#community-help
1 messages Ā· Page 7 of 1
Could someone explain to me what's wrong with this prompt that, according to DALLE, violates the rules? Style+location+teqnique. I don't get it.
This is probably because the location kyiv is related to an active political situation.
Does it mean that Kyiv/Ukraine is banned from the global art community for that? The city itself is just a location.
Yes unfortunately since this can also lead to generations that can be harmful to a specific group of people. Ukraine should be fine.
Not that I am not an OpenAi employee and I'm another user helping out.
Do we have a person who is actually from OpenAI here? I want to raise this issue.
Feel free to contact help over at support@openai.com
it could be political which they dont want
Countries currently in active war are filtered out.
Are you able to use the api key for dalle 2 to make a python script for it or is dalle 2 web only?
DE2 does not have an api at this time and attempting to create one or a workaround it against the TOS.
k
Anyone tried to sell their raw generated prompts?
I've just generated the prompt to verify my account here, but it didn't return me that credit - any help?
You won't get the credit back since it's "free"...
The prompt you use to verify yourself can contain whatever
you want it to contain as long as the code is mentioned in the prompt.
The code doesn't really mess with the prompt itself. 
I wonder if "Tel Aviv" is ok, as it is in Israel
Which has the conflict (I just want peace, I just used it as a FACT, and as you see, I am not taking ANY SIDE)
is there any document thats like a complete guide to dalle 2? inpainting, outpainting, variations, collections, etc?
Maybe someone else knows more/better ones :)
Try the dalle prompt book and dictionary, let me know if you need help finding them.
Ah here we go...
ty !!!!
This is would apply with live action characters (TV shows, movies, etc.) as well, I assume?
The FAQ says "Please share prompts you don't think should've been banned in suggestions so that we can improve.", but where exactly?
My prompt to generate a rococo oil painting of a main coon keeps getting flagged š
It depends, Spiderman e.g would be fine. Bruce Wayne on the other hand is not since the actors face is mostly, well, his actual face :)
Alright, thank you!
The word after 'main' gets flagged since it's a slur
Mmmmm maybe but my prompt an "oil painting of a female silver classic tabby and white maine coon, rococo portrait" did fine though, didnt generate the rococo part though
It doesnt wanna do "an oil painting of a silver tabby maine coon cat portrait, dressed in 19th century suit "
The word 'Moscow' got flagged because it's related to a country of an active political situation
Huh, i took away "cat" from the flagged prompt, and now its fine o.O make it make sense
Others found success with prompting it as just 'Maine cat' too :)
Ah yes that works , ty š
If I enter the same prompt without Moscow, can I get a ban?
Instead of doing 'Red square' perhaps go for something like 'piazza' e.g :)
'Red square' should work though (correct me if I'm wrong)
ok, thanks!
I'd reccomend you to simply go for Piazza though because of the whole political situation etc.
Better safe than sorry 
Is it okay that I'm using the artist badge even though I don't draw? I'm a digital artist (edits) and also make music so I thought it would be okay lol
No worries at all, that's completely fine! :)
Cool thanks haha š
where can i find your music im bored
Idk if I can self promote but i'll dm you
Actually I disagree as I've got non-sense results with it, check the differences between my first prompt and the second one, which is obviously more clear for Dall-E, but still the first one gave me just some random images
How do I get it to download an image rather than example dot com
As you can see the code doesn't really mess with the prompt 
So I'm an artist, looking to do variations on my own work- but it looks like it's getting flagged as a possible violation of content policy when I try to make edits? Why is this?
It depends on the image you are trying to edit..
Have a read through the policies and check wheter something
in your picture could violate any of them
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
As far as I can tell, I didn't actually violate any of those terms.
Can you DM me the picture you are trying to edit?
What is wrong with this?
Instead of 'very sweaty' try 'a heavily sweating seal' or 'a excessively sweating seal' :)
Just wondering if I am using Dall-E 1 or 2 when generating images.
You are always using DallĀ·E 2 :)
DallĀ·E 1 was never released to the public
Ooo thanks!
no problemo 
Are there artstation artists in the dataset?
Do the images always get generated in square format at 1024pixel resolution?
For the time being, the images will always be 1024x1024px in size.
You can, however, 'combine' or upscale them!
Have a look into #inpaint-outpaint as well as these resources!
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y by @spring wyvern
I propose you look at the section beginning on page 55.
The topic 'combining' specifically is on page 63.
This whole notion-site is extremely helpful!
https://bit.ly/dictionairi
https://bit.ly/dictionairi-in-outpaint <- For In/outpainting
https://bit.ly/dictionairi-upscale <- For Upscaling
by @wheat zodiac

Take a look at this list! I don't know if it has since been updated though :)
https://dalledictionary.notion.site/400-Artist-Studies-with-DALL-E-2-73eba3f9411a4f6c8c9682f38a2a761c
Thank you @pine void
Thank you!
Hi everyone, how do you save images to your collection sorry for the basic question i've just started using dall e this evening
There's a save button in the top right corner above the generated image. And there is a download button on the image itself.
This is a cut image of something generated on Dalle, why did I get a warning about content policy? I want to generate the rest of the sculpture
My guess is DALL-E is probably reading in about the uncovered void spaces in the statue. I got one censored yesterday for describing a subject as "sweaty." That was resolved by substituting "hard-working." I think they're just really trying hard to keep things rated G.
Quick question, how many times do I have to get dinged by the content policy to get my account suspended.
Because there are several times I thought I was following the rules and got hit with the content policy breach prompt, and I dont want to accidentally get suspended because the program thought D&D was a form of weird sexual stuff.
Can image size be increased on generated images using certain phrases or words to indicate you require a wide or tall image size?
does anyone know how many violations result in a ban or how that all works? typed in fursona as a joke not knowing it was banned and i got warned :/// i really dont wanna get banned from dalle
What is this slider for?
Its the size of the brush when you correct stuff.
Thank you... I should've realized that
Are we allowed to invite people to the discord?
So I'm trying to ask for an image of a "Realistic photograph of a roomba made of hot pink latex." and I got a policy violation message, can anyone tell me which part is the violation? I'm trying to make an image based on an in-joke with some friends X3
wild guess, either it didn't like you using the name of a company or it thought hot pink latex was gonna be racy.
any suggestions for what to replace roomba with?
"robot vacuum"
fair enough lol, thank you!
Do my credits roll over month-to-month? Or is it a situation of "use them or lose them"?
i believe it's "use them or lose them"! the article on how dalle credits work says this
i forgot if i've asked already but does dalle get updates to its database?
Thanks "free credits don't rollover" - that answers it. ā
The Faq states that while updates aren't frequent, certain information can/will be added in the future. Proceed to #gpt-models for more info on how to propose an addition.
Negative, image size cannot be altered in DE2.
Are you referring to the dimensions of the output itself or how the content is placed. (Ex. A closeup, long shot etc.) ?
Once again, referring to #gpt-models it says that the exact thresholds will not be shared, but if your prompts were actually harmless, you have nothing to worry about.
Nope, please don't try to share your account/codes in any way as its against the TOS.
I think you miss understood the question
thank you! forgot to say that i read the faq thingy and found my answer, sorry
Nope, sharing access to get the code classifies as sharing the account. Can a volunteer please confirm this?
What's a way say well worn, like a object that is worn out. rusty, you get the idea.
Old? Try #āāprompt-help the people over there might be a bit more helpful š
Ok, thanks
Hi! The Discord is currently for users with DALLāE access.
Cheers
Out of curiosity, not saying I'm going to do this, if my friend gave me their code and I put it in a prompt, would it give them access.
That's for you to try if you're curious š Not sure if you'll get flagged by the bot for a breach of TOS for sharing accounts.
Ok...
@stark glen ā Generating Donald Trump is against the Content Policy, please have a look in the link below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Hi Lovely folks! Is there an API for DALL-E?
There is not
Thanks @vivid mortar . Do you know if there is a plan to add that ?
They have not given any details on that so far but Iād keep a close eye on the patch notes whenever it comes out of beta!
What rule am I breaking? =D
Try replacing Moscow with something else ?
Surprisingly works
Might be something to do with Russia and sanctions...
Hi! Countries currently in active war will be flagged by the filter.
Makes sense thanks!
Well.. the problem ist that you need all the tipps from Discord to prompt correctly but you wonāt get access before generating your first image.. I dindāt wanna āpublicly shareā my first like 10 images because they all were nightmare material.. so I āwaistedā all these credits.
But I guess thereās no easy fix for that
Can I share "my" generations under creative commons? OpenAI claims the ownership and only give me the right to reproduce and display it. Not to relicense it.
Why is āOlaf Scholz with body of peter griffinā not allowed?
do users get any invites to share to friends?
No, the best thing to do is to tell
your friends to register themselves, just like you did! 
Have a read through the policies:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
You will be able to find your answers there :)
(It's not allowed to generate public figures/celebrities etc.)
Thanks
The best thing for you to do right now is to not do it...
Legal-wise the ToS simply doesn't cover enough for you to
be able to do so yet, so it's still a grey area.
As far as I know, they're currently being updated/reworked,
so give it some time and keep checking back. 
Thank you. I hope they will change it in a way relicensing will be allowed, then I could use it in one of my open source projects
We already have a lot of learning material, but I agree that it's
not obvious enough to find everything yet.
