL-E?
just through external software
guys whats the dpi for the generations?
The standard if you simply download the picture will be 96 dpi :)
got it thankss
Hi, what is the color tag (yellow, light blue, green, orange and blue) at the bottom of the images? Can we remove it?
That's the logo as well as the watermark of DALL-E
And yes, you are free to remove it with an external editing software
Thank you
Is there an option to increase it?
From Dall·E itself, no. There are ways to do it though, through external upscaling software e.g
If you want to know more simply send me a dm! :)
Please notice that it's only informally fine to remove the watermark.
It's still a grey area in the policy
Hi, I'm having trouble buying more credits, my payment keeps failing
Can someone help?
contact support@openai.com we can't help with billing issues here. sorry
Is there a pixel perfect way to do this?
You could try and erase the bit you wouldn't like to have in an external
program like Photoshop e.g if you want/need pixel perfect results :)
lets move to #❌┃prompt-help
they really are!
yeahh, but it sort of counts into that category
Wouldn't want to spam questions for this :)
Will Dall E ever be integrated into the discord like mid journey? It would be kinda cool but I would expect it to build up clutter
Can dall-e generate wallpapers? Like for pc(1920x1080), 95% answer - no, because is too high resolution and they have much more users generating arts right now. But can generate wallpapers or not?
You can combine images, that way you can work yourself up to 1920x1080 e.g :)
hm
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y
Take a look at this, everything starting page 55 should be of relevance to you!
also check out #inpaint-outpaint
does rendering the ' ' in DALL-E 2 use the entire model +- backend safety modifiers?
god I hope not...
Is there a hotkey for outpainting?
the left, right, and middle mouse button remove the image from the canvas, but can you remove all of the image except the paint tool selection?
Is there something I can include in my prompt to help scale things down? The subject matter in a lot of my images are cut off at the edges.
I think it is to do with model maximisation?
Like DALL-E 3 or 4 would be less of an issue in this regard?
Like bigger models will fit the image, better. Uncertain.
The training set probably has alot of images that aren't square.
I don't know.
but some people have success with photography modifiers
like wide shot
thanks, I'll give that a try
Can anyone tell me how can I create long form content using GPT 3 ?
Yep, like that user pointed out, terms like wide shot help and I recommend you check out the prompt unofficial book for more ideas... ✨ https://dallery.gallery/the-dalle-2-prompt-book/
could mods maybe add that prompt book to the #gpt-models channel?
oh wait it is
just very tiny
maybe give it its own embed?
has anyone created a animated cartoon snippet using DALLE? have seen capabilities for character design, storyboards etc and seen animated zoom videos, wondering if cartoon has been made yet!
Question: If I’m publishing a book with one of the NFTs, does it need to give image credit to openAI in the copyright page? Am I allowed to take off the small watermark logo on the bottom if I do?
yes you need to make it clear
also the generations aren't nfts....
you can remove the watermark but you need to state it was made using the open AI Dall-E tool
Do all the images save onto the site automatically?
no you need to save them to your collection or download them
The history tab only covers the last 50 prompts
Can i upload pic and then describe some details?
If I don't use all the credits, can I collect the free ones to have more?
No they will reset
Oh, It would be great to collect and having a lot in a time.
is it allowed to add your own watermark to generated images? the dalle watermark is easily removable and anyone could remove it and claim to have drawn it or rendered it and so on
I imagine they have proof of who rendered what if there was ever a legal challenge
so make sure you save to your collection
If you buy 115 credits for the 15$ do you then also get 115 the next month or is it still 15 every month?
They expire after a year
Free credits expire every month, purchased credits expire after one year
Oh ok
But I meant does it change how much credits you get every month if you buy the 115 or does it just give you 115 and nothing else?
I think you still get the 15 every month on top of what you bought
There is no way as of now to clear a specific creations but you can create a new session to clear everything in the recents section

Yup everyone gets free credits
Yes, you can remove the watermark but give credit to OAI/DALL•E 2
I just read 18th tip and trick. And I don't know what it means when they say they need an openai key. Wha is openai key? It is my invite male code?
who here has experience with uploading a photo of a object into dalle, and having the outcomes that are taking that object into different settings/art styles? everywhere i look i cant find good examples of this
you cannot do this yet :( if you are implying style transfers
can you screenshot what you're talking about?
what about same style into different backgrounds and objects in scene?
Oh, I'm sorry. I just solved the problem. You don't need to help me.
i want to put my jewellery product photo taken in professional setting with plain white background, to upload into dalle, and have it sit inside a jewellery box whilst increasing the shine of the jewellery
you can erase the background (in the dalle editor or a photo editor, saving as a png, and then upload again), and describe the full image you'd want to see!
i'm not sure you can increase the brightness of an object through the application im afraid
instead of brightness, is it possible to add a lensflare effectt? im wanting to make a diamond seem more valuable thru higher shine etc
ive had trouble finding good examples of this! will dalle be able to directionally rotate my product around and generate what it would look like from slightly different anngle, or will it be capped to looking exactly how it looked in the uploaded photo, just with different visual style/backdrop?
I actually had accidental success with something like this. You can put your product into one corner and try to make it a diorama or other image with multiple panes. I had it happen with this astronaut. I got this original from DALLE
and then shrank it and put it back in with the same prompt
and it gave multiple images of him
wow that's amazing! i don't see it work this well usually
I got a question, I have been trying to remove some trees/land from an island and it just doesnt want to, i have tried many prompts
(I have selected the area)
Is there another way to tell it to remove the trees?
the diorama tag did really good with this for some reason https://labs.openai.com/s/gqvX2hwyPWtt2zd5QnTm1lNh
Can you post the image you're wanting to remove the tree from?
I have been trying to remove the trees from the right side so it turns into a peninsula
first of all gorjus generation
second, why not try describing the right side of the photo in your prompt (ex. the right side of the photo is sea, giving off the impression of the piece of land being a peninsula)
Giving a natural language description of what you want to do TO the image won't get you the result you are looking for. You need to describe the image as a whole as if it already exists.
So realistically you would want to do something like "a photograph of the shore of an island with tall trees and a sand covered peninsula"
alright, thanks for the tips, ill try them out
I'll see if I can get the prompt right as well, trying a few now
may take some testing
The problem is, it may not have a good idea of what a peninsula is from such a low angle
is there a way you believe that this could be 3d rendered? or would the only real option be to change the environement around it?
