#community-help
1 messages · Page 10 of 1
the download button doesnt include the new one until i click accept. then I lose the other one.
hmmmm
then yes, screenshot it I guess lol
all those small variations will be saved in your prompt history, so just download the one you like the most and you can edit back in the other one in paint or photoshop
ok all good
my prompt history just updated now for some reason. I see them all there! thanks for the help anyways
Wait. Did i just break a rule by extending a cosplay photograph of myself with the new feature? I was just experimenting with my own art (photography and digital paintings)
To see how Dalle would extend them.
Please forgive me if i did break a rule.
Maybe it detected your face? If so, no worries:
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
No it didn't. But in question and answer there's a q about uploading ones own face. Which i technically did.
The pic got edited by Dalle (sloppy but impressive) and i shared it here. But i removed it as well, scared for breaking rules.
As long as you're following the content policy then you should be fine 🙂 Check #old-server-rules under Rule 0 for a link to the content policy.
When Outpainting, If I cancel instead of accepting, will my credit still be spent on that generation I cancelled?
Yes, it will still be spent
A fast question - how do I put the Dalle-2 bot into my own channel/ server?
It’s still not working
Hi! Can you describe what's happening?
You can only generate images via the website; labs.openai.com
Yes I know thanks no images appear
That was a reply to another user 😅
What browser are you using?
Safari it’s been reported but still not working
I highly suggest using other browsers for now.
I'm on Firefox and it works fine
Have trouble accessing keychain all saved on iPad long story do you think it will ever work on safari
Let me try with safari now, hold on.
Thanks
Works fine for me! Have you tried clearing your cache?
Nope didn’t work thanks anyway
İs it enough to say " created with AI " for Commerical Use
Worked with Firefox thank you
<:_:979610619286274058> — The #gpt-models channel has been updated.
Added information about the server's reporting systems, more community-made guides, and more FAQs. Be sure to check it out!
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
@weary tinsel
Correct. That's the official statement from OpenAI at this time :). Just don't lie about it if asked.
Yea sure i will say it on my collection descraption ✌️
I'm trying to extend an image I have, but is there a way to make the generation frame larger/the image smaller? If so, how?
Dall•E's generation frame will always be 1024x1024px, at least for the time being.
If you simply want to reduce the size of an image, you may either use the !pad
function in the #dall-e-bot channel, or an outside image editing software such as Photoshop, Gimp, etc.
You could also try scaling the picture after uploading it to the Dall•E editor and see if that works for you 
Can Dall-E 2 remove the background of a picture and make it transparent?
As of now this is not possible, no. :)
You could carefully try and erase the area that you wish to delete with the eraser tool;
however, I recommend you to use an external image editing software such as Photoshop, Gimp, or anything similar for this process. 
Good morning! Thank you to all you volunteers, staying with us through the night.. Newbie question.. : How do I get to the place where my images are? And how do I just go to that instead of channels? Not that I don't enjoy learning.. Mostly been reading the user guides. Wonderfully put together, may I say.
Also, I have a bunch of images I do not want.. any way to delete them?
Anyone?
Your saved images will be in the Dall E 2 website under "my collection" you can also hit the little clock on the top right of the DE2 homepage to see recent generations
Thank you so much! I am still not clear where stuff is. I appreciate your help.
are weapons like guns and such banned on battleships? because I made one and accidently got some and I don't really know the policy of it
Yes, weapons are banned. Be sure to 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/
I read that, there's nothing about weapons. But thanks for the answer
hey! how to get better results in outpainting? i outpaint into one direction only and it looks FAKE
no smooth transition between og pciture and newly added outpaint. really frustating
make sure that some of the new generation frame overlaps with the old one
Weaponry would fall under the violence category.
How do you use the edit image feature? I don't seem to be getting the best results. Is it possible to do it without erasing parts?
uh, why does "Mohammad teaching unbelievers in Mecca" break the content policy?
is it the word "Mohammad" or "unbelievers"
done but still no transition 😦
Is it possible to upload a picture and, for instance, add "realistic" to the description so I get a realistic version of the picture when I click "generate variations"?
you'll have to describe what you want the photo to be of I believe, but that might work
Would uploading a part of a human face (eg. the forehead or hair) count as a violation of the Content Policy?
Asking this because I’m still unsure if an image I uploaded and made variations of counts as a violation of the human faces policy 😬
I'm confused, how is pepe trying to save humanity considered a policy violations:(
@keen hamlet It's funny because I did a very hard one yesterday, the coffee mug in form of a potato. This was surprisingly difficult, because the concepts of potato and mug would usually appear separated.
I used something I called "Minimalist prompting" to create the mug and then took that as an example for said strategy in my guide.
(Sorry for the wrong channel, but the question was originally asked here.)
Thanks -- when helping users in this way, please give the prompt that worked for you so they can try it themselves.
I will remember it!
Since Pepe was associated with alt-right groups for a while,
OpenAI restricts prompts including the name 'Pepe' to
reduce the risk of racist imagery being generated. 
It can be hard to tell what the content filter will do in that case.
I'd reccomand you to avoid doing things that could result in accidental account suspensions, better safe than sorry :)
Yeah, I’ll be more careful next time
What does the bot's composite command do?
The composite feature is kind of defunct at this point, I believe. If I recall, it has the same functionality as the current Outpainting feature available on Desktops using DALL E 2.
@short salmon i have a doubt, how can i sell dale pictures ?
I don't know how to answer that question. There are many ways to sell artwork.
can you mention some
can you mention some
I'd suggest looking online for tips on selling artwork. Remember, if you do sell your DALL E 2 generations, to follow the Content Policy, the Terms of Service, and ensure you disclose the role of AI, and don't deceive anyone into believing you created the artwork yourself
do you know anay people sell dalle images
do you know any places people sell dalle images, then i can learn how they are doing the business.
No.
do you know any websites
do you know any websites
We're volunteers here to help people use DALL E 2. We're not experts on selling digital art. Please do your own research. I've told you multiple times I don't know any websites to sell digital artwork.
okay
In the future, please don't keep repeating questions.
Hi, can I generate an image with an specific text? for example if I want to generate an album cover for a band called, let's say, "Avocados", How can i make it to generate that specific text? when I try it generates similar text but not the same exact, like "Avoado" or "Acodo" but not the one I'm looking for
The current model struggles with text quite a bit. You likely won't get a perfect word straight out of a DE2 generation
Got it, thanks
how to make a trail cam image in dalle 2?
Hey I was wondering where did the generate variations of uploaded image functionality go? Or did I dream up it was there
It's still there 🙂
huhhh that's so weird... when I upload an image it takes me straight to the outpaiting editor? :/
It takes me to the outpainting editor when I click 'Edit Image'
ahhh I got it now, I think you have to crop!
if you skip cropping it goes to the editor as well, thanks!
is there any way to look back at my recent history once I close the browser tab?
Yes, for now, your recent generations are available in your browser's history.
thank you
what does this do in the generation frame?
It adjusts your eraser brush size.
rip no idea how to actually get inpainting to work. should i have specified just lantern if I erased the left hand area?
#1015941757218869331 message — scroll down to the bottom, there's a user-made guide for the editor 🙂
I would've prompted "a man holding a lantern", and erased quite a bit around the hand area, even extending to the man's body. Just make sure to give DALL∙E more space to work on than you'd think it needs.
legit! thanks
it even gave his left leg a mini sci-fi lantern
Beautiful! If you need more help with prompting, there's #❌┃prompt-help for you. For outpainting, #inpaint-outpaint is there 🙂
Is it possible to remove the 5 color squares in the bottom right corner of images?
For now, you'll need an external software to remove the watermark.
Hi! Is there a specific keyword to modify the aspect ratio of the output or is this enforced to 1:1 ? Thanks!
In its current version, DALL·E can only produce images in a 1024x1024 square. But OpenAI released the Outpainting feature for easier generations of various aspect ratios. https://openai.com/blog/dall-e-introducing-outpainting
@livid kayak
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
No worries!
dang, Wikipedia links are banned, fair enough
What I said is in a nutshell because I'm too tired to type it all out again. What would happen if I typed "blue footed booby" into the AI? would I get in trouble or would the AI recognize it as the bird I'm talking about? It's a question that edges the line but at the same time I think it's important to ask these questions.
I would link the Wikipedia article on the bird but sadly Wikipedia is a blocked linked which is understandable
DALL∙E's filters may be sensitive, but it also has contextual understanding rather than just a list of blocked words.
Scientific names are well known in dall-e
Can I provide a streamer friend with an image to post on their twitter? They did not ask for it, this would be from me to them
Yes 🙂
Will of course remember to include that it is "generated with OpenAI's DALL·E"
interesting and really good to know, it was something that concerned me a bit the more I thought of it cause someone might've seriously been wanting a picture of the bird from the AI or heck, a picture of anything with something in its name that would be considered inappropriate outside of context.
I gotta go to bed but I'm glad to hear this.
No need.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Is it still ok to ask them to include it?