The team is working hard in the background to provide
the best possible experience for users; so for the time being, patience is key 
Take a look at these recources, I'm sure you'll find them heplful!
https://bit.ly/dictionairi by @wheat zodiac
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y by @spring wyvern
I have a couple of questions involving uploading images for the AI to edit. I wanted clarification on the rules before I did so. Okay, so if I upload my own artwork that has a copyrighted character, is that acceptable? Also, if I upload a personal photo with a photo op of myself and a celebrity to edit the dull background or see what other things I can do, is that also allowed because it's a personal photo?
I understand the rules on deepfaking and no-no celebrity images. It's basically because it's a personal image. The situation is unclear.
Hey, is it okay to use the AI generated images as covers for music?
Yes that's fine! Have a read through the policies though and make sure to follow them:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
Oh, great!
is there a way to use my computer power to skip buying credits?
To prevent deepfakes, DallĀ·E is currently prohibiting uploads of all realistic faces,
even when the face belongs to you or if you have consent.
This is because DallĀ·E doesn't have a scalable way to verify whether you have consent,
or whether a face belongs to you yet. So all face uploads are prohibited . 
I can't give you an answer on the other question of yours since the ToS aren't clear about fanart etc.
Thank you for clarifying.
No, the only way to obtain credits is
to purchase them from the official website
or to wait for the free monthly credits. 
Does DALL E save all the generations or will they disappear from the recents list?
I am saving everything to the collection but am wondering it i have to
You can always check your browser history for urls with labs.openai.com/e/ or labs.openai.com/
and find old images that you haven't saved to your collection via the links!
So you don't have to add them, but it definitely saves you a lot of time :)
hi guys, is it possible to help train DALL-E for a specific style, for example so not famous artist?
I would like to generate images similar to https://www.behance.net/samji_illustrator but I don't know if the system would know what it means, it's not a famous painter like Van Gogh
Take a look at this :)
#āātips-n-tricks message
Quoting Faq here:
⢠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(https://forms.gle/kK2dSL4AzqQ2NnjR6).
^ This is where you can suggest new data, there is a workaround for the meantime though...
It might not be 100% accurate at all times but works great if you want to mimic a certain style 
So as an example, you can erase everything from an image and
only leave a blob left to convince the AI to mimic the style of your art.
That works great for me whenever an artist is not recognized by DallĀ·E yet
#āātips-n-tricks message (for reference)
Yep, thanks for adding on to that buddy š
Thats a great tip! Is there a limit to how the long the recents list will be before it clears?
Yeah, the llimit for recents is 50 :)
Anyone have any idea how long it usually takes to have your account reinstated after your account is deactivated? A change was made last week that cause everyone using DALLE at the time have every prompt be a policy violation, and I really want access again.
The OpenAI Team will manually review your account to ensure you weren't banned accidently :)
and there currently is no timeframe on how that might take afaik.
Just to make sure, you have reached out to the support for an unban, yeah?
OpenAI grants you the exclusive rights to reproduce and display such Generations and will not resell Generations that you have created
So legaley I may not post edit the images currently? I really hope they do change the Ownership of Generations paragraphā¦
heya! is there a way to upscale images with dall e 2?
if not, any good free alternatives (preferably online or pc)
Hey, is there a way to set the aspect ratio for generations?
Edit: Searching through recent messages tells me that no, not really. That's kinda weird š¤
The only aspect ration for outputs is currently 1:1.
hello! i have some questions please š
- why i only see 2 creations on my collection, if i made much more?
- i see other creations look more "pro". its because i got the 50 free credits? i should buy more credits? the paid credits make better creations?
- i can use dall with spanish language? its accurate as writing in english?
thanks so much!
Hi there
To answer your first question, an image is only sent to your collection if you click the save button (image below).
It sounds like you've spent less than or equal to 50 credits so far, so all your previous generations should exist in the sidebar, since the sidebar keeps the last 50. To save a set of generations after they leave the sidebar, either add each image individually to your collection, or simply store the labs.openai.com/e/ addresses somewhere (I recommend bookmarking them) - they'll also exist in your browser history if you have it enabled.
To answer the second question, there is no difference at all between free and paid credits. If you're not happy with results, I recommend reading through some of these useful community made resources for tips on how to craft prompts. 
#gpt-models message
hello, is there a way to specify a prompt for an uploaded image's variations? for example to turn a 2d art into 3d
thanks so much for you clarifications.
one last question: dall e understands spanish as good as english? or its better to describe and write in english? im from argentina š
definitely better to write in english
One more question and thanks so much:
Every creation is unique right? Its impossible having the same creation in spite of writing the same words?
I believe so.
Hello all, just recently received my invite so I'm new to DALL-E and have some questions regarding commercializing the images you have generated.
- when you have a finalized image; are you free to print, mint, sell, ... these images?
- when you'd want to commercialize these images, is there a way to remove DALL-E watermark? it's not that I don't want to credit the tool but more for the sake of the overaal aesthetics
TIA
You can download the image without the watermark using the inspect feature of your browser, or you can remove the watermark in an image editor
As for your second question, refer to this image from the DALL E Beta Blog, found here (https://openai.com/blog/dall-e-now-available-in-beta/)
Why is my prompt against content policy?
Prob cause you wrote "Joe Biden"
I'm wondering if anyone has gotten their artist assistance credits yet? I filled out the application well before the deadline and was told I would receive them today but have not gotten them yet.
Thanks for your reply and resources! I have tried using the link through the inspect feature, but it forces me to save it as a WebP Image... there is no option (that I know of) to let me save it as a jpeg, png, ... Any tips?
AFAIK you need to use an image converter
Theres ones you can download free, and others that are entirely online
I'll give it a try, thank you!
is there a feature that lets you pick which of the generated images makes the most sense and is most 'accurate'?
Right now there isn't. You can suggest any
ideas that you have for DallĀ·E in the channel #suggestions though! 
Paid account question. Iād like to share my password with a group of colleagues at the school I work at. Is this allowed?
Should I flag results that were requests for a gendered noun ("Empress") that produced the opposite gender? On one hand I think it's kind of cool that it's inclusive enough that when I say "Empress" it also gave me a male Empress (and he's definitely dressed like a fabulous empress). But on the other hand I was specifically trying to generate a picture of a female empress.
As far as I know this is NOT allowed with a personal openai account that got a dall-e invite, but they have instructions somewhere for specifically requesting access for your colleagues/classmates/etc, I'll see if I can find them
If you can point me in right direction that would be great. Thanks
Ahhh I was thinking of requesting organizational access to the API for research purposes, not for dall-e use. Unfortunately it looks like right now you cannot share your account:
#community-help message
You can have them send you prompts and you generate them, though.
Quote from the terms of use:
(c) Access and Credentials. To access the Services, you must register according to our procedures and provide OpenAI with certain information (such as identification or contact details) as part of the registration process. Once registered, you will be issued certain access credentials to access the applicable Services. You must keep the access credentials confidential and you may not sell, share, transfer, sublicense or otherwise make the access credentials available to others. You are responsible for all activities that occur using your access credentials. You will not misrepresent or mask your identity or your credentials when accessing or using the Services.
-removed-
Hi, for terms & policy inquiries, please direct your concerns to support@openai.com.
Will do, thanks!
How to use dalle bot?
Yes, but what to type there
I'm not sure what you mean
That explains it all
Thank you. Where was it?
Please check the link I sent above
Is there any specific reason why you can't seem to buy credits from Serbia? I get hit with a "Unfortunately we are unable to offer services to the country you specified." while pretty much every other country in the region doesn't get this, even the non-EU ones.
That's most likely a limitation on the payment processor that OpenAI uses which is Stripe.
Alright, thank you very much, I'll continue looking into it
I haven't received anything today either, and I also signed up (and got confirmation) well before the deadline.
Hmm. I'll keep you posted and see if anything changes.
Given DALL-E's structure, could it not be used as an image classifier? I'm very interested in if and how the DALL-E model would describe images, both those created by the system and other sources.
DALLE does not currently have this feature, but there is a good open source version that works quite well that runs in a google colab
attempts to give you the text description of an image
does dalle 2 build off of the dalle mini database? or are they separate?
DALL-E mini is not made by OpenAI.
ohh huh, why are the names so similar then?
There are numerous articles about that online, feel free to make a search
okay, thanks :) i'm assuming online articles rather than things like a research article?
alright so dalle-mini was inspired by dalle and that's why their names are similar?? just to make sure i'm getting this right
Yes, the name dalle-mini was inspired by the original dalle back in 2021. When dalle-mini went viral this year, OpenAI asked them to rebrand to avoid confusion. They rebranded to "craiyon".
alright, thanks so much! that clears stuff up
hey! im using the 50 free credits, today, im running out so i bought 115 more ahaha. i cant make more than 50 per day right?
There's no limit on how many you can create in a day anymore :)
yaaay! super š
question but what is patch-e?
A very helpful tool to create composite images! For all the details have a look at this 
#āātips-n-tricks message
Hey! J was just wondering since i couldnt find anything about it but if you dont use your credits by the end of the month do they just like reset or does it get added onto?