3D rendered in what way?
Like obtaining a 3D model of the shoes?
something like these
Do you have any example images for what you'd like the final image to look like?
but i guess without the angle change from the original white background one i sent above
Hmm yea Dalle would not be fully capable of doing something like this unless it's similar to this #community-help message
You could put the image of the shoe in the top right corner and then see if it will generate the same shoe from different angles
no need for it to be different anngles, more onn the background and ability to innteract with the environment similar to the top black image
Ill test something, lemme try
Ill add the fog later but something like this
ahh ok so you just want the water to be there, not ALL trees
yep
u sir are a legend! when u say put shoe in top right corner do u mean resize it into corner or crop, bit nooby question i know but ive struggled to find solid info around uploading photos and the capabilities
Yea they will definitely be adding some tools to assist with that sort of thing on the web in the future, but right now having a 3rd party photo editor is a big help (GIMP and Photopea are two good ones)
But as for the "top right" thing, thats more of just a hack to see if you can get it to generate multiple different angles or something for the same product
so I took this image and rotated it and erased the other shoe, it gave some interesting results
Lol but yes in all seriousness if you do it correctly you may get some good results like these
just may take some testing
So in this instance this was the image I uploaded
and here was the prompt I used: "a product photo of a black tennis shoe surrounded by a dark burgundy silk, studio lighting, cinematic"
damn this is awesome! its crazy how it was able to figure out the different noses of the shoe, im impressed - i believe my main use for dall-e will be use cases like this, taking product images and inputting them
damnnn
Definitely! Looking forward to seeing what you come up with :)
Had to cheat a little bit but this one came out ok!https://labs.openai.com/s/RdfUbWkMOA3Pr7kaf9DbDnD5
oh wow, thank you, is it ok if I use it?
So what I ended up doing what trying to help dalle by giving it some of the sky in the area where it should be so I uploaded this image:
With the prompt "a photograph of the tip of a peninsula, water surrounding island"
So I inpainted some of the sky using photoshop and just let it blend the horizon and the water
ohh interesting technique
But sure! Feel free to upload that image and see if you can get some better ones :)
Here were the other two:
this is essentially just adding drawing black or grey over the parts you want dalle to focus on?
Discord just makes it look that way, those areas are actually transparent
So with your shoe image here: everything is just erased except part of the shoe
So when you're on the "edit" page, just erase everything except the parts you want to keep, then in the prompt text, give it a picture of the whole scene like "a product photo of a black tennis shoe in a pool of lava" or "a product photo of a black tennis shoe in a fish tank"
I inputted this - "a product photo of a black shoe placed upon dark burgundy silk, studio lighting, cinematic"
these are the results
Yep, so on the edit page you will need to erase all the white around the shoe
so you'd expect it to look something like this:
You'll want to erase part of the shoe as well to give it some space to blend in with the scene
ok so its important to slightly creep over the boundary of the shoe? i was tryinng hard before to make sure i didn't hit the edge haha
Yep! it may not be 100% accurate to the product as dalle will try and guess what the rest of the shoe looks like
but using that example I showed, it gave this result with your prompt
@carmine girder hop over to the #❌┃prompt-help channel, I dont want you to get yelled at for asking all this here :)
thank you much appreciated haha
you were close…
Anyone else has their card getting declined for no reason when buying credits?
The card works fine anywhere else
Definitely triple check all of the information you are entering. DALLE uses stripe as its main payment processor, if they're having an issue with your card definitely consider reaching out to them!
Thanks for the info 👍
I tried searching but using "quotes" in the search bar isn't helping much with results. Does putting word combinations in quotes have any impact on the results? For example if I were to say "Liver Brindle Pitbull" would that work better than without quotes cause the word liver might somehow be interpreted differently? EDIT: Or would commas function the same way I guess?
yes punctuation of all kinds definitely has an effect on the outcome
quotes definitely help with getting text or numbers to show up accurately in a prompt, but surrounding a phrase like that Im not sure
Thanks
possible to have more than 4 results show up? would be happy to pay say 2 credits for 10 results etc (bit more value) at once rather than having to generate more instances
how do i change my account settings ? like adding a profile picture or changing the name added
If I want to use a generation as an avatar, is it okay if I can zoom/crop it to how I see fit, provided I use DALL-E Bot to watermark it again?
It's **informally **fine to remove the watermark. :)
It's still a grey area in the policy though, so just keep that in mind.
Also, please don't decieve anyone into thinking that your creation wasn't AI-generated.
Have a read through these:
https://labs.openai.com/policies/content-policy
https://labs.openai.com/policies/terms
Refer to this :)
#community-help message
unfortunate
am I allowed to use the generated images for my own purpose like as a pfp or album cover. and am I allowed to share the images I generated to others
yes you are
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/
Yeah, I figured as much. I was actually going to use the bot to put the watermark back in.
For sure, share them, use them!
By the current ToS you need to disclose the role of AI though, and the watermark is the easiest way to do that :)
(keep that in mind for the album cover e.g)
You should also definitely read the Policies, which are linked above.
Yup, that works great 
What is the best way to generate art using the style of uploaded image but that isn't really a variation or edit of that image?
I have often noticed that it's best to keep the most defying bit of the picture in there.
Have a look at this #community-help message e.g.
Simply give it a try and experiment! 
Oh I thought I had seen that exact example but couldn't find it anywhere, thanks!
Fs!
I don’t have access to a PC at the moment so i have to use my phone but is there a way to see the 50 most recent prompts created on Dalle. Website mobile, im not finding them??
is there any way to find your older generations that you didnt save
that arent on the recent list anymore
You can use the desktop-mode for mobile, that's the only way I know for mobile right now. :)
I tried that, and no luck. :/
#❌┃tips-n-tricks message
Take a look at this 
Mh.. have you tried using a different browser and giving it a try there?
Yes, safari and chrome for iPhone
So far I only have tried it for Firefox, maybe give that a try(?)
If that doesn't work either you could check your browser history for the URLs, which is of course more tedious.
Did you try turning your phone sideways? That’s what I do on my iPhone
Thats a good idea checking for url history thanks, i just needed the most recent
Yes
This was perfect thank you!