Of course. You are welcome to include it and talk about it
Spread the word 🙂
I prefer to, so that more people know about the project
Me too. It’s pretty amazing
Hey guys so I'm trying to get dall-e to visualize a song but if i put the artist's name with the song title it seems i get an error saying that it violates the policy. Is there anyway to get the A.I to make it without including the artist's name?
Ex: "Coldest Winter - by Kanye West"
Better to use the lyrics
oh i see then, so should i just like copy and paste the entire song?
The main lyrics you want to visualise
ok sounds good thank you very much Alex!
Thank you! Here's the prompt:
32x32 pixel art of an old grumpy ship captain, white beard, white cap, black coat, yellow background
Is it possible to change the generation frame size? thanks
using the "edit" function you can expand on an image to e.g. add width or height through "generation frames"
but the default operation is always 1024 x 1024, so you're only ever stacking frames of that size
Is this form still active? As in, do you guys still read it?
is there any way for me to share my pictures with on-website Share function that wouldn't involve my friends creating account just to see the link I share them? that is, if that is actually the case?
I have generated quite a few pictures, though mostly variations, and downloading each of them, to then send via SMS or email would take really long. I can't just sign in to my account on mobile device that I can take anywhere with me and show others my generation history either, because generation sessions are stored locally in the browser on each device
please, I would love to know if there's some easier way
i'm sure they will collate data continually and then look into it when DALL-E is next being trained; so it's definitely useful information, but don't expect the model to be updated with new stuff in the short-term
the links you share are public and your friends don't need an account to see them
ohh, thank you so much for explaining. will I be able to see my collection of images that I saved from current session on PC on mobile and see things like variation from, edit from to help me distinguish certain images in history of my generations?
or are those descriptors not available on mobile version of the site/they disappear on both PC/mobile when session is cleared?
I am hoping that those stay if a session is cleared or device changed (the variation from, edit from)
that information will be retained if it's saved in your collection, though you only see "edit from" for example if you hover over it with a mouse as far as i can tell. opening the item from your collection on mobile will show you the original image rather than a prompt (but it won't explicitly say if it's an edit or a variation)
I just tested it and it does seem to show all the information. Possibly mobile website update as of recent?
well, except for the edit/variation - didn't quite notice
sorry for double ping
yeah that's all i meant. no biggie because you can infer whether it's an edit or a variation from context
it would still be cool if mobile was updated to show that info too. it seems you also can't click on original image to show it bigger which is too bad
PC version is a bit ahead - only about edit/variation though
are the original images not saved (in collection view), hence why they can't be shown bigger, and only thumbnail?
I have a problem with buying DALL-E credits, it says "Unable to add a payment method"
i have 12 free credits to use until 28 september, what happens after that?
do i get more?
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Email support@openai.com for assistance with that
Hi, I'm totally new here. How do I generate an image here?
Thanks, I didn't know that. I always thought pepe was a cute frog meme
Go to the dall e 2 website
Hi, I submitted my name and email to the DALL·E Artist Assistance Program Mailing List a couple weeks ago but haven't gotten any response or a form to fill out. Is there someone I can inquire to about this?
hello guys, i need help with something, I ran out of credits but I want to see my previous creation , how can I access those images?
Is it possible to get dall e to shade flat color art?
You guys know a good prompt for cute chibi style animals
I'm not too sure why you haven't gotten a form to fill out, it might have something to do with the date you signed up for the mailing list though. (The program stopped accepting new interests since August 10th)
AFAIK another window will open in the future though, so you can come back every now and then to check 
You can always check your browser history for URLs with either labs.openai.com/e/ or labs.openai.com/
You'll find your old images that you haven't saved to your collection or downloaded via the links!
Additionally, there should also be an icon in the top-right corner. If you click on that you should be able to see a history of your last 50 prompts.
why do I get a warning for these prompts? they seem totally innocent to me
Take a look at this
#gpt-models message
Maybe one of the really helpful guides mentions something about this.
You may also find out more by asking our really helpful community members in #❌┃prompt-help 
Try ultra-wide-angle-shot instead 
So as an example:
A motorbike made out of glass, ultra-wide-angle-shot
or
a photograph of a motorbike made out of glass, ultra-wide-angle-shot
I was confused because I used that part of the prompt before and after the warnings with different subjects
in these cases i did not get a warning
anyway, I will adjust the prompt but it still confuses me
thank you so much
You're welcome, let me know if it works!
In case it should not work I suggest you to give asking in #❌┃prompt-help a try 
it works, but I am afraid I will get a warning again if I change the subject or object again
I just do not want to get a automatic ban, based on this
Use the new Outpainting editor functionality! Move the 'generation frame' to the edge of your image, and be sure to include a piece of the original image within the frame to ensure dalle continues the same aesthetic of your original image as it builds outwards.
do i have to describe the image?
Well, that will only help you. What do you want to see as your image is extended? Tell dalle. You can just generate with an empty prompt field, but you might not get what you want. For images I just want to 'extend' I say something like "the image continues with the same aesthetics" but even that can give mixed results.
thank you!
no worries 🙂 good luck, have fun
thank you@
Your punctuation style is likely making it difficult for DALL-E to understand the context. Using the suggestion already offered will help, but another way of writing it would be just use more natural language: an ultra wide angle shot, a motorbike made of glass.
how do i get it to continue the image in outpaint
isnt it supposed to like make one big image or am i missing it
all im getting is stuff like this
#1015941757218869331 message check the guide at the bottom
You must include a piece of your original image, "overlap", with the new square so dalle knows where to continue from. The canvas itself is not aware of all elements of the canvas, context is only taken from within the 'Generation frame'
Thanks I figured it all out! Check out what i put in dall-e gallery!
It’s the word “shot”. No violent content. Find other ways to say it. Wide angle lens maybe?
Hi! haven't had the opportunity to keep using DALLE since the beta ended, and i dont know how to generate variables from an uploaded image like before
since it gets me to the edit image new space
Are generating old politician's banned?
I keep getting the content policy warning when i mention old politicians and i don't wanna risk getting my account banned
Public figures and celebrities will get flagged. Please check the content policy to avoid further warnings: https://labs.openai.com/policies/content-policy
Understandable
How can I reset or change my password?
The answer is in the FAQs 🙂 #1015941764261089313 message
Thanks
Hi, I am not able to buy credits. Can someone please help?
Hey! What seems to be the error?
Please send us a ModMail and we'll try to help from there. Check the side bar for the ModMail bot and send it a DM.
Sure, I've done. Thank you c:
Hi, I am able to access credits now! Thank you 🙂
I have another concern, in the paid version I am not able to download multiple generations. I can download only one image which I accept..but I want to accept multiple images.
As of now it's only possible to download pictures one by one.
You can downlaod your entire collection though, take a look at this post for more information 
#starred-ideas message
How to I access the new Outpainting feature? What's the link
Here you go: https://labs.openai.com/editor
Thank you
Only available on desktop, this explains so much! I’ve been trying to figure out outpainting on iphone (chrome) for half an hour lol. Thanks
Same with me 🤣
Hopefully it would be optimized for mobile/tablets soon! 🙂
Hi everyone, I wanted to ask you which term to use to get black and white line drawings to be able to color them, and in particular how to get a result with the figure of the character (for example a knight with a sword) whole, avoiding being half head or half legs. Thank you
Ask this in #❌┃prompt-help 🙂
Hello ! What did you type for this render please @verbal crag ? 🙂
Needs more refinement to have it output a glass cube consistently 🙂
Many thanks ! 🙂
i just wanted to know how the credit system works, does my credit get used every time i generate or is it a monthly thing?
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
1 Credit = 1 Generation
thanks a lot, it makes sense why i had to refill the very first day xD
You might want to check out the photorealism guide in #gpt-models, and then head over to #photography channel. If you need more help with your prompts, #❌┃prompt-help is the best place for that.
Hi, is there a way to save outpaintings in collections while in the editor? All I can do is save to disk. Am I missing something or is it not available?
As of the moment, for the beta, you can only save your outpaintings locally. Hopefully we'll get that feature soon, stay tuned!
Might want to post that in #suggestions
Thank you! Appreciate your comment!
Will the 100 free beta credits ever expire?
Exactly one month after you first gained access, I believe.
They're paid credits, so they expire 12 months after receiving them
Well, I still have them now and I got access in June
Thanks!
100 credits were granted to users who participated in the research beta (April 6 - July 20)
does anyone know what would be wrong with this image?
I think dalle hates guns I believe
I tried to generate one and it didn’t make one
So that might be it
It can do water guns though
do you think this'd be ok?
Probably try it maybe
ye it works
even adds the censoring
which is just insane
Hello, I am not sure if this is a bug from the update, but when I try to edit an image, the eraser feature is not available across the entire image. Are there any other users are experiencing this or does anyone know how to resolve?
Below is an example: I cannot erase anywhere in the top half of the image
Try using the pan functionality to make the top of the picture lower down on the screen. I believe the text box is interfering
Hey guys, I want to create an orange with arma and legs made of twigs, doing a certain action. Any tips on how to make the prompt.