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
hey guys i see abunch of people taking images and extending em with the use of dalle, no matter how many prompts i try or editing the image i canāt figure out how to extend a image. anyone care to tell me how to do it with a image
Purchased credits expire one year from purchase
Use an image editor or #dall-e-bot to pad the image with empty space around it. Upload it to DALL E, and erase anywhere in the empty space. Enter a prompt of what you want to see, and voila
thank u sm! needed this padding thing i see
got my credits ā¤ļø hope you got yours too ser
Does anyone know if we can invite others still ?
Nope, invite only through the waitlist
got it, thanks
how deterministic is DALL-E? if you provide the same prompt, will it use same noise seed for all 4 images? will two users get different results via different noise?
do my 15 free credits reset every month?
yes
after ive used them all
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
I was just wondering, do images get deleted from 'recent' after a period of time? I generated a few images but didn't save to my collection, but I thought they were going to remain in recent for awhile.
It shows the last 50 generations in the history. If you need to go back further you can use your browser history
Thank you! I think I figured out my issue, I believe I clicked āNewā next to sessions, which probably deleted my previous images. Oh well, lesson learned:)
Yeah, I've almost done that myself before @ripe meadow š¦
Haha yeah, but I overall love it and Iām happy that I was able to get access. I was expecting not to receive an email or Wait for months. Thanks again tho @short salmon !
We're happy you're here ā¤ļø
I assume there are no credits refund? Got some atrocious generations that felt like a credit wasted
Is there an API to be able to integrate DALLE into external applications, or at least an interface specification for future integration?
While this is highly requested, there is no credit refund system at the moment.
Once again, highly requested but there is no official API at this time. Attempting to create one is against TOS. Check #gpt-models for more infom
Q: Is there a publicly accessible API for DALLĀ·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALLĀ·E, which is against our Terms of Service.
Thank you! Somehow scrolled right past the faq for the questions channel
is there a way of deleting something from my gallery if i dont like it?
Hey, gang - can anyone here help me understand why, when I enter the following as a prompt, I instead get the attached image?
"A 25-year-old Caucasian woman, medium build. The woman has short and messy black hair in a feathered haircut. The woman has a round face and no makeup. The woman has intense violet eyes and is pretty. She is aboard a spacecraft with many computers around her. The woman's hair is very short and parted to the left. The woman is wearing a black flight suit with a high collar."
I mean, the character as a whole looks pretty good, but the AI has completely ignored most of the details I provided.
Haha no problem!
Are you referring to the gallery of your device or your collection In Dall-e?
Hmm yeah, try #āāprompt-help and the lovely peeps over there might be able to figure out.
Is there a support section? I am seeing alot of posts about an artist program and it seems to good to be true and wondering if this is some sort of scam. Thanks so much!
here is the email in question. Doesn't appear to be asking about sign in info and the form someone linked is just asking for your email and if you already have dall-e 2 access.
Not sure what you mean by support section but the artist program is legit(many including community volunteers have been trying it out), through which you can get access to credits.
Oh ok so this is a program that is legit that is mainly what I wanted to know I was worried someone was phishing to get a dall-e account
I mainly just wanted a way to verify if this was legitimate and thought a support section would be a good start.
I applied so fingers crossed!
I think you received your first set of credits too :)
I was worried to even apply. Oh my friend sent me the photo I did not get so lucky haha
If this is yours.
Oh! Haha just make sure you don't enter your password anywhere except on the official website and you would be good!
Oh absolutely! I saw it ask for just the email so that made me a bit more comfortable but I was afraid if I attached my email they would be like we need your account info to validate and I would just rather scammers not have my email haha. Thanks so so much
Yep, no problem!
I have legit spent way over my budget this month so fingers crossed haha
The artist assistance program is not a scam. Just to be clear.
Thank you very much! Glad to know I already went ahead and replied hopeful! it would be quite helpful. Do you all prefer I have published things to my dalle page? I have almost all of it still in just my collection
i mean when you make an image it stays in the recent tab
I do alot of infinite zoom stuff so alot is repetitive but needed.
Thanks!
We discussed the image in #āāprompt-help in some detail. Thanks for the referral!
š
Are other people (like my friends) allowed to use images I made if I give them permission and they indicate the image was made by dall-e (for example: a profile picture with a watermark using the '!watermark' command)?
Yes, feel free to use them in anyway you want a s long as you don't deceive anyone into believing it was not made by Ai.
Hi, if my free credits expire on August 19, is that the last day I can use them or the first day I can't?
If your free credits expire on the 19th you will no longer be able to use those credits.
You should use all credits before their expiration date
Do purchased credits have expiry dates too?
Paid credits will expire after 12 months :)
Ah ok
does the word creepy break content policy now? cant generate anything with 'creepy' in the prompt even tho i could before
Creepy is working just fine for me, what is/was your prompt?
Is there any possibility to delete anything from recent list?
Not singular generation, no. You can however delete
your whole recent list and start a new session :)
How to do it?
Next to 'recent' you should find the button 'New'
Thank You so much
Where I can find recent list on the phone version of DALL-E 2?
there's a phone version?
I mean the DALL-E website on the phone
can dall-e understand japanese?
Try rotating your phone 90° :)
To some extent, it can, yeah. Some Japanese artists, for example, are being recognized when using Kanji characters. 
I am aware that I can mint Dall-e creations as NFT, but do I need to credit it? And can I remove the watermark before minting it?
whats the best way to ensure that generating different images follow a similar art style, color set or simply background style?
I don't want to specify a specific artist or anything, but would love to be able to have a set of images that look consistant š
Thanks. It'work
You must always credit DallE and are not allowed to deceive others at any time 
The removal of the watermark is still a grey area in the ToS, and we only have Joanne's unofficial confirmation.
#community-help message
Best thing for you to do is to read through the policies and decide for yourself afterwards:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
#community-help message does this also apply for this statement?
Have a look through
#inpaint-outpaint #top-tips-and-tricks #āātips-n-tricks
but also #āāprompt-help, you might be able to get some better help there 
Keeping a comparable art style across numerous generations might be tricky, depending on the style, etc.
What might work is to remove everything from an image but a blob to convince the AI to replicate the style of a specific image/generation.
#āātips-n-tricks message (for reference)
It does, in my opinion.
Particularly when selling your images/generations.
While Joanna is a person of authority, her statement is "only" a statement and should be treated as such.
There is nothing specific in the ToS regarding the entire topic, which is why I consider it a legal grey area. 
That being said, I am in no way associated with the OpenAI team or the company itself.
This is simply me trying to help while sharing my opinion :)
Which we appreciate and value
Alright, thanks for the info.
I just find it weird that prompts can be used for commercial services with but it has to be attributed. Noone discloses a photoshopped photo. Idk what's the diff between that and this.
Confusing technology
I received mine overnight as well, thanks! Enjoy yours! š
I'm getting warned that my search may go against policy. I don't think I've put in anything offensive but I'm not sure what's triggering the warning, how can I find out without wasting my credits?
What's the prompt? :)
"noir monkey driving a train"
The combination "noir monkey" is getting blocked
It will, yeah
Perhaps go for A monkey wearing a suit?
ty!
Another interesting thing just happened: Generated an image, clicked on edit to mark some parts that could hopefully be fixed.
I used exactly the same prompt, but now it somehow violated the content policy?
Most "problematic" thing I can imagine here is the character holding a sword
prompt is/was: "A young male swordsman in heavy plate armor with blue eyes and short platinum blond hair, arrogantly posing with hands on a long sword, digital art"
The prompt itself is fine, DallĀ·E most likely either has
Ā· a problem with the face (as in, it's "too realistic" I'm not calling the character ugly nw)
or
Ā· it recognises the sword as a dangerous weapon
So yeah, further editing is probably not going to be possible
wait, the generated image is fine, but not modifying it?
Am I in danger if I try to fix it multiple times until it works? Does that consume tokens even when there is no result?
the image is waaaay to good, and nearly perfect of what I want. He just has 6 fingers :|
No worries, credits will only be used if there has been a successful variation/edit/generation :)
Am I in danger if I try to fix it multiple times until it works?
Main question here would be: Is it really fully automatic or is there someone actually checking if it wasn't just falsely classified as problematic
e.g. me in earlier tries with a "bastard sword". There's not really another name for it, but obviously a blocked prompt
A few warnings here and there are acceptable; however, do not provoke it.
When you receive a warning, analyze the prompt, determine what triggered it, and try to resolve it.
If it works, great; if not, try rewriting the entire prompt.
(That's how I do it at least.)
As far as I know, the procedure of banning is automated and unbans are handled by the OpenAI team 
If you need help with editing check out #edit-inpainting
I'm sure you'll find a few helpful people there :)
ohh nice ^^
Hi to all! someone knows how send invites for Dalle?
You're not allowed to send invites :)
The only way to join is via the waitlist
Thanks dude! ā„
I am quite curious as to why there is this verification process in order to join the discord, why is that so?
The bot was made to make sure only users with DALLāE access can enter 
but, why?
surely someone interested could come here to find out if they would want to use it?
I am also fairly curious about the social media policy, I did read the ToS but didnt understand much of it, can you post generated images online to social media?
Since DallāE is currently in beta it makes sense that the discord is only accessible to those who have beta access.
Maybe after DallE is available to everyone, the Discord Server will follow suit :)
Yes, you can share your generations on social media as long as you keep true to the policies.
Also, never deceive anyone into believing that the image was not generated by an AI.
roger
It also makes sense in consideration of the need to moderate each channel.
So, for the time being, it's private in order to keep things as enjoyable as possible for the beta testers 
Does anyone have a high quality version of the Dall-e trademark by any chance?