Fs, happy that it worked out! 
hey :) my partner just got access to dalle 2 but she can’t log in, it keeps saying ‘authentication failed’. anyone got any ideas how to sort that?);
Can happen sometimes, give it a little bit and try again.
Also try navigating directly to https://labs.openai.com/ and check if that works
(in case if you haven't been doing that already)
thanks!!she’s tried that url, but can’t seem to get to dalle that way, only to the text based ai 🤔
maybe cos she’s on mobile?
weirdly it works fine for me 😭 i’m on apple and she’s on android is the only difference. trying a different browser is a good idea tho, ty
Huh weird, happy that it worked out at least on your phone though!
Let me know if the different browser does it :)
A friend of mine signed up for DALL·E 2 and she DID get the email, however her authentication token apparently expired; she's contacted support however it's been 4 days and no response. She thinks support is ignoring her emails because her email address, is it normal for wait times to be this long right now?
Any resources on copyright or commercial usage to read through?
Do I essentially "own" my renders?
Have a read through the polcies
#community-help message
Haven't heard of the authentication token to expire before tbh,
you can give the live chat on beta.openai.com a try! :)
how does the edit image thing work?
Take a look at this, starting page 55 should be relevant to you!
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y
I contacted support 2 months ago and haven't heard back. So either I was ignored, my request is very low priority, or waiting times are just that long 
How can I get Dall-E to create minimalist arts like this one
Probably the last option, long wait times
Try #❌┃prompt-help
#❌┃prompt-help! There are a lot of helpful people there :)
the best way to remove the watermark in the bottom right whilst still keeping the same size image?
via 'inspect-element' in your browser!
#❌┃tips-n-tricks message take a look at this video :)
Or, you can also simply edit it out in an image editing software
u r a legend sir!
A fast one at that 
Have a read trough the policies though, just to stay true to the policies
#community-help message
Is 15 dollars worth it? I don't really have that much money honestly, I apologize
It's an amazing tool, so yes to me $15 is worth it.
However, you should never spend money that you don't have on something like this.
It really comes down to you, what you do, etc. 
(no financial advice, all of this is my opinion, and I'm not associated with the team of OpenAI in any way)
They can be imo, but not all of them
How can we provide context aware inputs for images related to cybersecurity?
Let's say I need to do a presentation about a a certain attack vector. Rather than relying on existing images, I want to see what Dall-E can produce. How do I make a query for this without breaking the policy?
It depends. Does it involve language that you think would unintentionally trigger the prompt filter? Or do you think the nature of the generation itself might be against the policy? Read up on the policies and terms if you haven't already #old-server-rules message
alright, gotcha, thank you
I came here after reading the policy haha... There are many things in the gray when it comes to cybsec, so not sure what exactly can be considered as a policy breach.
As for the language, I think words like attack or bot might trigger it.
Wish there was a way to confirm that might part without getting the account suspended
Not sure if this image is again the server's policy, but I'll share it just in case. Mods please let me know if it is not allowed and I'll delete it right away.
"attack" trigger word
I'm not sure exactly what you want the images to represent, but there are probably ways to create visualizations of attack vectors without including language specific to it. e.g. instead of including "DDOS attack", describe exactly the diagram you're visualizing, piece by piece. Keep in mind that dalle isn't great at detailed diagrams or text.
What would the ideal end result be for you? Like what would it look like?
uhh not sure how to explain exactly but a diagram with arrows pointing between the multiple components
I see
I think the main issue you're going to run into with that is that you're not going to get a correctly-labeled, realistic diagram. You would get something more abstract that represents or looks like that type of diagram
Funnily, ||Charmander Killing Bulbasaur|| wasn't flagged haha
Something I've noticed: re content flagging is that changing the context that the flagged words are in can allow it through. For example, you cannot generate aliens abducting humans, but you can generate a UFO abducting the eiffel tower - which I'm guessing has something to do with the latter involving inanimate objects.
So if you're e.g. want to generate an abstract representation of that type of attack vector, you can abstract it, like: "evil programs invading a cyberspace world" might work.
oooh
It just has to be (I'm guessing here) abstracted enough so that it doesn't represent an actual real-world thing, like cybersecurity - it's just a tron-like illustration of a battle between good and evil programs, you know?
But your earlier point was right. Dall-E might not be suited for factual concepts
Yep. Same with text AI right now, you can ask GPT3 to write an informative report about something, but there's no guarantee it will use actual real facts (unless you specifically feed it a report of those facts)
(Edit: actually I only have firsthand experience with gpt2 so maybe 3 is better about this)
crazy
Dall·E knows everything 
Here's my favorite example of that btw >_>
I gave gpt2 the text prompt:
"Here are some unbelievable facts about eyeballs:"
And its reply started with:
"Here are some unbelievable facts about eyeballs:
- In the human body, there are approximately 250,000,000,000,000,000,000,000,000,000 of them. And that’s not a misprint."
I have... Absolutely no idea what I'm doing here honestly, and could really use some help.
I cannot for the life of me figure out how people have been creating such amazing things from this AI while every prompt I give it ends up looking like it was drawn by a third grader, is there some kind of hidden rules for how to structure a prompt, some settings menu I can't find, or something else I'm missing? Or am I just being impatient and not trying enough iterations?
Im happy to help! What did you have in mind?
I'm working on something like that but nothing quite like you're looking for yet. If you can show us a prompt, what it created, and what you were hoping for though we can definitely help in here 😄
everything
Why do you have over 100 emails on the waitlist?
Thank you LJ
Hi i am new to Dall.E does anyone know what we are allowed to do with the images generated? A friend of mine liked an image i got so much he wants to get a print of it
i remember hearing somewhere that you can sell your images? that doesnt sound right though
You have to state credit to openAI DALL-E tool, they have a disclaimer on the website about using images
@rapid knot As long as no attempt is made to misinterpret it as human-created, you retain commercial rights for your image
oh thats really cool
(and as long as it meeds the content policy guidelines)
yes its just a dinosaur being interviewed on tv in the 1940's
If you print bulk and sell then the disclaimer on the website needs to give credit I would believe, but ask an admin
Kinda like this lol
Caveman zoom
The only time you need to provide a disclaimer that the work was AI-created is if you are demonstrating your use of the DALL-E tool rather than just the output images you generated. Other than that, the rule is just "don't deceive people into thinking it's human created" - at least according to the content policy on the website.