What I have tried so far doesn’t work
is it possible to buy credits for DallE with paypal? im currently in a different country and my bank wont let me buy anything online whilst im abroad
can you scroll to see the rest of your images on the list
If you zoom in you can see that this image is made out of 6 individual frames. How could I get rid of that?
Here it's very visible:
Is something violent, like beheading, allowed in Dall e 2?
No.
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/
Not at this time, sorry. Hopefully soon.
Does getting a content violation use up a credit?
I'll take the risk, 99% of what I wanted to do with the AI would get me a content violation
Actually nah, I'll do it on my last credit
If you don’t get a generation, it doesn’t cost a credit
Cool
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
You get 15 a month after the first month as well
I'll probably just wait till the content restrictions lighten up to use the AI
Cause my main goal was to create horrid gore beings
My goal was to create Hitler rapping with Stalin
But hey
Im not mad with Saul Cheesy
Hi, I'm writing my dissertation on analogue computing and AI in the context of visual effects and computer graphics. I'd like to get in contact with a developer if possible. does anyone know if developers come on here and speak to people?
This might be a silly questions but is there a way to gift Dalle credits? Asking for selfish reasons ☺️
not yet, sorry. OpenAI is aware of it from #suggestions though
you can put that in #suggestions - the people with yellow names are OpenAI staff, or you can email support@openai.com
Hi! 🙂 Newbie here! I am just wondering, is it possible to upload a photo to DALL-E and then have it generate a new image based on that + a prompt? For example - upload a photograph of a pet dog and ask DALL-E to generate a painting of the dog in military portrait style?
I just got a "It looks like this request may not follow our content policy." for a prompt but I dont understand why?
same here, i think mine was to cruel. i tried making a screaming heart popping out of a mans chest
I just wanted to see what dmt trip person sitting in cozy apartment would look like
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
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/
another question: anyone knows how this works with copyrights and everything
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
thanks
any suggestions for prompt making?
Why when I’m creating an account it says “This phone number is already linked to the maximum number of accounts. If you’d like to exceed this, contact support@openai.com” and what can I do to make it stop?
Have a look at the community guides in https://discord.com/channels/974519864045756446/1015941757218869331
and choose one about your desired art style.
?
I tried to do some variations from a photo made by Dall-e. Anyone know why?
How long is the beta?
Similar question:
Dall-e generated a face, but won't let me render an angry cloud above it
It's eaten a credit for this boo
Quick question: one of the things I'm wanting to generate for my 3D Art concepts/inspiration are weapons and military vehicles/warships/aircraft (steampunk, modern, fantasy, or otherwise). Would stuff like this be against the content guidelines, or is it perfectly fine to do these kinds of generations?
I tried using " russian tank shooting at a sea monster emerging from the coast" and it didn't let me because it went against the rules?
so anything related to "current politics" is a no go?
i just wanted a model of a russian tank because I was trying a scene from the metro games
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 removed the word "russian" and it worked
like it has nothing to do with the war right now
yeah but still
thats why I'm a bit bumbed about it
Anything that involves war, violence, weapons, and so forth is usually not allowed.
Please read the following Content Policy

That being said, military vehicles can be generated as long as
there is no obvious act of violence, such as an actual combat, etc.
Simply read and understand the Content policy, and you should be fine :)
sure
Why do you need to verify your phone number when signing in?
Sounds good! I wanted to check on that before I did anything military related (wouldn't show any combat, just a concept art style that could be used for model creation), thank you for the clarification! 😄
is it fine to use something created (for personal use) from DALL-E?
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/
alright thanks, I will read it!
Great!
As long as you follow the policies, you have full commercial rights to your generations!
That also includes personal usage 
When I create a new account it says that I’ve reached the limit of accounts that I can have with my phone number and need to contact openai’s email, am I contacting them to extend the limit?
Lol
Currently, every user is only allowed to have 1 account.
Oh ok
Huh?
Is there a way to delete accounts?
If you want to close/delete your account, please contact the OpenAI support.
Do note that account hopping to acquire extra free credits is not allowed and will have consequences :)
E-Mail: support@openai.com
Live Chat: beta.openai.com
Ok awesome thank you
this appers when i click generate, am i the only one? im new in dalle btw
appears*
Yes, this does happen from time to time and is usually caused by server issues.
Bring some patience and check back every now and then.
Take a look at my message above 
This should answer your question as well.
What is the meaning of the 🌍 in #hall-of-fame ?
Hello how can I make a very realistic image with prompts
Who would I contact to arrange interviews for a news piece
I asked it to create an image of an owl wearing glasses, a black-tie and drinking coffee
The results were as impressive as always, but the third image is a creepy woman looking at you with a person next to hear holding a cup of (probably) coffee
Can someone explain, why this happened
And most important: Are those results against the Contest Policy? (No humans, however, I didn't ask for humans)
any way to generate more than 4 options at a time?
would legit buy an unlimited sub if offered
dall-e 2 good
Hi, this has probably been notified but has everybody notices an enormous drop of quality these past days? I find myself enable to use the outpainting and generation option for good use so much the results are terrible, like the image weight and diffusion passes were hyper low
Please contact support@openai.com
DE2 is a static model and there have been no changes to the underlying image generation system as far as we know.
That result of something that is far removed from what you asked for is typical when the system is unable to converge on a solution for the goals in your prompt. When you get these, it is helpful if you report them using the report button associated with each image.
okay, thanks for the advice
Many guides to prompt construction are listed in the external guides listed in https://discord.com/channels/974519864045756446/1015941757218869331
The photography and realism guide may be of particular help to you: https://destiny-sherbet-aa9.notion.site/DALL-E-Photography-Realism-Guide-045c50ded4c54fdfb52f55acaf613f1c
Hmmmm. I haven't got a clue. I will investigate this mystery.
Turns out that globe was a side effect of a setting that has now been turned off for the server. It may show up on a few older posts, but should not be appearing on new ones going forward. Safe to just ignore it.
any tricks to make a subject face the right side of the camera?
like my pfp, just need a gnome facing slightly right or left of the camera
mb 45 degree turn
Hey, how come new users are getting 50 credits when they sign up, and I only got 25?
You mean like a profile side view?
Not all the way to profile. Halfway between face forward and turned 90 degrees to either side
👆
Have you tried something like 45 degree pose/profile/angle ?
Does generating variations of a result use a credit?
Yes.
Effectively any time DALLE has to generate any material whatsoever, it costs a credit, sadly.
We are all volunteers here and have no way to answer this type of question about your credits. Please send mail to support@openai.com for help with this.
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Ok thx
can't download my photos to my phone
apparently its against content policy to write "slavoj zizek" on the prompts, not sure why
Hello. I want to use outpaint but the options for editing are grayed out. anyone know what the cause might be?
You can't replicate or generate images based on the appeareance of a Real Life Celebrity, it's on the policies
How do i get Dall E to generate variations of an uploaded photo? I can only see how to erase parts of the image
Upload your image from the standard text prompt page, not from within the new Outpainting editor. Uploading an image in the standard way, you will be able to choose "Generate variations" or "edit" once uploaded and cropped.
I've been skipping the crop, is that causing my problem?
Yes, maybe so. Since that's probably taking you to the Outpainting editor.
It's being glitchy in general, for instance won't let me erase the top 25% of my images
Yep, skipping the crop was the problem.
Is there an equivalent to Midjourney's Upscale here?
No - but you can expand your generations through https://labs.openai.com/editor
I am requesting images with animals in them. Every single one has a deformed animal; they're very disturbing. I've reported them, but I'm seeking an explanation.
I can try to help you with that, as you've mentioned the images are disturbing, do send us a ModMail with the images attached and I'll try to help you from there.
Can I get some help with understanding how to use the edit/erase feature? When I erase part of the image is it supposed to fill in with the prompt I give it?
That's right. Make sure you erase enough of the image to facilitate whatever you're asking for.
Oh, just to clarify, you will have to press "generate" again for it to fill in the parts you've erased. You can also, like me, use the erase feature as a kind of 'content-aware fill' for cleaning up images.
I am not seeing a way to attach images in modmail.
Right click the bot -> Press 'Message' — then refer to the image below.
how do credits work
is it per photo?
No, it is per generation. 1 credit = 1 generation.
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Can you count how many generations are in your account through the generation history?
I'm on my phone
Turn your phone to its side, on landscape mode, and that button should appear.
Oh, no no, go back to the page where you can generate prompts. Exit the account info page.
Hey i have 50 credits now if i use them all today then will i get new credits tomorrow?
next month u get 15 free
Oh i have to wait a month to get more 50 credits
no u will have 15 credits
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
The generation history icon should be under your profile picture when on mobile. I think you scrolled down, judging by the scroll bar.
also just wasted credits trying to look for it
nope didn't see anything down there than other pictures that always show up
it's whatever I'll just wait tell next month I'll be ok
is there a way to generate images larger than 1024x1024?