As in, the coloured squares?
and what would you want to use it for (if you don't mind me asking) :)
I'm personalizing my desktop and I like adding custom icons to the links
I'm trying doing that myself but somehow it ends up more like a weird colored square rather than the logo
Thanks!
^^
No problem :)
Many thanks for the clip interrogator collab, that helps lots. Wish DALL-E would expose that but this is next best.
No problem! It definitely helps get a few good keywords I may not have thought of
I am planning a short book with (generated) illustrations accompanying each page. Is there any kind of openai review process, or is it just an informal "follow the terms of service"?
Is it allowed to publish an DALL-E image as a cover artwork on streaming services? (Spotify, etc.)
Yes it is
The "Edit" portion of DALLE is what you are looking for :)
As long as you are not claiming that you drew the pictures yourself and give credit to AI for the imagery somewhere in the book, there is no review process :)
When publishing your generated images, do you have to include the watermark in the bottom right of the image as long as you still attribute the images to DALL E?
Anyone having trouble right now? Generations arenāt going through
guys where can I find the people generations on the dall e site??
You must always credit DallE and are not allowed to deceive others at any time 
The removal of the watermark is still a grey area in the ToS, and we only have Joanne's unofficial confirmation.
#community-help message
Have a read through the policies and decide for yourself:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
Do you mean these? :)
I mean the place where I can see the construction of the community
Hi all--new here! Just got my invite and have been having a blast playing with the generator. Apologies if this question has been asked before, but is there a way to change the aspect ratio of the image? Some things I've got in mind would definitely look better in a shape other than a square.
At the moment, the only aspect ratio you can generate an image is in the 1:1 Scale.
Welcome to the server and DE2 Btw 
For the time being, the images will always be 1024x1024px in size.
You can, however, 'combine' or upscale them!
Have a look into composites as well as these resources!
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y by @spring wyvern
I propose you look at the section beginning on page 55.
The topic 'combining' specifically is on page 63.
This whole notion-site is extremely helpful!
https://bit.ly/dictionairi
https://bit.ly/dictionairi-in-outpaint <- For In/outpainting
https://bit.ly/dictionairi-upscale <- For Upscaling
by @wheat zodiac

If you mean like on Midjourney where you can search and browse through other peopleās images, I donāt think DALLE works that way. (Images are not public by default.) Your best bet is here, the subreddit, or on social media with the appropriate tag.
great, thanks for the resource! I'll take a look
fs :)
Is it allowed to generate photo where someone is in underwear/swimsuit, because I want to generate photo where someone swimming in the sea
thank u, i'll search on reddit maybe i'll found something good
one more thing .. can i buy the photo that i made in dall e 2?
Have a read through the policies:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
You own the rights to commercial and non-commercial usage
for your generations. (as long as you follow the policies) 
Can I invite friend?
No
Hi everyone, I'm new here, please be gentle š
I read the FAQ but didn't find a way to create pictures in non-square proportions. I tried to read something on the net about this issue and found out Dall-E simply can't generate rectangular pics, but there's some way to ask Dall-E to draw the rest of a picture in a new picture, like shifting the camera up, down, right or left. Anyone knows how, please?
I think this is the anwer!
although I don't quite understand how to do it.
Can anyone answer my question?
I would like to change my user profile photo on Dall e. Please Help
On the discord or on the DALLE website?
As long as you are not making intentionally sexually suggestive photos, thats fine
Once you have made a photo you are free to do anything you wish with it, buy a print, sell prints, etc
Not at this time
Thanks
Hi folks. I'm completely new and lost, so apologies! Can we generate images here on Discord, or only on the site? Not sure of the relationship btwn the server and site. Thanks.
The server just serves as a place for like-minded people to share neat tricks and cool generations. You are only able to generate the images on the site.
On the website. It transferred my Gmail photo. Which shows the logo of my place of employment. I don't want that as my user profile photo.
Please reach out to support@openai.com they will be able to assist you with your account :)
Is that an email address?
Yes it is
Thanks so much!
HI, I really like the art that Dalle2 makes, if i wanted to use it for album covers for my monetized music is this allowed?
do the 50 credits you get when you first got access also go away by the next month, or do you keep those, and it's just the 15 every month that reset?
Lol.. sorry I donāt have an answer for your question and Iād be surprised if anyone could tell you from experience š
You should be able to see when your credits expire by going to https://labs.openai.com/account
@pine void "noir" is black in French. Probably getting interpreted as a racist thing.
never saw this error before! what am i doing wrong?
refresh the page and try your prompt again
it worked! thank you
Any tips for making manga-style drawings look better? Right now they look very "reddit fan art made in 5 minutes on an ipad" when I ask it to make a manga or do something in the style of manga etc. Wondering if there are any magic words to coax it into doing better?
black and white manga or like colored anime?
either!
Here are a few I found searching our discord for "manga" or "anime"
https://labs.openai.com/s/di6JSI0NrXFb4SoxBlMulZvx
https://labs.openai.com/s/XBRKdV8kBir8TkvU9qzwDpk6
https://labs.openai.com/s/iATcKCELTnABUwUNdfFCrAaz
https://labs.openai.com/s/xJLJssJfGGchpck0DSYcSO2L
a few more :)
yea even just searching for panel has some good results
very helpful, thanks! anime key visual seems to have good results.
what about mafia monkey? Is that racist? I gave up and used the "monkey in a suit", I managed to get something close but no cigar
It almost looks like there's a cigarette haha
I think it could just have been the word noir, but that kind of is a style so I don't know. It might have read "noir monkey" and interpreted it as "black monkey"
do drug/alcohol/tobacco references flag the content policy?
NSFW content will result in an instant ban without warning.
DALLĀ·E technology is new (especially in a publicly accessible context), and there are millions of other concepts to be explored that are not NSFW. Weāre taking baby steps and slowly loosening restrictions as we learn more.
accidentally deleted images from a previous session š¦ anyway to recover?
misinterpreted the 'new' button :/
You can use your browser history to go back to old generations
Are there any plans to add Paypal as a payment method? That's where all my "oh crap I forgot there's money there" money is, which is better for me to spend on this sort of thing than "rent comes from this account" money.
Hi I just wanted to know
in the edit function
can I change the medium
I wanted to change a photo to a oil painting
In this document: "https://docs.google.com/document/d/1DZJX3VvC9ZeGBWTxWRB0wLf4Xqp1x8yQDRSg-uappCw/edit#" -- it says "Transparent noise
This one is really cool. You can add transparent noise to an area to basically get targeted variations. You can select an area, use an image editing software to add some transparent noise to it. Try being careful how much noise you want, as this can affect how different the generation is from the original. Because the AI recognizes anything that is even slightly transparent, it will take any pixel that isnāt 100% opacity and take it as something you want filled in. This acts as a sort of controlled variation, because the text input could decide what looks similar to the original, so if you have an astronaut and you apply the noise, you could say that itās a spider or something, and DALL-E would have to work with that, so it would try to make that spider with the same colors as the astronaut. As a more broad use, we can actually style transfer something, for example, we can take a picture of a dog, apply the noise, and then say ādog, pixel artā, and thatās all we have to worry about. As with text/visual hybrids, I havenāt been given the chance to mess with these like⦠at all, but I will update this when I get access :)"
What is that talking about? There are no visual examples and I don't know what it means by "transparent noise", that phrase doesn't make any sense.
This twitter thread talks about it
Yes it will attempt to draw any transparent parts into oil painting, it wonāt modify the existing already drawn parts
ok but, then, what is āsalting the image with transparent pixelsā? I still donāt understand that terminology. I donāt know how to āsalt with pixelsā because those words donāt make any sense to me together.
Does it mean āseeding randomly scattered transparent pixels through all or part of an imageā, thatās the only thing I can guess from that and Iām by no means sure
also if so, wouldnāt DE just replace the transparent ones and leave the solid ones? So it should just fill in the same.
that twitter thread has a message buried further down that seems to say that sometimes the opaque pixels can be replaced too. Perhaps by proximity?
If thatās the case, I could very easily make an ImageMagick script that would chequerboard the entire thing in transparent, making exactly every other pixel go. Is that the idea?
DALLE seems to draw in portions of 16x16 squares, if enough transparency happens within a given 16x16 grid it will redraw it
Youād enjoy this conversation surrounding this message #988918555288408064 message
sorry iām not just experimenting. 14c per and all though.
Gives some good insight
thank you.( i think i should probably read it again when i donāt have a stupid fever.)
Get well soon! Pat pat
Hey guys .. Iām sure this question has been asked already .. but when will we be able to download higher resolution renders?
The reason random noise (sometimes) works is because if you divide your image into a grid of 16x16 squares, DALL-E will completely erase any square with a single pixel missing from it
If you remove every other pixel it will treat it like you uploaded a completely blank image
In this twitter thread you can see the only part of the original image that it actually kept because it stands out as an artifact in every generated image, it's hard to tell if any other pieces got saved at this resolution
What happened to the Dall.E dictionary website? Itās not working for me anymore!
Good eye, I think with some acute and smart pixel manipulation and seeding of the colors youād like to see in a specific scene you can give DALLE a starting point for where to begin
They stand out to me because it was driving me insane trying to figure out how to get it to blend them into the image better when I was experimenting with this
Here's an example where you can really see the squares from the original image in the generations because it can't match the texture
They have been updated!