I'd be happy to help, just let me know :)
But that may be different in cases where the watermark is removed as indicated above
This is correct
Yes
Then credit must be stated somewhere if it’s removed right
My leg just fell asleep
Well you need to verify your phone to use it so those are all useless anyway
Take a look at the prompt guides that are listed in #gpt-models -- you will see many examples there to help you get started.
And probably breaches some kind of TOS
Spam
It does indeed
I mean I'm gonna be honest I think it'd be a shitty thing to do if they're limiting numbers of people
If you want to go through the trouble of paying for multiple VPNs and phone numbers rather than just buying credits, thats your own concern
😂
Polar this is kind of like walking into a police station and asking "It's okay that I committed fraud, right? Like I won't get arrested?" 🤣
So for someone to spam loads of emails on waitlists could take up space for a different person to use it
Wow
Most people are not concerned about others, unfortunately
So many dollars
Why are you in the official discord talking about it though?
That should be the button “Just Buy the Credits”
So, I've been trying to create an image for a character of mine for an RPG, using the prompt:
a muscular, middle-aged man with mid-length brown hair, reddish brown eyes, and a slight tan, with a scar across the left side of his face over his eye, with many intricate tattoos on his arms and shoulders, reading a book with a serious expression
And have tried in several styles. For "digital art style", the images were of decent quality, but not exactly what I was asking for, and one (first attached image) wasn't even tangentially related to the prompt
Nah I hope not. Hope their emails are found and banned anyhow
Such a shitty thing to do for a closed beta
Where access is limited
I tried again in "anime style" and these were what I was given
They're just in here to rile people up, you can just ignore them @rose glen
You can in here asking about something which is clearly going to either be against the TOS or against the spirit of this place
Maybe type in “Aged man with scar on face reading book with stern face”
#❌┃prompt-help for this please! 
Digital art *
You're the one bragging about trying to game the system?
I mean they've actually been posting their gens
FYI systems have been developed to prevent this, and the mildest outcome is all accounts lose free credits in perpetuity.
This particular conversation is now over.
So I assume you are looking for an anime style specifically?
Yes
Hopefully this is the right channel and not too stupid of a question—but when I saw others using DALL-E, they generated 6 images in a grid. I’m getting a line of 4. Is this a new update or am I doing something weird? My invite was for v2
When the beta came out they changed the number of images from 6 to 4 to accomodate the number of users they are adding
Awesome, thank you so much!
I’m curious how many people have signed up for the beta, but it’s alright if you can’t share
more than 80, fewer than 17 billion
how do I know if I'm at v1 or v2
i'd say its a heckload, friends of mine who found out i have access have been signing up and getting in within the same day, so, is a lot
I got it two days ago 😌
Question: Can I use a picture of my home and then add in the promt “construction man on the roof building house digital art”? My dad does construction so to make an image of him making the house, as he did, would be neat
Carpenter *
Why are there color squares at the bottom of the generated pictures?
trademark?
I'm wondering if it's that, like a signature kinda thing.
The logo shows up on the tab, so yeah.
it's the watermark!
You're meant to make it clear that images are made using openai/Dall-E when you post them/use them elsewhere
and the watermark is the easiest way to do that
you can remove it so long as you have a text disclaimer
Is there a good way to force Dall-E not to try to generate text on top of a picture? It can be kind of annoying
Hello I have a question, can we sell or publish as an nft collection of our ai produced images?
you can indeed
are there any legal restrictions about it? I really would like to get a legal answer, I see some other people done collections but Im a bit afraid tbh
it doesn't feel right
They have a sharing policy on their website
afaik right now, you just need to make it clear it was made using OpenAI/Dall-E
In the TOS, OpenAI say that whilst they retain ownership of the generations, so long as you state that the generations were generated by DALL•E 2, you are able to sell and share your generations.
you can even remove the watermark so long as there's text stating that ^
I see thank you guys, so as long as I mention that the art is generated by the openai dall-e, we all good
yep 🙂
❤️
Hey all! I just have one question.
I've been having fun generating pictures of frustrated cartoon badgers storming out of supermarkets, but when I try that prompt without specifying a cartoon style, I get dinged for content policy.
Is there anyone who can help me understand what part I'm running afoul of? I don't want to get auto-banned. Thanks in advance!
Do they plan on documenting the HTTP API? I noticed you can ask for different batch sizes.
It would be helpful if you could share the exact prompts you've been using
Sometimes some words in some contexts trigger the filter unexpectedly
Hey, yep - so it's literally 'A frustrated badger storming out of a supermarket, cartoon', and if I subtract the word 'cartoon' I get dinged
is there any aspect/ratio control in dall.e
Not in-app, but you can expand images to any aspect ratio using outcropping and composite techniques. (Will cost more credits)
Here are two good resources for getting started with it
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y (Page 63)
#❌┃tips-n-tricks message
Thank you.
I doubt it. A proper API is in the long term plans. See below.
🤖 Q: Is there a publicly accessible API for DALL·E?
• 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.
Weird that they say that since I'm using the publically accessible API.
By using the browser.
I wonder if this was written before the credit usage feature was released.
They mean an API in the sense of being able to access it using Python or something
I know what they mean it's just a bit nonsensical in the material, there's very little difference between a browser and curl making an http request.
Sorry this probably isn't the place to discuss this, thanks for answering my question.
How do the credits work? I know I get 50 free credits, but I am trying to do a D&D campaign and want specific images and may have to make multiple tries. Does each try count as a credit or does one saved image count as a credit?
dang, so I really have to be careful what I write to avoid using up all my credits?
The only thing you have to be careful of is pressing the generate or variations button. That's what takes up credits.
If I add a friend to my OpenAI organization will he have access to the DALL-E beta as well?
Is there a way I can view my creations older than the recent history? I know now to save and add to my collections.
Beyond the sidebar you'll need to have saved the /e/ addresses of your prompts. They should be in your browser history if you have it enabled
@split hazel Oh nice I do see them, glad it didnt clear it, thank you
Is there a limit to how many you can have saved to Your Collection?