Not at this time. But you can use outpainting for that
ok it would appear that another AI generator is capable of massive images before this one, thank you for your answer.
Is there a list of words that are banned? For instance, "Daenerys Targaryen burning a slave master's tokar in Astapor" isn't allowed, I'm assuming because of the word slave right?
There isn't an official list
Oh okay, it says you can get banned if you keep putting in banned words though, how many times does that take?
That is also not explicitly stated
But I've also never seen someone who's asked about getting unbanned mention any difficulties if it's not a genuine attempt at breaking rules
As 'OMGNinja' already said, there is currently no list of banned words.
Please read the Content policy to get a basic understanding of what is and is not allowed. 
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/
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
Thanks!
Anyone else have the issue of you have to DL every single image because it will not display on the webpage? And this of course means that I can not see any other image on the site either. Using a mac (catalina) with safari. Any help appreciated, TIA.
Is it ok to upload the generated art on a social media like Instagram? Obviously properly referring to Dall-e of course.
I'd stay away from Safari if you're using DALL∙E. Been seeing lots of issues with safari. I'm on macOS Monterey and currently using Firefox.
Yep, please refer to the links below for clarifications:
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 checked the “sharing and publication policy” and I do not quite understand this point: “Attribute the content to your name or your company.” Care to explain? Thanks.
Is it possible to remove this on the corner if I'd like to choose one generated pic as online portrait
For policy inquiries, please direct your concerns to support@openai.com. Though in general, OpenAI's goal regarding the sharing is to avoid deception. Meaning, when someone asks who made the art, people should mention the AI.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
You can remove the watermark with third party softwares.
👍
Thanks for letting me know. Any recommended third part softwares which easily to use without destroy the other part of the pic
In the interest of server rules instead, would referring to a Instagram account solely focused on sharing dall-e art be braking the “no self promotion” rule?
Don't bother. I got it. Thanks to all.
Yes, as that may also break Rule 5. We're currently only whitelisting links for DALL∙E specific guides.
Is it normal to sometimes get expletives in your images for no reason?
Those are rare, but can totally happen. Please report the images so it reaches the team.
Ah ok, just reported it, it was a little funny though
does support answer? support@openai.com right?
Yes, they do. That's correct
Just wanted to ask, does verifying that I have a dalle 2 account for getting access to this server link that account to this discord account?
Also, do the 15 tokens per month get added even if my current balance is not 0?
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
I don't believe it does. Maybe someone else can chime in and confirm
In simple terms it brings your credits up to 15 if less than 15 every month
I’m pretty sure dalle uses free credits before paid credits but I’m not sure
Does anyone know if the makers of Dall-E are exploring a 3D AI generator? or one for music?
You can use CLIP-Mesh ai for 3d generated models
thanks, I'll have a look! :)
It’s pretty new and open source so if you don’t have experience with PyTorch models you might have trouble
I'll figure it out 😅
Hi, I'm an author and own the rights to the book cover art I had made. I tried mutating one of the base cover art I had for one of my books, but I got a prompt telling me it violated Dall-e policy. Is there a way to inform Dall-e that I own the rights? Or does the system itself just filter that out automatically?
Hey! Does your cover art contain faces or other objects or body parts that could be considered nsfw or violent?
seriously... I wasn't even allowed to write cocktail 😅 Dall-E is quite harsh
The word cocktail by itself works just fine, what was the full prompt you were trying to generate? 
something like "cat and dog making a cocktail at a bar, logo" or something
That prompt works just fine as well.
Let me know if you face any problems in the future :)
huh... maybe it was a different program? 🤔 its been a few days
how do you create variations of an uploaded image? it seems you can only edit/add to an uploaded image
I think my problem was that I wrote "catdog" 😅
there are 3 little dots in the top right of the image... click on it and itll say you can create variations
even "cat and dog" isnt working
As HitchHacker has said, Dall·E may identify something that goes against the Content Policy.
Read through it and see if you can spot anything that may be the cause of this 
😩
Having a look at it now :)
drug use? alcohol... drug?
maybe they don't like animals and alcohol together or something lol
The word Cocktails seems to be the issue. Replacing it with 'cocktail' worked for me.
Perhaps change it to 'drinks' or something similar, or, keep the word in singular.
i did change it to drinks and it worked.... weird that cocktail works though 😅
Weird indeed...
Do you have any other questions? :)
not for now, thank you 🥰
ah, so it might be something else. Thank you!
Simply upload your image and click on the button Generate Variations 
i'll check. ty
ohh i see, i had to crop it first, it doesn't allow variations if you skip cropping. thanks!
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/
Check the message above for a direct link to the Content policy :)
how do you use commas correctly?
Ene has some explanation regarding prompt structures, you can find Ene's guide in #📚┃resources 
@torpid glade — Aspect ratio is currently only 1:1 for DALL-E, however, you can use the outpainting feature to go beyond your generations. https://openai.com/blog/dall-e-introducing-outpainting/
oh sorry... thank you! :)
How can I refer someone to join Dall E?
No need to be sorry, hope it'll help you! :)
There currently is no referral program, no. 
will we be able to buy custom amounts of credits? currently to buy credits we have to add/subtract fixed *115 credits
it would be so much comfier to buy any amount of credits we wish
As of now, OpenAI has not released any information on this topic.
In case you have any suggestions, such as the possibility to purchase a certain amount of credits,
feel free to post them in the channel #suggestions 
I see. Thank you, I will try to post that there!
A friend of mine just got 50 free credits to try the new outpainting feature. Were those credits only for certain people? We signed up at the exact same time so just making sure.
Everyone should get 50 upon first log in
Yes. This was an additional 50 credits. The email they received said they got the additional credits to try the new outpainting feature.
Hmmm
Hello:
I have a policy violation for this prompt and no matter how i write it, i still have the issue. Can someone have an idea for the reason why?
"A man and a woman with curly dark hair, holding hands and lying in the grass while looking at the night sky and the Milky Way"
I'm not speaking English 🤣
Does* anyone have any idea why?
People who tried outpainting during an outage period were given some credits to compensate for the trouble
Damn. 50 extra credits for being on at the right time! Missed the boat on that one for sure.
Just needed some punctuation. Try #❌┃prompt-help for these kind of questions, but this worked. https://labs.openai.com/s/dPatEfDDKTgy3Ajm7lCmN1GX
Thank you 🙂
Where can I find documentation about switches I can use with DALL-E?
wdym by switches?
You won't find switches here. You can look in #📚┃resources or #❌┃prompt-help
Like in MidJourney. "Cat sitting on roof --s 1 --c 6" <- those after -- are switches
Ok, thanks. I try to look
is there a way to pay with paypal?
Unfortunately, no
Is Dalle working? It's failing every time I prompt, even when changing the prompt entirely
The site is timing out when I try to load it. So it's not just you.
Site is down for me right now too
OpenAI is looking into it.
Hi thanks, but nope after many tries and other users doing the same: the quality of Dall-E has drastically lowered. This has nothing to do with the model but the diffusion passes on a generation that seem to have been lowered
I haven't really noticed anything to be honest.
Hey everyone, I'm looking to inpainting a whole folder of images. Is there a way to submit a mask through the API?
The API is currently private.
DALL·E is not yet available through an API, and users are permitted to access DALL·E exclusively via our public-facing website. We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use. We will continue to evaluate how we release this technology as we learn more during the beta.
Btw Im a huge fan ❤️
Thank you!! (Sent ya a DM 🙂 )
What prompts should I use if I want to create a realistic mythological character image? pls help me
please use #❌┃prompt-help 🙂
Hi I'm new, I just needed to know, is Dall E page we are using right now is Dall E 2?
Hey all, I've been away after only dabbling with DallE for less than an hour...Has DallE2 gone public yet or is it still beta? Is there a way I can send an invite to a friend? Thanks.
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/
See the above
DALL E 2 is still in beta. Currently, the only way to join the beta is to join the waitlist (https://labs.openai.com/waitlist)
Hi there, do we own the images Dall-E creates for us as it's our prompt or not? It's fine if not, I'm just genuinely interested.
Take a look at the Terms of Service, it has all the answers regarding that
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 wonder what this means for outpainting. For instance, I recently shared an image I on this Discord that I generated. Then, someone else on the Discord decided to outpaint it a little more than I did because they wanted it to be a better aspect ratio for their computer background. Who would own the new image? Originally, I made it and created the prompt, shared the prompt with them, and then they expanded it. I don't know if there's a definitive answer, but it does make me wonder.
With DALLE Mini, a thread came up with a bunch of similar looking creatures if you asked it to generate "Crungus". This thing just... doesn't exist if you search it normally. Has anything like unusual like that been discovered with DALLE 2 yet?
Is it allowed to use images created in doll-e 2 for commercial use?
Yes, you have full commercial rights to your images
Is a text prompt required to generate a zoomed out version of a picture? Referencing something like album cover that I dont quite know how to describe by text
Someone know how to make dalle understand how comic works? like panels and such?