Take a look at these links and let me know if they work for you
https://bit.ly/dictionairi
https://bit.ly/dictionairi-animation
https://bit.ly/dictionairi-upscale
https://bit.ly/dictionairi-in-outpaint
by @wheat zodiac 
Hey so is it ok to create images using fictional characters in a sfw manner?
Yes, as long as you don't violate the terms of use as always : )
Gotcha
So I'm trying to generate variations of the first image, but dalle keeps giving me low quality versions, is there a way to get a more accurate detailed variation? (both images were made by dalle)
Sadly variations always (in my experience) are lower quality than the original
@vivid mortar Darn, I've been trying to figure out a trick to get the same character doing different things (not in the same pose) but keep reaching dead ends xc
Your best bet would be to try something like this:
Use that image and do what I mention in that message linked above
Bahahaha
ok maybe not
š you get the idea, may need some work on the prompting
I'll try it out right now xD
Yea the method works better with less detailed characters apparently
Yeah It's not working out xD
has anyone tried to upscale a low-resolution image (from craiyon or wherever) and then use dall-e inpainting to add the intricate details? not sure if that would be a use case
Dalle Currently generates images at 1024x1024 px. However do check out this link to find out more about tools you can use to upscale any outputs : ) https://dictionairi.notion.site/Upscale-f8fb441b92b54c16a64aef156a747bdc
God saw that @pine void š¤£
Interesting, I hadnāt heard of specific models for that! I had used waifu2x, which is supposedly based on neural networks, but I was curious about the idea of something that did more of an attempt to āguessā the details, since waifu2x is pretty āconservativeā and works best on cartoons (hence the name)
if one of you runs that webpage, I would add waifu2x to it, itās free http://waifu2x.udp.jp/index.pt.html
Afaik it's run by our friend @wheat zodiac : )
no pressure, just a suggestion
Yep, very much appreciated!
Sorry if this has been answered before but can I post 'my' images on Deviant Art? Obviously I will be super clear that they are A.I generated and that my involvement in the process was minimal. I just want to spread the word...
yes
You can!
As long as you follow the policies and, as you said, give credit to the AI/DallĀ·E you're good to go 
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
how do I get a credit on prompts that just didn't do anything? I don't like to blow money on "it just didn't do anything at all"
If it didn't produce an image at all, contact support@openai.com
No, I mean it just completely ignored my prompt on an infill. It infilled with just matching the background. I could do that with content-aware fill in Photoshop for free. I verified the prompt works in isolation. I tried another prompt. I wasted a dollar's worth of credits trying to get it to fill in anything other than "more wall and fog".
(and there's nothing in the image that should satisfy the prompt in any other way that I can see. not even a little)
This just does this no matter what I tell it to do.
Just saw this, I replied in #bugs-and-issues.
gotcha
Do credits roll over month to month, or do you have 30 days to use them?
Free credits don't roll over, paid credits will last for 12 months :)
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
Heyaa... I saw that someone was able to extend an image... like a character that originally cut off at the legs,. but they could chuck it up in dalle2 and it generated the legs and feet to match... how to i prompt that?
So I posted this in new-user-creations, as I think this image is really aesthetically pleasing. However, I was wondering if anyone had any advice on how to get the buildings in the image to be a bit more clear. It looks like the city I want from a distance, but the buildings are really abstract up close.
https://labs.openai.com/s/BeBKSkyvowgquqqAFklnYYUX
You can write, full body after your prompt (including comma)
Muchos grazias amigo
de nada
First of all this prompt is really aesthetic and beautiful
Heres the image done for you. I added lighten ground in the prompt
https://labs.openai.com/s/YgUyfa1qPW1Gu4WOM6ZkgLh0
https://labs.openai.com/s/PYb1KJXoqh0dqUiZeclH1xio
https://labs.openai.com/s/avzIoD1BCVmdKDJ1vW9h7nGd
https://labs.openai.com/s/5IcyF56UVN1HwkyMNJhbLqDD
Or you can also add the name of any light source near buildings (lamps, street lights, candles)
It will work well
Thank you, those are stunning. I will try that.
Can someone tell me if there is a way to download every image on your collection at once ? If not, tha could be an useful feature to have
There is!
Have a look at this 
#āātips-n-tricks message
Thanks !
I'm trying to make a nice picture of a frog, but only black people are being generated! What should I do?
Pastel green and yellow frog man, anthropomorphic, furry character, trending on artstation, cute art, top hat, suit, profile picture
Hi, does anybody know how to properly copyright dalle2 generation im using one of my generations for a book cover for a publisher
As for Copyright, I'm unsure
You might want to try contacting support@openai.com for more information
Hi thanks where did you find this info
found it
I have a question, what would be considered as unsolicited bulk content?
Great question, please drop an email to support@openai.com regarding clarifications to the content policy.
Done so. Ty
Is there a way to access past creations? I went a little overboard recently and noticed only too late that my old creations were no longer accessible. Even just having the prompt history would be great, as I truly had some interesting items in there.
Yep! For now, you can access them through your browser history
Can you please clarify what you mean by through my browser history?
You can always check your browser history for urls with labs.openai.com/e/ orlabs.openai.com/
and find old images that you haven't saved to your collection via the links!
So you don't have to add them, but it definitely saves you a lot of time :)
@earnest arrow its the part of your browser you use to see what websites youve visited in the past
For chrome, youd press Ctrl+H
Yes I am familiar with the concept and I use Safari. Are you sure each creation would be saved in the browser history in this manner? Is there no other way to retrieve the prompts?
They only way for you right now is either by saving them in your
collection, or by going through your browser history 
Thanks for letting me know!
is this the right discord for API assistance?
There currently is no API available to the public :)
This is the Discord for DallĀ·E though
Q: Is there a publicly accessible API for DALLĀ·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALLĀ·E, which is against our Terms of Service.
ah whoops, i'm actually not using an API now that i think about it š
i mean general technical/programming help for accessing dalle š
my requests to images are being rejected on 403
Programmatic access is currently not allowed, this is against the ToS.
How do I get dalle to generate things in a Ps1 style. I tried low poly with rough pixelated texturing but that didn't work. I also tried PSX, and many variations
Could you give an example of what youre looking for?
In this case, a Ps1 style 3d model of a liver. Here's an example of an orange.
Ignore the fact that it's a gif of course.
Hi, could someone help me? I wanted the AI āāto create an image of a character, but the AI āāends up generating another character with the same name. Does anyone know how I could be more specific?
Could you provide a screenshot of the Prompt and the output so we can help you better š
Can I send a screenshot of the screen here?
Sure!!
Yeah not really sure how to generate that type of an output tbh, but do check out: https://dallery.gallery/the-dalle-2-prompt-book/ for more reference on different art styles etc.
It just doesn't know how to uv unwrap I guess
Yeah I'll check it out
The fourth image had the same outfit, but the character itself is not the one referred to in the sentence.
asdksakd
While it is very hard to identify the exact cause behind you not getting the expected output, it is very likely that DE2 does not currently identify the reference you made to what I'm assuming is a TV Show.
It was supposed to be this character:
In this drawing style:
I will soon delete the images.
Ohh gotcha, chances are it doesnt identify the reference : (
No problem! 
Hey. Is it required to attribute to OpenAi or Dall-e 2 in our commercial use of the generations ?
Yes! Make sure you don't deceive anyone into thinking it was not made by AI too. š
Got it. Thanks for the help š .
No Problem! 
Hey Guys, how much different prompt words like "3d render of..." exist?
Is there a list of words?
Happy about feedback!
Yes! Do check out the Prompt book to identify different types of art styles etc. https://dallery.gallery/the-dalle-2-prompt-book/
Do note this is an unofficial source
Thanks!
No Problem! 
Does anyone know if plugging "skeletal remains" into my prompt will flag it?
Given the AI crazy flagging history, I'm thinking there's like a 95% chance.
Nop, skeletal remains won't get flagged 
Huh. It did work! Thanks!
Will dall-e always be paid? Or will it be free or cheaper when beta ends
At this time, we do not have any clarification on the post beta days so we can't really answer that question till Open Ai does.
Thanks
Hi dall-e Discord. I'm looking for an overview on "who can see what" regarding sharing images.
'Published' images are viewable only by who has the link.
Unpublished images (default) are only viewable by the prompt creator unless 'Published.'
Do I have this right? Are there any other categories? Thank you
Yess, you got that right 
Awesome. Thanks š
Is there a specific person I ping for what could be seen as a major bug for users of Dall-e?
You can post any bugs/issues in the channel #bugs-and-issues 
If it's a huge problem, as in, really problematic
please contact the support of OpenAI and/or tag a community volunteer (the names in pink)
A simple post in #bugs-and-issues usually should do the job though :)
E-Mail: support@openai.com
Live Chat: beta.openai.com
This has happened repeatedly with figures and animals. Whatās the fix?
This is a common issue. The easiest solution is to "outpaint" the image using #dall-e-bot and the image upload and edit feature of DALL E
You can always try adding camera angle language to your prompt, but it is hit or miss (at best)
I just learned to outpaint but sheesh. All those extra steps to do what should happen to begin with. I mean, I asked for a trout, not an excerpt of a trout
I tried ālong shotā but it had no effect. But, thanks for trying to help me jpcguy89
Keep at it, it's worth it.
thank you!