10,000
Got a question, some prompts are getting me dinged and I am wondering why. I’ve read through the content policy and I don’t think, “the creation of life and death” violates it? Just got access so I’m still playing around with Dall E
Can someone explain crops and image editing? How can I request the Ai to re image the same image but in a different art style?
would it be possible to further train the content police model to be better at distinguishing prompts that technically violate one or more provisions but in a harmless way? example:
countless people have requested this lol
hopefully its fixed some day
did u try "capitol building in washington"
no, i'm pretty sure what actually flagged it was "protest"
ah i see
is there a character limit for prompts?
i've never run into a formal character limit myself (though it probably exists) but i've noticed the more convoluted i make the prompt, the more likely it is to return random photos that don't match
I've had 'protesting' pass on a prompt
I have a question regarding content policy violations. When you enter a prompt that you don't think is a content policy violation and you receive a message stating
"It looks like this request may not follow our content policy."
"Further policy violations may lead to an automatic suspension of your account"
Does this statement means you have received an infraction on your account and with sufficient infractions or multiple infractions in a short amount of time your account will get suspended?
yeah tried it again and you're right. it must have been the mention of the Capitol that got me
I believe if you're suspended you can ask for a human to review your history to see if you were intentionally trying to circumvent the restrictions.
I see, alright.
Correct! You can get in touch with them at support@openai.com
nope, wait, maybe it's the combination of "protest" and "Reuters"
it flagged me when I included modifiers to make it look like a news image but not before
I feel like there might be some sort of combo system but can't confirm exactly. I was having trouble with torchlight in ruined churches the other day
i would really love for rejected prompts to say specifically what i was doing wrong
i feel like J. K. Simmons at the end of Burn After Reading
Same but I also understand why they don't:might make it easy for those actually trying to bypass filters.
400
Is that allowed to use prompts like "~ in <Company or creators name> style"?
I think this is unlikely, but I wonder if this is actually allowed.
Ethics aside, it works, it's allowed, and it's often incredibly effective.
I wonder if there are any legal issues to watch out for in this regard. (i.e. Intellectual property infringement in Terms, etc.)
Hey question does dalle-e have an invitation system ? Asking for a friend who wants me to invite him lol
No but the waitlist turnover is fast right now. Send them there.
Can someone tell me the name of the website where I can find better prompt ideas?
I'm not sure if this is what you are referring to but try this: https://dallery.gallery/ and be sure to check out their prompt book! (Unofficial) ✨
A guide to OpenAI's DALL·E - prompts, projects, examples, and tips
New to this, and just trying to figure out how to add to the collection, any help is appreciated, thanks
You can find this button in the upper-right corner once you click on an image :)
Thank you, much appreciated!
Yes! This is what I‘ve been looking for, thank you!
np!
can i use the open ai logo on things i create with dall-e?
Are you referring to this -> 
or the actual Logo of Open-AI? :)
this guy
the dalle logo i assume i can use because it's stamped on every image :P
okidokes :)
is there any limit on the amount of images stored in collection?
If I try to prompt "Tiananmen square during a beautiful sunny day of May 1989" does it break any content policy? I suppose one could argue a case to support it, but I don't want to get banned 🐣
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/
Political: politicians, ballot-boxes, protests, or other content that may be used to influence the political process or to campaign.
10,000!
Oh, right, thanks
what happens after the 10k? the earliest one disappears into the ether?
Not sure, I'm at... 281 😅 I don't imagine so, but I can't confirm. Might need an old timer from back when it was 1,000 to chip in.
I guarantee it flags stuff like that to avoid the spread of fake news
Hi! I just got started using dalle but I already tried almost a 100 prompts, but now I see that on the "Recent" tab I can only see the last ones. Where can I find the whole list of my prompts and the outputs I got? Thank you all very much. (If I lost them for good I'm gonna cry...)
(If I didn't add those to my collection)
Ah unfortunately the history tab only keeps the last 50 generations
You'll need to get into the habit of saving every generation
But also your browser history should still have the URLs!
Oh god
any idea on how fast? my friend got in a day or two after me and I've had access since the 27th. Do the filled out job titles/social media platforms make a difference?
let's hope i can retrieve them, thank you
Less so these days, I believe. I've heard anecdotally of users being added within hours.
For the use of their logo, the best case would be to email OpenAI
I see; are invites being sent out daily?
Hello there everyone, when I upload an image, should I have text in the promt?
Yes, if you're editing it's best to reprompt. (so erase areas and add phrasing to focus on what you want ADDED)
how do I see my old creations that don't show up on recent list?
if I didn't save them are they just lost?
They're still in your browser history! :)
the answer to this question is probably obvious but since "furry" is kinda a blacklisted term on dalle, would it be bad to make variations of art i have of my sona orr 
#❌┃tips-n-tricks message
Take a look at this!
Basically, yeah
Or bookmark/save the ones that you'd really like to come back to 
Pretty sure it's fine if they're SFW and they get past the facial detection. If it starts spitting NSFW art back though, report it and stop 😅
aight gotcha 
if the generated image doesn't match the description i wrote do i get the credit back? if not would that be a fair addition?
Add this to #suggestions if you believe that it'd be a good idea!
Also, please flag the images that you think weren't accurate to your prompt 
(see picture down below)
done both
Hey there, everyone! Is there any way to redraw an image in a certain style using the AI?
Are the credit prices in USD?
This is not possible as of now, maybe in the future though, who knows? :)
Yes they are!
Also be aware that the $15 do not include taxes.
Taxes will be ~ $2.50 on top, that depends on where you live though.
Hello! I'm new to Dall-E 2, I just got access this morning and it's exciting! I don't want to break TOS, but I also want to create fantastical creatures and illustrations for inspiration! One thing I found that Dall-E mini was great at was creating strange, twisted horror movie like creations. It provides the best inspiration I've ever had for modeling creatures beyond our imagination. Is this allowed on Dall-E 2? Or is it something that I should avoid in order to retain access to Dall-E 2?
Currently there is not
That's fine honestly, I've realized variations do a good job too ;) Thanks guys!
There shouldnt be any issues, please read the Content Policy for the full rules
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/
I think the only thing you gotta keep in mind is the AI's G rating, so monsters yes, horror probably not
Sweet thanks guys! Can't wait to see all the creations here!
Hey is there any admin here that can help me with my login? I initially signed up with login with google but I want to switch to a traditional email and password. I also opened a ticket on the website and waited a week and received no reply. Thanks in advance!