I've been trying but I got... Mixed results
Its best practice to describe the square you're in the text prompt. It's not technically necessary, but personally I wouldn't just enter a space in it. I don't know that the overlapped area would be enough information.
Try #❌┃prompt-help
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/
Question on commercial usage; does the image need to keep the dall-2 watermark (rainbow logo in bottom right corner)? Or is it safe to remove?
Btw this is the most incredible piece of technology I’ve ever used. Literally feels like magic, something that only comes by rarely in one’s lifetime.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL∙E 2.
So, yes it is safe to remove?
Attribution makes sense, I’m just using a dark image and the rainbow sticks out lol
Can I pick your brain, in a sense of showing you a picture and then you describe it. So I can understand the avenue of thought I should have when doing this. I don't want to tunnel vision like I tend to normally do
Hi all -- I have a weird question that I couldn't find an answer for with my searches, so far. This might be due to poor choices of keywords. Anyway the question is this:
Is there a way to know if a particular thing exists in the DALL-E 2 training data, without burning a bunch of credits testing for it? My attempts to use "bacteriophage" produced interesting results, but nothing like what is seen in electron microscope images. Doing a search on wikipedia (since I can't give a link) will show you what I mean. A really good structure ref is the SVG image by Adenosine, which can be found part way down the page.
Is there a dataset test call, of some type, that doesn't chew up credits?
The only way to test for what is availble in training data is to ask directly using a very basic prompt such as "X" to determine what it knows about X, if anything.
so burning credits?
Check your friend requests, we can do that in DMs
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Yes, testing in this way spends one credit.
ok - thanks
Can we have cheaper credits?
Hey so. im just trying to edit a photo that DALLE created for my book, and have a generated face stay the same but it wont let me edit the photo and as far as i can tell the prompt is fine as far as flagged words
@help
photo and prompt in question "Upper body front view of a young man in tattered sorcerer robes with a desert backdrop. Portraiture Painting by Norman Rockwell"
i mask everything but the face and try to get something new and it wont let me
and im of course not trying to get banned for something
Consuder asking in #inpaint-outpaint which handles this in detail. Uploading the realistic face is a common problem.
Are historical figures like George Washington allowed to be posted here? How long does a public person need to be dead for?
No clear guideline on that, but 200 years seems safe. Concern is particularly with faking things. I did George Washington crossing the Delaware Parkway once.
Is there a way to create a collection folder to categorize our images?
Not directly on the server. This feature has been suggested in the past.
Ah, will that be something you guys will maybe look into?
i did george washington meeting ramses the great in a white castle
I need to see this. Be sure to post it in an appropriate gallery!
We are all volunteers here and have no direct connection with OAI. Consider posting the idea in #suggestions -- OAI does look at that channel.
Oh ok thanks will do!
I have a dumb question. Are DALL-E and DALL-E2 like kinda the same? Just like a different version, but if the webpage says DALL-E, then I'm probably like, using DALL-E2 ?
I did say it was a dumb question. 🙂
I just haven't seen anything that specifically describes the difference between DALL-E and DALL-E2.
@river cargo
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/
(Thanks, whoever did that; can I also ask what command you used so I don't have to ping people next time?)
It was generated by a moderator command we have for this common question. 🙂 Just ignore the person behind the curtain.
COOL! Thank you!
normally i use this on my computer but the thought just came on my phone so i took a quick screenshot, am i using Dall-e or Dall-e 2 ? i believe it’s dalle-2 but it’s odd that the title is just dalle instead of dalle-2 when on their site it is called dalle-2 & says they have dalle in open beta, just a bit of a weird understanding for the way the words are structured ig
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/
@spring osprey — the background will be challenging, to change the background, you'd need to select the whole thing, and describe what you want to see. Adding an artist or an art style similar to your goal will help.
ok i will try thanks!
also is there any way to make the camera go a bit further?
like "Outpainting" ? is that the correct term?
Yes, for your case, go to https://labs.openai.com/editor, upload your image, and a resize feature will show up.
ok, ty
Well, dalle is pretty amazing, but is posible to make this image more "clean"? For example the characters looks good and with clear strokes, but when looking to the top part the tree looks so diferent drown and painted, more like oil painting in stead of "cartoon realism", dont know how to call it
Yes, you'll have to find the perfect prompt for your goal. There are many guides in #📚┃resources for you to check, you should find a useful tip in one of them. If not, you can always check #❌┃tips-n-tricks and #top-tips-and-tricks.
ok i will check resources, ty
Hey. is there data on how unique a generated image including its variations is? Is there data on the probability of the AI creating the exact same picture twice?
exact same picture? almost impossible. very similar pictures? depending on the prompt it can happen, especially if there's limited training data for the subject in question. but you can run a prompt 10 or 100 times and you'll never get the "same" result, unless your prompt is something like "just a totally black empty png"
ok thank you
Regarding copyright: Do I own all of the variations generated based on the initial creation?
Normal copyright rules would apply. If you take another artists work and just change a few pixels, that would probably be a violation of copyright law and just plain uncool 🙂
Derivative work rules apply
Best youtube video for getting the most out of this? I tried a long discription and ended up getting a child like drawing?
See the #📚┃resources channel
External links are not allowed here
What is the question?
Thank you.I´m talking about the DALL-E2-generated variations of my own creations. Read something on the internet and am a bit confused
To be honest I have looked for tutorials myself in the past and there is hardly anything better than the selection of guides you find in #📚┃resources
Yeah, I read one or two and they immediately made a big difference. Still hard trying to pinpoint what I want... but it is AI so not like I expected something perfect for me the first time. I do see how you can burn through your first 50 runs real fast though.
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/
The content policy details rights. You have full commercial use, but do read the details there @sick mango
Why is my gender identity a banned term? It feels super invalidating tbh
Log it in #bugs-and-issues it may be being suppressed so that people don't use it as hate speech.
I will. All I wanna do is make femboy art of boys in dresses and stuff
Hi there. As a non binary person I can assure you they mean no harm by it and probably put it in the filter to avoid it being used as a hate thing. Just use the full thing of feminine boy or feminine man for now and you’ll get your result. Do log it in bugs though please 🙂
Not exactly what I was thinking (less Caillou and more SFW Astolfo) but ye I did
Got it. Just showing that you can word it so it works.
Thank you I'll try messing around later to see if I can get it to do the socks and painted nails and such.
Unrelated: do Tony Soprano and Danny Devito count as public figures? Devito might but Tony Soprano is fictional
Does anyone have a good prompt or Tag to make dalle design a useful tattoo template ?
#❌┃prompt-help is the place to try. If you give them more details.
What hashtag is used on Instagram for DALL-E works?
hello, what should I add to my prompt to make the picture more zoomed out, the subject is always near the edges
or is it required to do inpainting
“Wide-angle” or “full shot” sometimes helps, but inpainting is the best method thus far
Sorry I’m not sure what you mean here
Usually just #dalle or #dalle2 but #aiart has some as well
How long does it take for invitations to be approved? My friend has been waiting for 1 day. It only took me 4 hours.
i was interested in popping out some images to support one of my assessments im doing but the requirement for the images is that its either copyright free or i own the image rights and i'm pretty bad at reading legal jargon in terms of services and im just wondering what the policy would be for that with Dall-e? not sure if this would be the appropriate place to ask this specific question
it took me a week and i think it can take even longer
👌Thx bro <3
Thanx! 👍
Hey I generated a texture is it safe to say we can use these in our 3D modeling projects?
Absolutely
Alright what's the license schema on this?
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/
Check the Terms of Service
Will do, thank you.
Very useful for textures! Saves me hunting through texture sites. I wonder if it will generate normal/UV and bump maps too, I'm going to try.
Does all the generated art you create disappear unless you download it? I clicked My Collection and didn't see anything and had to hit the back button because I thought I lost the art.
It should all be available through your browser history, if you didnt save them to your collection
Thanks
In order for them to appear in your collection, you'll need to press "Save"
Do we have to pay to remove the DALLE watermark? I obviously don't want to have that on my textures.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Sweet, thanks jpc.
You can use an external photo editor like Photoshop to remove the watermark, or you can use the "Inspect" feature of your browser, to download the image without the watermark. However, that image will be in WEBP format, so you'll need to convert it
You'll see DALL-E in my next indie game project. 🙂 No problem.
Appreciate it. Is there a limit on space for collections? I would assume so
10,000 images
Ez ffmpeg cli. Not an issue to me to get webp over.
This might be of interest to you https://twitter.com/IRCSS/status/1554088473434509313?s=20&t=ZxEY0lG_o5Q73kZ6DJCXrQ
Oooh yeah, I've got photogeo on my iPhone, I love it. Saves me a ton of time!!!
clay model, subdivide, use instant meshes to clean up the geo, its magic, I won't ever go back to per quad modeling since I discovered the generative way.
Hi all, newly accepted to Dall-E here, two qs: (1) is it possible to generate a 16x9 image and (2) any way to get higher quality than 1024px?