Try this technique, I think it'll work for you #āāprompt-help message
Does anyone know why the Dall E dictionary is down? Thanks!
its working for me!
does anyone know how long till your account will be activated again? I reach out to the support, but am curious how long it will take for them to respond or read my e-mail
Hmmm really? Could you maybe send me a link, Iām using one from the tips and tricks chat but it keeps saying the owner needs to update?
Not a problem, glad it was a simple issue to fix.
Hey guys! I have question, may I ask for your help? How do I ask for words to Dell e?
Every time I've seen someone try to put words in Dall-E's mouth, it tends to mangle them
e.g.
same problem
Hello, I had some questions regarding the Credits system.
- what time specifically on the day that you signed up do your credits expire?
- when you generate a prompt, does it use the free credits first before the paid credits?
Thanks
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
DALLāE consumes the free credits first
ah, thanks
I just wonder when the credits will expire - eg, midnight on that day or is there a specific time?
Unfortunately I have no information regarding that, if you'd like to know more, please drop an email to support@openai.com
Here is a prompt that got flagged as not following content policy:
cat in painting style of Yoshitomo Nara
Any ideas why? Thanks in advance š
Probably since it doesn't know about the style Yoshitomo Nara and thinks you are trying to use the term to generate an image of a notable person.
Instead of "style of" try saying "painted by" or something similar
i'll give that a try
cat painted by Yoshitomo Nara gave the same error. I've got to conclude that @stark lantern was right.
It's more likely with the artist's name. A vulgar word is in it.
Do you think it get's triggered even when it's embedded within another word? 
Yep!
Ohhhh yikes...
can the image generated be changed to 1920x1080? Wallpapers would be cool
Try it with the Kanji characters for the artist's name! 
That worked for a couple of users that wanted to generate a prompt with the name of a Japanese artist
Right now images can only be generated in 1024 x 1024.
As Avin already said, for the time being the images will always be 1024x1024px in size by default.
You can, however, 'combine' or upscale them!
Have a look into #inpaint-outpaint as well as these resources!
https://pitch.com/v/DALL-E-prompt-book-v1-tmd33y by @spring wyvern
I propose you look at the section beginning on page 55.
The topic 'combining' specifically is on page 63.
This whole notion-site is extremely helpful!
https://bit.ly/dictionairi
https://bit.ly/dictionairi-in-outpaint <- For In/outpainting
https://bit.ly/dictionairi-upscale <- For Upscaling
by @wheat zodiac

ok thank you
If that were the case then AI would object to this prompt
shitake in a peaceful hippy valley
but that prompt works, no problem
shitake may have been whitelisted in the filter but afaik, the correct spelling for it is shiitake
OK,
computer programmer painted by ćØć·ć㢠ćć©
and
computer programmer painted by Yoshitomo Nara
are equivalent phrases. And they are both blocked by Dall-E.
It is much more plausible to conclude that Dall-E interprets both of them to represent a request to portray a living human, than it is to conclude that Dall-E would object to a word in one language, because if the word is translated to another language, a substring of that word is a swear word in that other language.
a computer programmer, painted by å„čÆ ē¾ęŗ
That one works just fine 
ok, I will test that phrase, but without a comma ... just to see if the comma is significant
and ... the comma didn't make a difference. So it seems that Dall-E recognizes the artist's name in Kanji, but not Katakana, and not in English either. š¤·āāļø
How to increase the number of images searched?
Im curious about when did you, the creators & developer's of Dall-E decide to create, and why to create Dall-E?. Also how did you decide on using OpenAI?
(Sorry if this is not appropriate)
The default right now will always be 4 images.
However, if you're creating something such as textures e.g you could ask for a
3 x 3 arry of tileable ... 
You could try using ";" supposedly some other guy made a anime dall-e library with all of knowledge for anime/manga based images
Like (your prompt; artist name or style)
What is this service and what do I get when I buy it.
1 prompt costs 1 credit each :)
Meaning, so far you have had 8 generation and got 42 left 
115 credits (generations) currently cost $15
Is there a similar discord server for OpenAI's text-based APIs? I want some help designing text prompts for GPT3, etc.
Hi, does anyone know why I would get this waring for this prompt? A full body editorial fashion shot of an edgy, alternative female model, shot on film ā the warning: it looks like this request may not follow our content policy... I don't get why this type of prompt violates anything?
Yeah I don't why either, my account got suspended for recreating a character that has a muscular body, so watch out, stay a way from any type of that kind of prompts. If you get the warning to many times you will get suspended, btw if somebody knows how to get my account back that would really help
@warped basalt I saw they give you some kind of documentation in which you can appeal to it and they check your prompt history and return access if you have nothing wrong. Check what you received.
Support@openai.com can review it.
How will I get trusted-user role?
hello, i joined the waitlist in may-june and been waiting since, in early august one of my friend got into the waitlist and got accepted the day after, so i thought there were something wrong and decided to re-apply to the waitlist, and my second attempt got into it in less than a week.
Yesterday my first attempt finally got accepted, but now here's my question, can i use that second account ? (i still didn't use the link in the e-mail to create the account)
From what i read of the OpenAI/Dall-e 2 TOS (https://labs.openai.com/policies/terms) i don't see any mention of this
Quick question, what is the term for taking an image, scaling it down outside of DALL-E and deleting content around the image so you can upload it back to DALL-E to render a larger picture. Negative selection, or something like that?
outpainting, the opposite of inpainting
#suggestions message
This is the most recent info about the trusted-user role that I know of 
Maybe someone else has a little more info though!
That's it, thank you. I'm new tot he terminology
One account per user is the general rule... so no, you can't use that second account for yourself.
You might be able to give it to one of your friends who hasn't signed up yet, but I wouldn't do
that before getting confirmation from a community volunteer. 
For now, leave the second account be and simply use the one you have :)
i was expecting that yeah that's why i am asking, but is that rule written somewhere as it's not in the TOS?
else yeah i was also about to ask this as i have friends who are interested
how can i get in touch with openai members to know more?
E-Mail: support@openai.com
Live Chat (available from Monday - Friday, at least I think so): beta.openai.com
thanks
The mention you're looking for regarding multiple accounts can be found here: https://openai.com/api/policies/terms/ ā this link can be found from the link you sent, at the very top.
How do credits work? Is it like 1 credit = 1 generation?
That is correct
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
This role will undergo a revamp, expect updates about this in the coming weeks!
api tos also counts for on website usage?
okay i did not know
As per the link you've sent, yes.
okay so considering these lines, if i buy credits can i use the second account?
and if not can i give it to a friend?
Scroll down further and you should find something regarding lending your rights with regards to the services.
If you have more questions regarding the policy, feel free to drop the team a message at support@openai.com
okay thanks
I have been banned for my prompt below. I would like to have my 100$ back. I do not like Dall-e from the 10 min of using...how do i get a refud
refund
You've been posting this in so many channels. Next time, only one post will do.
How to make infinity zoom videos with dalle?
Hey I have heard mentions of something called patch e but I can't find it on the internet anywhere, could someone clue me in
I used this video to make mine https://youtu.be/tux81Z3uXMY
In this tutorial I will show you how to make a multi-layered infinite zoom using DALLE, Photoshop, and After Effects.
Audio by Tyrone Estephan
Chapters:
0:00 - Intro
0:18 - Generating images with DALLā¢E 2
3:27 - Fixing images midway through
3:59 - Generating the rest of the images
5:24 - Setting up the comps
6:08 - Nesting the images within ...
The site you are referring to is patch-e.com
I did the verification thing which used 1 of my credits and I wasn't credited back like it said I would be - can someone help with this?
You don't get paid back for the verification image.
Instead the verification can be done regardless of prompt, effectively coming combined with your first image
Hi! It's not supposed to give it back, please refer to the following screenshots:
I'm writing a popular media article for a uni task. It might be a silly question but can I use ai generated images as photos in it? I skimmed over the tos and copyright isn't enforced - I want to make sure it's ok.
Yes, you have all the rights to the images as long as it follows the Content Policy.
I see I see. Thank you :)
Hi, Thanks for answering - I guess I don't get the 'you wont waste a credit' part. I took that to mean it would replenish it.
The prompt to verify can be anything you want. So you'd use one you were going to generate anyway, and append the code for verification
Hey i haven't Photoshop and after effects. Is there another possibility?
Tools for animating DALLĀ·E generated images.
Thanks!
Asking on behalf of an acquaintance: does Dall-e recognize Pantone colors?
I've seen this on hex codes appearing to have an influence
Ty!
Obviously recognises what a Pantone color swatch is. Will it effectively implement it in your generation? Only 1 way to find out. xx
That's pretty much what I told them
how can I look at all the bot commands?
Thank you! I have sent them an E-mail, let's hope they reply soon!
The answer is "yes, aside from Dall-e color issues"
Just a head's up that the bot does not, in any way, mention that you should add this to a real prompt to avoid wasting a credit; the instructions are only to paste a code into DALL-E:
hey guys, it seem the edit feature is not working for me consistently
im trying to remove the black background and add something but results usually turn up blank or with random letters
Can you show us what's happening and the prompt you're using?
its just an image of a creature but the background is black so i edit out the black and whatever i type doesnt make a difference
sometimes when i delete the initial prompt and just write what i want in the panted area it works but sometimes not
and when i keep original prompt and add what i want changed still absolutely nothing changed in my results. what a waste of my money on credits lol
is there anyone at openai that can address this? It literally is robbery lol
We can't even speculate what the problem is without seeing the actual image and the prompt you're using
Hey, Anyone able to help with this question?