Unfortunately not. OpenAI have exponentially increased their new beta participants and are swamped right now; hence the delayed email response. Best bet is to check back as often as possible for any #announcements that might relate to systemic changes with e.g. account management, which are almost certainly coming considering how often we see issues like yours.
Is there a list of "bad words" that will trip Dall-e out? i.e. Fight, War, etc...?
I actually got a warning for breaking the TOS because I tried to generate an image with Michael Jackson in it. Assuming because he is no longer alive, this is considered harmful ?
In this case... No.
MJ's banning was the direct result of an especially troubling image shared on this server.
Because I can assure that my intentions were not meant harmful
Of course! But the safety net has to be very wide on this particular issue.
Totally understand, thanks for the feedback!
There's no such thing rn, but anything related to war e.g will trigger it, yeah
Have a look at the policies for a little overview! :)
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
How many images can you get per credit?
1 prompt (4images) per credit
Okay I gotta ask
Hey, so I wanted to make some more noise on this, evidently something has changed.
My original prompt "A frustrated badger storming out of a supermarket, cartoon" is now getting dinged for content policy, whereas before it was not, unless the word 'cartoon' was not included.
I really love this thing, and I very much don't want to get banned, but again - I have no idea what part of the content policy I'm bumping into, here.
Hi everyone, I have a question about TOS, I'm assuming that Public figures are not the same as historical figures. I understand I cannot prompt for a living public figure but what about Jesus Christ? I made a Terry Richardson photo shoot of Jesus Christ, to me it respects TOS but want to make sure before I post... any insight would be appreciated, thank you!
Does anybody have any strategies for generating images in bulk? Say you have 2.5k text prompts you want to generate, what would be the fastest way?
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.
Hey! Can I use generated artwork as a cover art for a music release?
Does anyone know how I can add someone to my account but they have their own login info?
Hi everyone, I have a question, I'm trying to create some images inspired by Goya or Lovecraft, so dark, gloomy and frightful. I've written something like "An old man sitting at a table" and then added adjectives from page 14 of the DALL-E2 "Prompt book" v.1.02, under "Negative mood, high energy", like terror, frightful etc. After loading for a bit I got a warning telling me that the image was not ok or something and it told me I could get suspended. I'm confused, I thought words from the prompt book were specifically used to give us examples of what to write. So how can I prevent that?
First of all, have a read through the policies
#community-help message
It's currently a grey area, so have a read through the ToS and make your judgement based on them. :)
Hey, I am the owner of a beautiful Maine-coon cat and wanted to make some images with my favorite cat breed, I put in "Maine coon" and got a warning for possible content violation, I tried again before realizing that it was probably flagging the second word, it said there is an automatic suspension thing if you did too many so I was just wondering how many flags it'll give you and if its something I need to worry about, thanks!
No that's not possible as of now.
Add this to #suggestions though if you think that it's a good idea! 
Let's move this to #❌┃prompt-help 
Can credits carry over to the next month?
Free credits will last for one month from the day you have registered. Paid credits last for 12 months! :)
oh ok
There is no numeric threshold that we are aware of that causes someone to be banned :) you shouldn't need to worry but I wouldn't keep trying it over and over :)
Here are a few other names you can try!
I would leave "coon" out of the prompt for now as it may set off the content filter
I assume this is for some sort of NFT generation? There is no current API for DALLE so this is not possible in an automated or quick fashion
Ok thank you
As long as you are not clearly abusing the content system you will not be banned, this prompt is innocent and should be fine to test which word is being picked up. You have nothing to fear if you are using the system as it is designed and not trying to bypass the filter to create NSFL or NSFW imagery
Does Dall-e make slight adjustments to the portion of the image that hasn’t been erased (in edit mode)? I was hoping to zoom into an area of an existing image and let Dall-E improve the resolution while filling in other areas on the side
according to joanne the banning is automated but reinstatement is manual
maine coon cats are on the list for being unfiltered #❌┃prompt-help message
once they have a chance to get to it
Apologies I mis-spoke, I just meant that there is no numerical amount (that I know of) that if you meet, you get banned. So it's not like you get 15 warnings and then you're banned.
I could be totally wrong :) but I've mistakenly got the content filter 20+ times and Im still good to go as well. It happens!
I imagine it's probably a leaky bucket
where hits expire after a few days
or get less important
Very possible! I'm sure they will give more information about it in the coming months
since otherwise, given the chance of getting random flags from gpt3, all the old accounts would be banned by random chance
(admittedly I'm only assuming their ML model is GPT3 but c'mon what else could it be)
I can't seem to edit images with DallE on mobile? I entered my prompt and erased the area I want DallE to fill in but when I click generate I just get the spinny loady circle in the generate button, but it never proceeds to generate
is it down for anyone else?
Thank you! I will begin testing tonight.
just refresh
thing happens to a lot of people from time to time
pretty sure that I unintentionally verified 2 accounts with the same phone number, will that cause any problems? I don't plan on using the 2nd one
So out of curiosity and I don't mean to perhaps ask a dumb question, although how would one go about getting on the waiting list for DALL E 2 ?
Is it okay to remove the watermark and either add "Made by OpenAI" or move the watermark from its original place?
does anyone else not get great results when wanting something with multiple heads ie 'a cat with 12 heads' or 'a 12 headed cat
I normally just end up getting images with mutliple cats in rather than a single cat with mutliple heads
Why is the term "guns" or anything pertaining to firearms banned? It doesn't seem to be in violation of any of the content policies.
Unless it falls under "violence" then it makes sense.
yeah I was gonna say they probably did it to be safe so it doesnt conflict with 'violence'
Never mind I didn't read the top "Do not attempt to create, upload, or share images that are not G-rated or that could cause harm."
in what context would a gun be used not for violence LOL
that being said swords arent banned?
I'm sure I've done 'cat holding a samurai sword fighting a fish wielding shurikens'
I just tried to create a Chrono Trigger team picture and DallE doesn't seem to know those characters, so I described them instead and it lead to a warning. ugh. Didn't know the word "Gun" was a trigger.
It's informally fine to remove the watermark. :)
It's still a grey area in the policy though, so just keep that in mind.
Also, please don't decieve anyone into thinking that your creation wasn't AI-generated.