I read an article today that claimed Dall-E has finally approved the usage of faces in its app. But when I read the faq here, thats not true. Thats kinda weird that they would write that.
I am concerned with the publish option of Dall-E. If I publish my images, could someone just copy it and use it? From what I read we are allowed to use our own images in our projects commercial or not, (don't think we own the copyright), but how do we protect our gallery from being "stolen". Any thoughts?
Is there any way to make the same character appear on different places? Like to generate a series of images where the same character/s do different things.
you could try the variation option, not sure if it will keep the original character
I wish the faq were a bit more extensive, there are so many things I want to know. For example, the way I interpret it about the watermark: you can remove the watermark but you can't claim you draw this yourself. You can however just not say that it was created with Dalle. Isn't that again deception?
ok I found this, it may answer some of my questions:
Disclose the role of AI.
You must clearly indicate that images are AI-generated - or which portions of them are - by attributing to OpenAI when sharing, whether in public or private.
You may post these images to social media. Please refer to our Sharing and Publication Policy for further details.
Hi — I have a little events app that currently pulls event images from royalty free sites. Would be cool to have users generate images using dalle— any plans for such integrations?
DALL·E is not yet available through an API, and users are permitted to access DALL·E exclusively via our public-facing website. We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use. We will continue to evaluate how we release this technology as we learn more during the beta.
Currently there is no way to create "an asset" in DALLE and store that asset (character or object) for ongoing use in multiple images, like an asset in a game development system. Even if you specify many details of the appearance, there are some random aspects to the generation process. This results in variations that are very difficult to control. You can get a somewhat consistent art look for a series by developing a set of reproducible art style tags, then apply those style cues to various changing subjects using a consistent prompt structure: [subject][style].
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Thanks for the info!!
is https://www.dalledictionary.com/ going to be accessible again? Was looking for a resource with Artists and Styles with examples and it looks like that might be the place
hey, I lost a credit when veryfing my accoun on the dalle discord
is there a way to get it back?
The site is still accessible at https://dictionairi.notion.site/, the domain https://www.dalledictionary.com/ is/was a redirect domain
The verification process allows you to use any prompt you want, you just have to append the code given to you
oh, well I made a prompt and verified it just took my key
I was wondering if you could change the aspect ratio of your creations?
All gens are 1024x1024 but you can expand borders via outpainint. See #inpaint-outpaint
tysm
i verified but it still took a credit for the prompt needed? i thought it didnt take one
Sweet thanks!
Hey all, is there some sort of prompt generator/creator that you all use? Like something that you would pass a character and scene to and it would generate some prompts to use?
There is currently nothing like that in DE2.
Why can i not erase anything in the top 1/4 of my image?
Try panning so the image isnt so high up, I believe the text box interferes
hay, what is the profit fo me if i boost this server ?
Pride and accomplishment.
Why is every image I generate an oil painting? I can't seem to get away from this, even when explicitly stating digital art, photograph, etc.
can I see an example
quick question, how do you expand a image. for example i have this photo i want to expand but every time i add it it only animates it. thank you😊
Check our the #📚┃resources channel! There are guides that caters to every styles. There's also even a Photorealism guide.
Currently, no way to animate with DALL∙E. However, you can expand your generations with the Editor: https://labs.openai.com/editor
More info here: https://help.openai.com/en/articles/6516417-dall-e-editor-guide
Any advice on how to make the photos come out this way? My generations don't come out so good
Have a look through #📚┃resources; there may be a guide that covers this art-style, or something comparable!
Additionally, try asking in #❌┃prompt-help; perhaps someone in the community can help 
How can I get portraits and faces to look more human-like? I like to create images with people, but they always looks kind of smeared, like in this one. My prompt was:
"Beautiful elegant woman with wavy hair, in a black dress, surrounded by mirrors. Film noir style, cinematic, realistic."
dall-e doesnt know what dynamite is... ???
#❌┃prompt-help has some great people in there. Usually you start with a portrait and then use the outpainting editor to build out around it. Either that or there are some apps that fix faces post AI generation. ask in #ai-discussions or see #📚┃resources
Hi everybody, I try to generate a realistic Claymore Sword but it always not in full size on the image. Do you have some advices ?
DALL-E can generate images of smoking, but if you try to upload or edit some it will give a warning.
bruh
it's a known issue that dall-e can't discern realistic faces you upload from its own.
I actually think it is the cigar ^^
possibly. let me try. I've been hit a few times with paintings of faces
I just did a similar test, edited out the cigar, and got the same violation. not sure. Drop this into the #bugs-and-issues channel please @neon prawn
Those images are most likely not from Dalle 2
But from another AI named Midjourney.
This is my interpretation based off the style the images have, i could 100% be wrong though
Are free credits used before purchased credits, when both exist?
how can i get a more 'zoomed out' result?
but i'm getting more zoomed in stuff like
'far view' and 'wide angle' actually seem to make it even more zoomed in
I looked through the resources and tried "detailed", "cinematic movie photography", "4K/8K", "photo of ...", etc.
All of my images came out the same as yours. Even the simplest "woman holding flowers" looked like a crappy oil painting. Not sure if I need to tell the generator that the images don't fit my description to train the model in my favor or not, but I have to say I'm quite disappointed thus far with every attempt I've made.
There's got to be something I'm doing wrong, right? And the faces are distorted too, like not even a coherent image.
Is there a way to use DALL.E to generate an image in a style I uploaded?
Example: I upload a set of icons and ask it to generate a new icon in the same style
Yes 🙂
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Not yet
check out #📚┃resources and ask in #❌┃prompt-help . Note - images from other AIs are not allowed in channels other than #ai-discussions
hello, I had a question if getting variations of the photo also use up credit
Yes it does
does anyone else is having issue to load the generated images?
here's what I see in the console (attachment)
I'm using chrome btw
I had that last night. I waited like a minute and refreshed then no issues
hi, i am building a clothing brand about drug i have read the terms of use and politicy... that means y cant do any desing on that way?
Idk if this has been asked yet, but is there an ETA for when dalle 2 fully releases?
Not at this time
I would definitely use the content policy as your guide. It states you stand to lose the use of any generations that violate those terms.
Quick questions. The announcement says: "DALL·E 2 can take an image and create different
variations of it inspired by the original."
How do I do that? When I "upload an image to edit", I can't figure out how to use it as a base for variations.
Check #📚┃resources for a video on how this works 🙂
thanks!
Hi, every month when you get 15 credits, are credits from last month deleted or remain?
Also, at the beginning of the second month, are the 50 credits from the first month deleted or remain?
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Is there anyone or any company offering tutoring services? I am a volleyball coach and engineer and I have a ton of digital assets that I want to experiment using them with Dalle and I know what I want to do but have no clue how to get started. I would like to pay someone experienced to meet with me one hour a week or so and help me think through my challenges.
Hi. I'm back to ask about the "variations inspired by the original" ability that is in the announcement. I've attached a quick screenshot of the part I'm interested in.
At @broken wedge's suggestion, I checked out the #📚┃resources channel. I've looked at everything and feel like I still don't know how I'm supposed to do this. Am I overlooking something?
Sorry you are having problems with variations. There are actually two different variation functions, one when you have created a new set of images and inspect one, and another when you are examining variants in the editor. When inspecting a new image you have created via DE2 (not an upload) the variations button is at the upper right on the desktop interface.
Here is a good video on the operation of the edit functions, including the variations function in that: https://www.youtube.com/watch?v=juZqvZGKVi0
• Website: https://labs.openai.com/editor
• Background-Music: https://www.youtube.com/watch?v=SLK3SCwwax0
• I hope this helps some of you out :)
This is the output I got from variations on the image I posted above.
Ok, so lets say I want to replicate what is done in the screenshot I provided. The image on the left is "Girl with a Pearl Earring" by Johannes Vermeer. So, First I upload the image... then I... ? I'm not trying to outpaint or inpaint, just make a variation on what I've uploaded.
I gave a wrong answer here! Correct answer is below, with procedure. DE2 does not currently have the feature you are requesting. If you upload an image, you can edit it using the edit functions. Take a look at the video above that I just posted if you have not already done so. Editing lets you extend or alter parts of an image, but it needs some part of the original to be retained to obtain style information. I am not recognizing the layout in your upload.
It is in the announcement. Here https://openai.com/dall-e-2/
It says "DALL·E 2 can take an image and create different
variations of it inspired by the original."
The sample in the announcement is showing the variation feature for new gens, as I showed in my sample above.
All of the examples on the left are famous paintings, not new gens.
Perhaps someone else may be able to help you with this. I must be missing something.
Ok, thank you for your time, I appreciate it.
I will try testing some things, thank you for raising the issue. One problem with the example here is that DE2 will reject uploads of realistic faces. So the sample portrait being used might throw an error on upload anyway. But lets see if someone else has a more clear understanding.
Whoops...
There is a portion of the image "T**s on Cherry Branch" by Ohara Koson as the fifth example, that wouldn't involve any faces. Also "A Sunday Afternoon on the Island of La Grande Jatte" by Georges Seurat (the third example) also probably doesn't flag for any faces.