They say you are allowed to sell/use images you generate using the site.
The only thing I've not found is if you must keep their watermark on bottom right or you can pay them extra to remove it.
You don't need to keep the watermark, but you can't use it in a way that implies that the image wasn't generated by AI
@storm spindle just make sure I understand correctly sorry.
- I can't say it was generated by AI?
- I must tell it was generated by AI?
Sorry about that.
2
You can't erase the watermark and then post it somewhere acting like you are the one who created the image in photoshop or something
don't pretend like it's a real thing
Aww ye 100% thank you both for help on that question.
You don't have to keep the watermark, but you can't erase it with the intent to claim that it wasn't generated by AI
Depending on the content of the image and the context that it's being used (eg. a news article) it must be credited as being generated by AI to prevent spreading misinformation
In order to help, we really need to see the image youre trying to modify, the results, and the prompts you're using
Are you DM-ing the bot here?
I had to DM the bot to get access to this server, yes
That is not the intended way to verify. Users enter through the #verify channel and use the bot from there. The welcome channel and the bot there has the complete info regarding the credit that will be consumed upon verification.
I'll notify the bot dev regarding this, thanks for letting us know!
So uh. Is there some way to know where credits I got came from? Im in shock at the moment at how many credits just appeared in my account
Did you apply for the artist access subsidy?
I did maybe its that? I didn't get the email saying for sure they were applying it yet so maybe that's it was not expecting it to show up today if that is it. Gotta go shoot some more work then!
That's the old version of the bot
It's been updated in the past month
That should be it then š
It is just more than i expected even for that but either way very very thankful it means so much to me.
I just really wanted to say thank you too whoever it was even if it is the whole team.
oh? May I hear more about that?
Yep! ~~The artist subsidy form can be found at the bottom: https://openai.com/blog/dall-e-now-available-in-beta/~~
EDIT: The first phase of the Artist Subsidy has apparently ended last August 10th. Another phase will open up later on. Kindly wait for updates.
is it really only name and email?
I suppose i'll apply then
how is it decided upon further?
I beleive this was a pre sign up sort of a list where users can enter their information to get an email once more information is available.
@verbal crag is a direct link available or does it happen through mail once you fill this form?
that's what it looked like, but apparently people already got accepted
People who signed up earlier may have gotten an application?
Is "Pencil drawing of Saul Goodman drawn in Omori style" againts the rules?
hmm..
because he is a real person
are characters only played by real actors within TOS??
that's a really good question
Hi @coral hornet & @stark lantern ā apologies for the wrong info! Please check my original reply for the edit. #community-help message
š
Fictional characters are A-OK as per OAI.
I'm not so sure, that would make all actors fair game for generations despite being real people
Their content policy seems pretty explicit on this: https://labs.openai.com/policies/content-policy
Hmmm I think it's ok if it's a drawing (for example if you make a prompt "markiplier fanart")
because it is on the DALLĀ·E 2 Prompt Engineering Guide
Where can I see all my DALLE-2 creations? It seems the "Recent" section doesn't loads all.
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 via the links! :)
how hard is it to get literal text into an image? would the "quotes" thing work like in a web search?
thanks!
DallĀ·E struggles with the latin alphabet quite a bit.
Words with 2-4 letters sometimes work, most of the time though
it will give you gibberish (or Dallish as some call it) 
Fs! :)
the terms say it is allowed to commercialize the images, so is it allowed to sell the images as a NFT?
also, is it allowed to modify the images and commercialize it? like adding other arts to the image and removing the logo at the bottom
I believe the NFT question was asked before and the general consensus seemed to be that considering you fully own the image, you're welcome to do with it whatever you like. The same way you'd be allowed to do whatever you want with other art you might make.
the policies say that you can do whatever you want with the images, but that openai still owns it
- Use of Images. Subject to your compliance with these terms and our Content Policy, you may use Generations for any legal purpose, including for commercial use...
but
- Ownership of Generations. To the extent allowed by law and as between you and OpenAI, you own your Prompts and Uploads, and you agree that OpenAI owns all Generations
Has anyone tried to change their own pictures into different style? How can I do it best?
Context: I have an old family photo that I want to redo into watercolor. Or maybe a digital cartoon. My mom is turning 60 this year and I think it would be a fun present
Unfortunately, DALL-E doesnāt have a way to change the style of a picture. You can erase parts of the picture and generate new content for those pieces, but not regenerate an entire picture in a new style
Ok got it. So I could change the background into something else, for example
Yep
hiya, came here to ask a question to help clear up some ambiguity. there's a card game that allows you (as a player) to upload your own art (assuming not copyrighted material and you have permission) would DALL-E be acceptable to use in this case assuming proper credit? (i.e. line of text stating DALLE-2/OpenAI)?
Hello friends š
im trying to ask dall-e render the full object nad somehow even long shot, full frame and his friends doesnt work
maybe there is more tricks to prevent cutting?
a subquestion to this one, according to the terms, if a user were to violate policy and get their account suspended, would the art used for these cards be no longer acceptable to use (and would have to be replaced)?
Why can't I get a whole image of a subject? Everything is weirdly zoomed in or cropped.
Are triggers of the content filter reviewed by a human before account suspension? I've hit it twice today and I'm not sure how at risk I am. My queries were somewhat benign. I understand the filter, not questioning it, but curious about my risk of banning.
I've been doing a lot of cyberpunk imagery. One was a katana on a cutting board with little bits of sushi, lilies, and "flecks of blood". Tripped the content filter, that's cool, that one is on me.
Second time was (I assume) the phrase "bastard sword", which, you know, I get, but also, it's a kind of sword. Didn't think about that one.
Can anybody speak to that process?
Also, when did DALL-E switch from 6 results to 4?
Been trying to upload and edit a picture I generated earlier on dall-e, but I'm getting a "It looks like the uploaded image may violate ourĀ content policy." Is this common?
I wouldn't say so, could you share the picture and the prompt?
Not sure if there is a human verification process but you don't have to be scared if your prompt does not violate the content policy.
Hey guys sorry to ask, do you know if 16:9 does anything? I use it in all of my prompts but I do it just out of habit and was wondering what it changed
Hey guys! Iām new, just got access a couple days ago so Iāve trying some prompts to get a feel for the AI, and I read the content policy closely as I definitely never wanna be at any risk of being suspended. Today though Iāve gotten a few content policy violations and I was confused what it was that was in violation, mainly ākanye west eating an entire universeā and ānicki minaj performing for an audience of muppetsā lol. these are obviously dumb prompts and i probably should start focusing on the art for my videos that Iām mainly using the ai for, but Iāve seen and done prompts before that use celebrity names in the same seemingly non-violent or mocking way and i was just wondering if you guys have any advice on what specifically in these prompts was in violation? sorry if these are dumb questions guys, iām having a ton of fun with this thing and canāt wait to work with it more.
I think it really is just being celebrity names
Someone earlier today couldn't use an artist because they had the same last name as Trudeau
word Iāve def noticed that it seems to be mainly more current pop culture figures and less ones of the past, i did people like anthony bourdain and elvis with no issue
What is the process to make a composite? I see all the huge images in that channel but don't know how to actually do it
How would I make DALLE have characters smaller in relation to the rest of the background? Any time I describe a character, they seem to be the main focal point, rather than part of the background.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
This is the part of the content policy that's the issue
There's a great website patch-e.com that makes the process much easier. I recommend checking it out
Hey! Sorry you're having trouble. @keen hamlet can probably help you over in #āāprompt-help
Hmmmmm.
Your best bet in this situation is to use #dall-e-bot to pad the image, upload it to DALL E 2, and run the prompt again.
Q: I just got a content warning! What does that mean? Will I get banned?
⢠Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
⢠We chose to err towards our classifiers/block list being aggressive to start, but have been constantly iterating on them as we learn more. Please share prompts you don't think should've been banned in suggestions so that we can improve.
⢠While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
@half quartz
DE2 has some grasp of foreground/background, but it can be difficult to control. "a man is standing in the foreground. a group of other people is in the background." @restive river
I missed the faq. Much appreciated.
Not a problem. Happy to help š
Thank you!
Does anyone know what to type for a cool anime style
Ask in #characters -- they do a lot of that style there.
Thanks
Hey all, noob and have a few questions:
- I just realized the image creations on the right margins eventually overwrite, finite amount showing. Are those earlier creations gone for good or is there an archive of my creations?
- On the right margins there is āRecentā and ānew.ā I started a new batch of creations. What is the āRecentā tab for?
- Collections: how do I use this? Is this for saved images? Is there a max amount to save there?
Thanks for your help..
Hi there! I'd be happy to answer your questions!
- The recents tab stores the last 50 generations you've done. If you want to find your earlier creations you'll need to find them in your browser history.
- New simply erases your history and starts it fresh. Recents is where it stores your most recent 50 generations, as stated in #1.
- Whenever you click on an image you can press Save in the top right corner and it will be added to your collection. There's a maximum of 10,000 images that can be stored there.