Have a read through these:
https://labs.openai.com/policies/content-policy
https://labs.openai.com/policies/terms
I don't think that it uses proprietary characters from what I've done. It doesnt seem to know and if it does certaintly doesnt use any characters that are owned by other companys
No stupid questions here :)
This is Dall·E 2, Dall·E 1 has never been released to the public!
I havent tried basic ones like mickey mouse though
OH!
Thanks, I've gotten pretty close to a cool looking image of goku's facing away from the viewer, but that's about it.
It kind of feels that the images i am generating in the past two days are qualitatively worse compared to the days before.
Was there an update, or do are my prompts just unlucky?
Joanne has said it's fine in her official capacity, that said the ToS hasn't updated yet
Sounds great! Thank you so much both of u for help :D
why is my collection empty?
you have to click on "save"
correct, but here ill send a screenshot
oh, Im just a fool sorry for asking
it just took about 3 minutes to load it up for the first time
all my saves are there
Anyone know if I can share my 'collections' page publicly? Or if I can only share the images one at a time?
As of now you can only share one image at a time!
I like that suggestion though, post it into #suggestions 
cool, will do!
Hey i just got DALLE access 2 days ago and i used up 41 prompts because i thought it was 50 prompts every 24 hours, is there any way they can refund me at least some of those prompts?
No, unfortunately there's not ;(
Oh well, i shouldve read up more
Is there a way to save old creations? My first 15ish creations I didn’t realize they wouldn’t be saved to my collection, I had one specific one in mind I liked that I screenshotted but didn’t save
I found this online, not created by DALL•E. Is there a prompt that may create this? I know DALL•E doesn’t always type text out correctly, so possibly not
I recommend uploading the image and editing with Dall-e. Though It's likely that the text will turn into random symbols
have personally tried any typography art yet tho
I’ve had success with “worm’s eye view”
This would be hilarious with dall-e words
go thru browser history to find them
then save em
they still exist just not in the history tab of the website
Do you know if they have an expiry time?
I do not
Yes
Meiji-bot
Actually I think a Maine Shag involves Stephen King, Susan Collins, a Lobster, and a very awkward morning after
How long does Dall-E store images for?
How do I use the edit feature?
Erase a part of an image you would like to edit, change the prompt to describe the full image now and run it again :)
Right now, forever, but on the history you will only see the last 50
So if you bookmark some generations, they (should) stay there, but I'd save them to your collection or download them if youd truly like to keep them forever
At the moment, indefinitely - they should all be active links to prompts in your browser history. This is not necessarily the status quo going forward though (and I haven't checked my oldest ones from June for a while)
I tried using the bot to make the version of my image that's like cut along the middle and split and then the embed said to just click once in the middle but it didn't do anything
Can you paste the image you uploaded here?
Ok perfect, and then what were you trying to turn it into?
I'm not sure to be honest, I didn't know I had to supply a prompt
Yep! So you will take that image and go to the "upload" part of DALLE, and then it always requires you to erase part of the image, which you already have, so you can just click in that empty space. Then you will supply a prompt like "two crying boys surrounded by scary faces" and see what it makes!
Ohh I see where I screwed up
I accidentally clicked to make variations
Interesting, to say the least
Thanks for the help! ❤️
No problem!
If I buy credits, will I still receive 15 credits monthly?
But they will stack on top of purchased?
yes
Cool, thanks
Is it possible to tell it to add text to an image? If so, is there a specific way to tell it, like adding quotations like "this"?
Is there api for dalle? so i can make like a node script or something along those lines?
Ah
Not very artistic or expressive.
Is OpenAI considering to commercially and selectively license DALL•E capabilities for other purposes than creating art?
I also have a question concerning DALL•E art and your attempts at teaching it morals
Hello! I've just received my invitation for Dall-e so I am very new at this, making my very first prompts. I just uploaded a picture of scenery and selected a part of it to edit, writing my prompt something along the lines of a man wearing certain traditional clothing, but the results were just the pictures of slightly edited versions of the original picture. Did I misundestand the function of editing pictures or did the results turn out unrelated to my prompt because it included humans? It was not about a specific person, just "a man".
When you use edit, it only changes the area you erase
You usually need to give it a LOT of room
and might need to be more specific with the prompt
Hi there. How do I display 6 not 4 generated images?
you cannot
If you saw it generating 6 in videos, that was before it entered this beta stage
supposedly reduced to 4 to deal with increased demand from the extra invites
Ahhhh gotcha thank you
So… DALL•E outdid itself and now I can’t upscale my image because it’s a realistic face. Great problem to have, but any ideas? #dall-e-bot message
Don’t know if it would make any difference if I don’t flip it first or not. Edit: Nope, still too real. Might need to smudge the face and fix it later.
Does anyone know, on what basis does Dall-E produce images when i enter random numbers e.g. 12919211293 to generate something.
It seems (atleast in my limited use) that each time I use the edit the newly added sections seem more blown out, is there a solution to this at the moment?
It tries to find that string in all its data. If there's nothing, or conflicts between a few examples it can't resolve, the model just prioritises a cohesive image - usually what it knows best, photos of people, food and nature.
Bigger edit areas tend to result in more cohesion. Tiny changes might need to be visually salient for the AI to recognise it's done the job.
Do they stack with the starting 50?
@past geyser could I get an answer on this? Sorry to bother you so late
Hmm. Yes as far as I'm aware at least one of those accounts will not garner free credits - maybe both.
Are you referring to verification of Discord accounts? I don't think you can use the same number on two Discord accounts.
No probs by the way, I'm sitting in the (intermittent) sunshine by a soccer oval in sunny Sydney 🥰
nah the dall e 2 accounts, I forgot I had 2 emails signed up for the waitlist
and they both got in, and I think I used the same number for both when registering so
obviously unintentional so wanted to ask about it
As volunteers we do not have access to the back end, so perhaps you should send an email to support@openai.com -- I will let Jeremy help with any issue that may cause here in the Discord server.
Shouldn't be a problem, from our perspective!
Question - can DALL-E 2 colorize black & white images? I'm not sure how to engineer a prompt to colorize a drawing or image!
No, at least not easily. Style transfer is potentially possible and could be used for this but we're awaiting an update about whether it's in the future.
How many chances do you get before you get banned? I got 3 prompt warnings already
Is there a way to get the ai to create new art with the same "characters"? My attempts at editing/uploading images have resulted in wildly different styles and features being generated.
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.