I just tested the upload function with the Vermeer painting and was able to upload it, but I can't imagine how the variation gens were done if that is the process used. (Better answer on procedure is later in this discussion.)
This illustrates the edit function I am familiar with to extend the uploaded image, here adding something in the style of Picasso. This is what you will get if you use the "crop" option after the upload. To get the variations button you need to use the "skip cropping" option.
Variants of this are viewable using the small control beneath the picture.
How come after I uploaded my picture, the update part of the picture cannot be edit. Any way to fix this issue?
upper part
I actually don't do a ton of editing, so hopefully someone else will have a better idea of how the sample was done.
Thank you @keen hamlet.
I'm uploading an image of mine and I'm getting a warning, "It looks like the uploaded image may violate our content policy." The image is my own logo and is not vulgar or harmful of any nature. Any idea what I can do?
The uploader does have some false positives now and then. You can contact support@openai.com for specific help with this.
Awesome. Thank you.
I was trying to combine Chloë Grace Moretz and Emma Watson for a picture but it said that it was violating a term of service I was curious of which one
OK, I finally got some help from other volunteers here and now I understand the variation process used in the OAI sample. Here is my version of the variations that I just produced. The sequence to do this is to 1. go to your main home page at labs.openai.com and select upload an image. 2. Click "crop" (if you click "skip cropping" it will not work, which is what was throwing me off). 3. You then will see the generate variations button.
Hen frens any idea of how to include text inside my image ?
Trying to create a logo that holds company name with prompt like "a logo with text xyz in it" but not workings well
DE2 is terrible at text. It may get a few common words right, but that's about it.
Can anyone help me remove the white background in my pfp?
For some reason I can't make the background transparent 😦
You cannot generate faces of public figures
Got it, seems to be the same on every AI image tools, anyway dalle2 looks awesome
NVM
Is there any way to reduce the size in the generation frame from 1024x1024?
I am in a multiple prompt picture and I dont want to click out or do something to mess it up lol
is there any typical way that people fix groups of peoples faces? I often get the exact vibe I'm looking for but the image contains 5+ people and all the faces are a mess
There is really no solution to do it in one take. Only possibility I see is to generate the faces one by one via outpainting.
There is not, you can overlap the frame on your existing image, if that helps
Ohhh gotchya, just sucks because the frames I have already are close to the size of the default frame
Any way to get my collection to re-appear on the site, or is it just gone? Tried clearing cache, no luck.
Looking for verbage assistance on the colorful lines that are flowing throughout space. Prompt started from a colorful abstract interpretation of a lonely stoner freeing his mind at night
I want to fix the frame split lines and continue the colorful lines
Not sure what to suggest. Are you planning to adjust this via inpainting of this image? Not sure what you mean by the "frame split lines".
Correct, going to place the frame with the cut lines in the center of it so i can take the eraser and make it look like it flows better
Not sure what you mean by "the frame with the cut lines in the center of it ".
The colorful gas flowing around the person has a line where you can see its a separated frame
To the left, above, and below him
Here?
Yup that is one of them, and then there is one above and below the guy
So I need a good word in order to describe that colorful dust flowing around to make it mesh better
There I have erased one of the lines, and what I have highlighted in the prompt is what I want to replace in order to describe this dust better
That is how I have set up the prompts to make this image
I would think the eraser tool could reach those parts and just erase them? If you do a minimal erasure and use the same prompt you used before the style should fill in. I am not sure what to suggest more specifically. Perhaps someone else has an idea.
Yup, I just kept that prompt and it properly filled it in
Above his head looks amazing, makes the two lines to the left and below him stand out even more now
Making very small edits allows the system to just fill the small space based on contextual clues.
Have been having difficulty with made up hybrid animals that are part (a), and part. (b). Such as a hybrid animal that is part dog, and part eagle. Dalle seems to struggle...
Probably better off adding your own text once you have your image.
I got this message from an image Dalle provided me, likely extra careful or sensitive. It was of a female knight kneeling on the ground as she prepared her armour. Dalle wanted nothing to do with its own image lol...
Hybrid animals often present problems. This is particularly true if the animals have very unlike anatomy. The general cue "chimaera" works in some cases.
If I want to unlink my Google account from my dall-e account what do I do, if it's even possible
If I buy credits and I get 15 free ones on the 11th will the 15 just be added to my total?
I believe so.
Yes, the free credits will add onto your paid credits
(Paid credits do not stack up every month though, the maximum limit of free credits you can have at a time is always 15)
Awesome! Thank you!
I have not seen a procedure to do that. You may want to send a message to support@openai.com about it.
Is there any way I can make a nickname for my account profile so it doesn't share my actual name when sharing creations?
I guess I could screenshot, which does work but slightly more cumbersome.
I have not heard this is possible. Consider posting the idea in #suggestions. If the account was created by linking a Google account that adds another layer of issues.
Is there anyway to save my outpainting into my collection, or do I have to download it?
How could I save pictures DALL E and I generated when I'm using my phone? I had only been able to publish pictures but not actually put them on my collection
Can someone explain what the generation square on the edit page is?
so you can extend the image by adding a new generation frame with some overlap to the existing image. it will read the portion of the image inside the frame, plus your new / altered / same prompt, to fill the blank area, to convincingly extend it
currently not, you have to download it
it's at the top-right - a big bold "Save"
So I put this as a prompt and it gave me a content warning “ underwater city of rapture above the water” and I would like to know what it was against?
Hmmm. Hard to say. I just tested the word "rapture" on its own and it was not blocked. Must be some contextual issue. I am trying to imagine what this would look like since an underwater city above the water is a puzzler. "underwater city of rapture" produces an underwater city. I just ran "underwater city of rapture above the water" and did not get a content warning. Produces a rather blurry underwater scene.
I lost credits by moving my browser window while doing a photo edit. Is there anything i can do?
it put me back to my last creation with the main text prompt and there was no way to view the edited photo
Thank you! Wondering why it pops up when erasing tho
if it used a credit it should be visible in your generation history
Thank you it did not show up at first
Hey, I'm wondering if anyone faced backlash from other people after having used Dall E. Like getting accused of not being a real artist or not owning the work...
Not personally, but I've seen the occasional sceptics and bad mouthing of "AI Art" in general. Sometimes it feels like someone during the invention of the television warning that the television has demons in it.
not personally, but i've heard its common when posting AI art in artists' circles. frankly, i can understand why
Yeah, because we've just stuck a hole the size of the moon in the gate that is technical savviness.
I'm on Michael's side though, I think it's a bit backwards. Kind of like how electric guitars aren't real guitars.
It's also very tempting to keep all my prompts secret, but I'm forcing myself to share them with people who ask.
i'm not sure it's the best analogy, as the skills required to play electric and acoustic guitars are similar. this isn't just shock of the new, though it's part of it. it's also not quite like the invention of photography, which is a closer analogy - a new visual medium with a totally new technical skillset. the concern is that there are a lot of unresolved issues with models that are trained on unlicensed data, and the real work of artists. there's also the problems with fidelity and generally poor detail, which is inoffensive to most idle viewers but is no doubt grating to those who have trained technically.
like, i wouldn't go into an art forum with AI art the same way I wouldn't go into a writer's forum with AI writing. it'd rightly get savaged - i know, because i'm a writer. in my opinion, AI tools are fantastic aspects of the creative process, and interesting ephemeral experiments, but if we start to consider them as worthy products in themselves, it's a devaluation of art. but perhaps this is now better for #ai-discussions 😅
I'll catch you there.
any suggestions on prompts when editing to be able to fill back in an area that something was removed from
say removing a person who's standing in front of a wall and just filling in the wall behind
do I put in a prompt for the full image i want, or just what I want to fill in the empty space?
can I share about my thoughts about Dall-E 2 outside the server
I am almost certain the terms of use of DALL-E 2 doesn't sign over the rights to your thoughts and opinions to OpenAI. So, I would say you are very free to share your thoughts about DALL-E 2.
thank you!
The whole area you are editing
Are we allowed to mint our art on Opensea?
Yes.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
I want to do a sort of "Turing test".
I want to generate a face and use that same face in different prompts.
Ex: Generate a face. Use that same face in different situations, emotions, angles, etc. Like a real person.
HOW?
That's not really an option with DALL-E
Is there any AI that can do that?
that's a question for #ai-discussions
On prchasing credits. Is the cost of purchasing credits going up over time? I may be mistaken, but a few weeks i thought you received 150 credits for $15. Now I see it's 115 credits for same cost. Am I seeing this correctly? If so, will costs continue to go up?
The cost of credits has not changed, it's always been 115 for $15 🙂
Great to know! Thanks
Guys, why is adding a beauty mark so difficult?
It just keeps returning the same image... I'm trying to indraw it...
Inpaint it or whatever... I think I must have wasted like 10 credits so far...
I erase a little bit of the cheek, and ask it to add a brown spot... Nothing. Just erases it.