Very difficult unless you happen to stumble upon a really distinct series of descriptive words / adjectives that reliably create an iconic character effect. You could potentially use edits to seed a baseline of e.g. colours or hair textures...
How do you seed colours or textures specifically? Each time I use an image to try to seed a new one, I get something either totally different or the same image with no reframing/just additions to the base image, but worse.
MidJourney has an unlimited paid plan for $30 a month, what am I looking at spending with DALL-E to get something that feels unlimited? (I know they don't have an unlimited plan)
Depends how you use it. But yes a lot more.
If you upload a mostly transparent 1024 x 1024 png, with a few splotches of colour or texture, it'll work with those cues plus the prompt to build an image
Oh thank you that's amazing!
I used up 140 credits in 2 days
Definitely wishing it was cheaper to buy more.
For me im just messing around. If i were to actually use this as part of a business, for concept art for example. I could see it being worth it.
I imagine most other people will find the price a bit much
I used up my free 50 credits in like an hour. The fact that you have to make so many attempts to get a usable image is really putting me off buying more, plus the investment into learning how to get good results. It seems like I could get some good pics if I put in the time but I'd have to learn how to do that first, and it seems pretty pricey to pay for each iteration while I do that.
It's why we generally recommend hanging around here for a while and reading things like our #gpt-models links before you dive in. Often you're re-inventing the wheel.
Especially if you have a narrow vision. DALL-E can make beautiful images with a small amount of guidance, but if you're trying to recreate a naturalistic artistic process to get exactly what you imagined, it's definitely a credit-sink.
Yeah, I didn't know about the discord when I started playing around with it. I actually lost most of my credits to really low-quality results which I struggled to filter out. Other AIs have unlimited generations but they also have higher-quality baseline images to start with. Even if their detailing is worse, the images they put out look better.
Yeah i 100% agree, theres a lot of trial and error. I kinda feel like variations should be free or the first one free. That would help a bit
DALL·E is a finer tool. More control means more room for failure. The limits of what it can produce are still untested - that's what's makes it so exciting and worthwhile...
I definitely agree, but the 'more room for failure' combined with a pay-per-credit option actually makes this tool less desirable as a paid product, is all we're saying. Like, there's going to be more failures, and we have to pay for it.
DALL.E is definitely the better product, but it makes it almost prohibitively expensive in my eyes compared to other AIs. That's just my feelings on it after using a few different options though.
Best place for feedback on pricing is #suggestions - the pricing model is definitely being evaluated, based on both on our feedback and user behaviours...
Thanks, I'll chuck something up there!
I have a few questions about DALL-E 2.
-
Can I use generated images however I want (share anywhere, commercially, without attribution)?
-
Will DALL-E 2 get another cheaper plan in the future?
-
Can I use DALLE-2 in my own (public) app and if so, how?
1: You always need to stay true to the polices, so definitely read through those beforehand.
#openai-chatter message
2: We don't know, so far there haven't been any updates.
3: If you're referring to something like an API, no, not right now at least. They do plan to offer something like this in the future though! Take a look at this
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.
I hope this helps 
are we allowed to generate pictures of doctors? not doctors doing anything related to health, just a picture of a doctor
Absolutely! Something like "wearing White robe" etc. works just fine, simply make sure that there's nothing graphic and/or offensive shown 
perfect, thanks
???
is this because Ukraine is going through something? I'm just doing cool dog paintings in different landscapes
Yeah, Ukraine is on the nono-list right now.
Try something more descriptive instead! 
ah ok I see, removing Ukraine works
Can shared images be made private again?
sure, just use giga pixel lmao
Not afaik, but the only way they're "public" is that you created a share link. Technically all prompts / generations at this point are preserved by OpenAI anyway
I try to archive that dall e is adding a robe, but all i get is that my cat gets thinner or layed down. Any idea how i can archive my request?
In the prompt for editing, try describing the whole image you want to see, e.g. something like "A cat with a crown wearing a King's robe" or something similar.
I would have tried adding coronation robe, king costume, king robe to the prompt
Do you think the marked area to edit is correct?
Looks okay to me, you can also try by erasing the body but credits are precious
I didn't share the link, it's my favourite. I know but that's a bit different, technically.
Yes thats true the pictures vary every time you make the same prompt.
This is the only one I got with something resembling a robe, the other images where like yours.
Please dont waste credits for me, I wont be able to pay them back 😅
Don't worry I see it as a challenge.
Any idea how i can a whole picture into another artstyle?
Lile the picture above as oil painting
You cant iirc
Ah ok, because they adverted that feature in one of their videos
There are other models that can do that, like Nightcafe, but DALL-E at the moment does not. However, I should it should very well be possible for them to implement such a functionality.
Ok, then I will be patient. We get 15 new creds next month. Then I will try again
In my experience it doesn't matter what order you do. If you use the right prompts it will work
Probably all in your imagination
I wanted to change the pipe in Magritte's "the treachery of images" but got a message saying the image might violate the content policy. Do you know why this happened? I got it from wiki art but modified it slightly to make it square.
You gotta be more specific than that. Vagueness gives DALL∙E the freedom to generate anything.
Salute guys! Can someone tell me if I can use generated pics as cover for my music album? Or in other words commercially use
Read up on the terms for commercial use
#old-server-rules message
^
By the current ToS you need to disclose the role of AI, so keep that in mind for the album cover 
It's currently a grey area, so have a read through the ToS and make your judgement based on them.
I saw one of those videos today where some used a picture of a Tesla and made it older and older. Can someone maybe explain to me how I can do that?
Thanks! If I understood everything correctly, then the answer is yes
I will make an announcement in social networks 🙂
Is there a way to change account information for dalle 2? stuff like name, email, phone, etc.
The best way to do this would be by messaging the OpenAI support :)
-> support@openai.com
You could also try reaching out to the live support on beta.openai.com
The live chat is available on Monday again!
it's a python program somewhere on github
Thanks a lot
I can't remember the name, I'm looking for it
Can I change my dall-e account pfp?
If you registered via. a google account you could try and change your google pfp, yeah 
Besides though there is currently no way to do so (that I know of at least).
ah xD alright alright
!!!!
ty
also, is it ok to have multiple open-ai accounts or are they against it?
Keep at one account :)
You can lose your free credits afaik
Is there any way to change the resolution or aspect ratio of an image?