That's 8 credits of trying different synonyms of brown spot and beauty mark with no avail.
I'm messing around with inpainting and the parts I've erased keep getting random letters and word-like images as replacement suggestions. There is no text anywhere else in the image and I'm not asking for text. Is there a way to prompt that I don't want letters appearing? I've tried things like adding "textless" and "no letters" but it just keeps putting them in.
Maybe inpainting's just a bit meh at this moment.
What prompt are you using
I've had nothing but success using it, but obviously i'm biased
Why would you be biased?
I'm a volunteer for the DALL E 2 discord
We're all biased.
true
I'm using all the prompts.
"Beauty spot." "add beauty spot" "add beauty mark" "beauty mark"
I have a watercolor image of a man playing a guitar, but the guitar neck looks totally weird, so I erased that part and added the prompt "guitar neck, watercolor" which added a bunch of text nonsense. Then tried adding other things like "textless" or "no letters" etc. But pretty much every suggestion is adding lettering. Tried some other things to, but it's a pretty small part of the image.
Stylegan will be a better choice for what you want
maybe you could try saying birthmark or mole? but also, why not just use an editing app to simply draw a dot on their face? it would save credits and time
I'm a man of great stature
lol
It's not perfect, but it does understand what a mole is at least 
That's a great example of what I'm not doing, but thanks anyway.
Right, just showing that the AI knows what a mole is so it's prolly the wording or specific generation you're working with that needs to be fixed
No, it's actually me. I'm the problem. It's my attitude.
Oh, then that's something you've gotta handle I suppose
is there any way i can save all the variations?
I'm trying to upload an image to edit an specific part of it, but I always end up in the "create new parts for the image" and not the edit image
how could I edit the image just like the old days and not end up in the new feature?
choose this
Its basically the same
and the selection thing needs to be on it i think
The box just determines where the generation will take place
anyone?
One at a time They'll appear in your generation history
Sorry about that
How can I see past prompts from the phone?
I've seen before someone mentioned turning your phone horizontal, not sure if that still works?
Will OpenAI email you if you apply to be a community volunteer?
Thx i tried but not work maybe i can try safari
I'm trying to turn a previously generated image into an android face using the inpainting option. I am doing this due to the body looking subtly un-human - I have edited out the face but the image (made by dall e 2) is flagging for maybe violating the policy I assume because they think it is a naked person or smutty in some way. is there an issue if I repeatedly attempt to make this image work or will it result in a ban?
does the credits keep on adding 15+ every month or does it stop getting added after the first month?
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
why do paid credits expire wat
Hi, I am a new user. Question: is it possible to take part in the training of the AI? More precisely: I want to create images of parts of our city. So I would like to describe it by words such as: "town hall of city X", or "sky line of city X". Unfortunately, city X is not famous enough to be known by the AI
When do i get the 50 free credits?
You are granted 50 free credits on initial sign up
Anyone tried to use Dall-e to generate illustrations for a book? If yes, how did you manage to create images with a consistent style across the book?
Book illustration is a great use case. To do this, you must think of the prormpts as [subject][style], and first develop a set of reproducible style cues that are right for your book. That set of art style cues can then be applied do various different subjects. You also must consider if the book will be in color or black and white, and develop the style cue set accordingly.
To suggest something for the AI to learn:
• 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.
thank you very much for this form. It asks (optional) "do you know where we can get image-and-caption pairs (like a link to a dataset) related to those concepts". I would provide my photo collection of city X sites with captions. Which would be an appropriate format? I would create such a dataset and publish it on a website.
For that please send mail to support@openai.com -- and also please use the form to explain what you have in mind.
thank you very much, I'll do that
lame question but here we go anyways: i was curious that is it possible that get more credits without buying ? my friend is writing a fictional world for a series of books and we need visual resources
dall-e is one of the best for the job so yeah
oh this is the exact situation we are experiencing right now
• 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 after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
does DALL-E have an API that I can use to send it text and it responds back by the images generated?
DALL·E is not yet available through an API, and users are permitted to access DALL·E exclusively via our public-facing website. We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use. We will continue to evaluate how we release this technology as we learn more during the beta.
hi 🙂 im new to dall-e 2 and i wonder if its a way to add a picture of my own and make the AI do variations of that picture. (example, "a cartoon picture in a comic book" with a picture of my cat)
does dall-e team have office hours? or an email I can send with suggestions on building new models? I filled up the google form, but would love to find a way to enhance the models or work collaboratively..
We have #suggestion for new ideas, and support@openai.com for any questions you'd like to ask OpenAI directly
I believe what youre referring to is known as "style transfer", which DALL E 2 does not currently support
@short salmon : thx. i am already in that channel... let me email support@openai.com in a bit
if i have beta access can i get an invite for a friend who is an artist?
They must apply on their own.
ah i see my b
is there anyway i could make a case for him? he's a really good artist and his ideas for things to generate were amazing
just trynna help a homie out
Each application is independent
Does the app for Dall-E-Mini use the same dataset?
Discussions of other AIs are only permitted in the #ai-discussions channel.
As a community, we value dialog about the future of AI, including alternative technologies and their associated ethical issues.
@limpid harbor just have them apply, i had my co-workers join, and while it took me about a week to get an invite it was under 24 hours for them
Trying to make a character image for a dnd-type game and got content warning for some reason... "Armored caucasian woman with short curly black hair and evil, red eyes, digital art". Any ideas why?
Also i have a really hard time trying to get dall-e to output red irises. If I put red irises it gives me the plant. Red eyes often gives me a stye or a bloodshot look. Any ideas?
Hey, how do I do what they're doing?
Why did I get a reply that says “waiting for you to realize what you did”
It was the guy from I Know What You Did Last Summer. He's a regular on this server.
Was the dall e watermark you left in intentional
Huh
You left the watermark in the real or fake
I’m stupid
It was me
But without it looks real
That's the point, dude
Please use #openai-chatter for chatting.
Please treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.) Community volunteers will issue warnings and then mute, kick, or ban you from the server — depending on the severity.
How do you extend a picture like that?
click edit after creating the image
sorry for the delay just woke up
check out labs.openai.com/editor!
or press the edit button on any generation
will dalle ever do credit sales or discounts
Glad I joined here because I feel better about getting warnings for bad word choices (I was trying to do a circus-related image today and "shot from a cannon" got tagged by "fired from a cannon" worked (kinda, the performer just sat on the cannon instead of flying out)) and I'm glad that generated images getting warnings when trying to edit doesn't just happen to me (heck, I was trying to remove a smoking pipe that was floating mid-air, in addition to fixing an eye). It's good to know they're being checked by people. However, is anything being worked on so that the system recognizes images DALL-E just generated itself?
Hello! Is there a way to generate variations of an image without having to use the crop button first? I can't drag the corners to fit my entire image and I'd rather not crop it if I can avoid it. Any help would be appreciated!
You can try and pre-re-size your image to fit in a 1024 x 1024 square, so when you generate your variations, you're getting the whole image, maybe with a little extra border on either side... But indeed, you'll need your image to be 1024 x 1024 (a square) to be able to generate variations.
Great and I typed up a whole discussion about an content warning issue, and discord said I can't post that here either. So I don't even know how to address it.
And I can't even copy to my clipboard because it ate it.
Thanks will try that
Basically [anthropomorphic animal] style does not equal [explicit content]. Please don't make assumptions.
How do you keep the perspective and main object the same while outpainting? For example, when I am generating a full body, once the selected frame can no longer overlap the main image it starts to try to recreate the prompt ignoring the rest
You need to change the prompt to suit whatever it is you're trying to outpaint at the current moment, since the only thing DALLE uses as context when outpainting is whatever is inside the 1024x1024 generation box, and nothing else.
I make anthropomorphic animals all the time. Post in #❌┃prompt-help and we can help you with yours. The filter isn’t making assumptions, it’s responding to attempts people have made to create nsfw content.
It’s been addressed in #suggestions a lot.
@broken wedge Thanks I will try posting it there!
i know outpainting is in beta but will there be wider generation options in the future ? like 2048x2048 for two credits
Has anyone tried to use Dalle2 to make illustrations for a story? The challenge involved in this is to remember the characters, environment type and style Dalle2 used for the first image that it created and then use it to create subsequent images from new text descriptions.
That's something I was thinking of for a suggestion if it doesn't do it. Like creating tags you can apply to images like "character so-and-so" including images you upload. Then you could have "[Bob] riding [bob's horse] across [desert setting]" if use something to show referencing an existing tag.
Similar to a number of text based story making AIs, have world definitions you can add.
i've seen a make-shift music video made with it!
although i bet that would be one of the coolest uses for it
@obtuse tiger I'm there...
Do I need to add another prompt?
I just want it to "complete the image"
you will need to describe what you would like to be inpainted in that section (or the original prompt), and make sure the boxes overlap a bit.
I'm going to try "rest of the head of this monkey"
Does that seem like a good idea to you?
No, I didn't even make it on Dall E