Asking this again since it was buried a bit
Subject to your compliance with these terms and our Content Policy, you may use Generations for any legal purpose, including for commercial use. This means you may sell your rights to the Generations you create, incorporate them into works such as books, websites, and presentations, and otherwise commercialize them.
So from OpenAI's perspective, legal right to use the images is reliant on adhering to the content policy. edit: whether or not the account is suspended is irrelevant here.
No question for 5 hours š„²
Man I love answering stuff here to the point I miss it when no one asks something š
Itās work time for emea and really early for na
Ok so when using the generations I don't need to include the watermark? But I need to make clear the generations was made in collaboration with AI right?
I see that sometimes Dalle-e makes some weird letters...
Is it supposed to be some language?
If yes, what language is that?
take this pic as example
Dall-e is just not good at text so it's probably supposed to be English. I'm guessing there's something in your prompt that could be interpreted as having text. Happened to me when I used "middle earth" as prompt and got generations looking like LOTR book covers with nonsense text
What was your prompt for this?
A music band poster about a musical group made up of a bee, a dragonfly, a butterfly, and a fly
There you go, its the musical band poster part that does it
I would try something like bee playing drums, a dragonfly playing guitar, a butterfly singing, and a fly playing bass, "whatever style you imagine your poster be in"
But you might have to experiment a bit with other elements too, like "on stage", "Heavy metal" or "rock". Sometimes prompts you don't want int the generation will just pull it in the right direction too
it seems like there's no way to view history of previous searches without making a search (so you lose access them if you use all your credits). the side panel to the right is not visible. if i go to browser history i can access a previous search and the side panel is there. but nothing from the main landing page. is this a bug?
Hi there! Sorry to hear youāre having trouble accessing your previous generations. Are you on a mobile device?
@short salmon i'm on a laptop using firefox. i have the same issue on chromium
Do you see this icon anywhere in the top right?
oh wow 𤦠i thought that icon was a refresh action or something. my credit-frugal mind didn't want to click.
i suggest using a different icon to indicate it either would slide a panel out (like the one that collapses) or is more indicative of history
Thatās a great idea! If you want you can leave a message in #suggestions for the OpenAI team
thanks @short salmon i've left a note there
Yes, if you are not using the watermark, make sure Dalle is credited.
Hey there, do previous searches affect current the current search? Ex: Say I do searches for a knightās helmet with a unicorn horn, then i swap out to a baseball helmet with a unicorn horn. Then I search with a ski helmet with unicorn horn but now some of images showing up have hint of the knightās helmet shape. So do my past searches build up and start to direct my new results? If so, and I donāt want this, is this why there is a āNewā search option on the right?
Hi! I donāt believe previous generations affect future one in any way.
Dall E-2 learns to create different style of images from images available on internet or it generate from images and text relationship that is already available in it's database in form of neural network ?
Hello! DALL E was trained on a static set of images. It does not access the internet to search for new images
ok
Is there a way that I can upload an image and have DALL E try to expand it and generate what it thinks would be around the original image? I think Iāve seen those before but I donāt know how to do it
I was wondering this too because I was making a bunch of pig related generations and then I asked for a snail and one of the results was a pig snail hybrid
Can I sell products online or anywhere with images Iāve created on dalle , do I won commercial rights , do I have to mention on product description that I created online with dalle , thanks
Own commercial right s I mean
So if I want to participate in #daily-theme I get to use the prompt I want and try to get it to match the theme?
Yes, you have full commercial rights to your outputs. The golden rule is that everyone is aware that it's made by AI and no one is deceived into believing otherwise.
Yes!
Ok thanks š
It is!! š„
No probs!
š
What does 'request was canceled' mean?
Dang, is there an outage or something? I can't generate anything
Edit: yes, see #bugs-and-issues
Is it possible to change the name you use for sharing generations [name] x Dall-E
This isn't a feature at the moment, unfortunately! However, if you have signed in from Google, you can try changing your name on the google account or you can contact support via support@openai.com and they should be able to help.
alright, thanks
Np!
It has been reported by several users that while they can generate images with pipes, you can't upload one with a pipe. That's probably what's stopping you here...
I have many pipe monkeys, will test with one of those see if something similar happens.
Yeah, let me know what happens!
Yes, there is! The easiest way to do this is by using #dall-e-bot to "!pad" your image by a percentage. You attach your image to your message and type "!pad" followed by a percentage to pad the image. So lets say you wanted to add a 50% padding, you would type "!pad 50" with your image attached. Then you upload that new padded image to DALL E, and run it with a prompt of what you want to see in the new area.
Can't wait to see what you come up with using this awesome feature!
This may be a stupid question, but can you do Prompt + Image variation? I'm just wondering but it's probably only one or the other.
are you asking if you can generate a new set of images while editing?
No.
I just mean it'd be cool to be able to upload an image, then instead of writing something normal in the box, giving dalle suggestions or recommendations as to how the variations could be.
If it's not a feature it's probably likely to become one in the future.
Yep, feel free to use #suggestions to communicate this to OAI.
Okay so it isn't a thing yet, thanks.
Afaik, its not!
Are existing album covers off-limits to be used within Dalle and changed, or are they seen as "public domain". I imagine the former but wanted to check
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/
According the Content Policy, "Do not upload images to which you do not hold appropriate usage rights."
that's what i thought. thanks
I think I accidentally erased all my generated prompts by choosing a new session. Didn't realized those weren't kept. Any way to get them back?
Not to worry! All of your generations can be access through your browser history.
How do I go about doing that?
i realised i wasn't able to access my previous images on my phone, but on a pc all previous images are displayed on the side of the screen
yes! Choosing New initiates a new session and clears the generations you currently have.
Dangit. So no way to recover them?
Not if they aren't in your browser history
Like @short salmon said, you can use your browser history!
They do still exist and you may be able to contact OpenAI to get them to link them to you
Thanks, I saw that. Was hunting them down
Use the Save button to move any images that you actually want to keep to your Collection
A few still exist. I will try and maybe contact OpenAI. Thank you all for your help
There's no guarantee that unsaved images will be saved indefinitely
I'll surely be using the Save button going forward.
Tried a similar picture and it didn't go through. I think you nailed it.
Haha, yes š
I often find that using "movie poster" or "album cover" as part of my description gets me closer to what I want, except that it then includes (gibberish) words in the image. Other than not using those phrases in the prompt (which, again, often helps), is there any way to stop DALLĀ·E from including text?
Hmm, perhaps trying "concept art" might get you closer to what you're looking for?
Thanks, good idea, I'll give that a try!
Is the word explosion banned on dall e?
Itās context sensitive.
Certain things like explosion of Color would probably be okay
Alright so I have a question!
I donāt want to violate any rules and Iāve been searching for a concrete answer but I canāt find one.
I have a few friends who really love to generate images from AI but they only have access to crAIyon etc. Am I allowed to generate an image for them? (I would make sure it doesnāt violate any guidelines) I know accounts cannot be shared (which I agree with).
I have not told them I got invited.
All in all,
Can I generate a prompt a friend gave me then send the picture results to said friend?
If I buy more credits would it just have to be my own money? (I donāt want to make any money off of prompts etc) (Iām asking if I donāt have money and a friend wants a quick prompt is it allowed to get partial financial help from said friend)
Again I really want to stick to the rules and I donāt want to be doing something that isnāt allowed. I am very thankful to have been invited in the first place. If I cannot use anotherās prompt I will continue to test on my own to help the program grow :)
Edit: I'm not getting an answer from anyone and I didn't see anything against this in the Terms of Service so I'm going to assume it's ok? I'll hold off for a couple days just in case I do get an answer saying I can't
I have a question I just created my account on my iPad but when I go to log in on my computer with the exact email and password that is used on my iPad it tells me itās the wrong email or password even though I can login on my iPad using the exact same email and password
double check in your password keychain (if it's saved there) to make sure it's the same! if it doesn't work still, i'd email support.
I have done that already yes so Iāll have to email apparently
I have had some success using the keywords "without text" to keep posters letter-free.
Is it me or dall e doesn't like Twerk dance
somehow violated the rules
"mona lisa dancing twerk" is a red flag
Have a read through the Content Policy 
Since twerking is 'sexually suggestive dancing' the filter will flag the prompt.
https://labs.openai.com/policies/content-policy
@pine void if it is why it is allowed in public? Any dance in some way or another is an act of that kind. Especially tango.
It being allowed in public has nothing
to do with the word being blocked by DallĀ·E.
Simply letting you know why your prompt got/gets flagged 
But this one is totally fine š
In my opinion it can be casted as Political.
#whats-the-prompt message
@pine void Moderation is pretty tough on their side
@pine void Do they have any blocked list so i can check it out. Second strike and they will close the account.
Q: I just got a content warning! What does that mean? Will I get banned?
⢠Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
⢠We chose to err towards our classifiers/block list being aggressive to start, but have been constantly iterating on them as we learn more. Please share prompts you don't think should've been banned in suggestions so that we can improve.
⢠While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
^ @white briar
And no, DallĀ·E currently does not have a list
of blocked words available to the public :)
does dall-e not like Hong Kong? Heres my prompt: "A downtown shot of a bright futuristic Hong Kong at night, digital art"
gives me a warning
Possibly because of recent political conflict and the capacity for the model to be used for misinformation
Ok so nothing china related? That's weird cause I did one with the Russian Military invading Mars and it worked fine
You just need to be careful to avoid globally political buzzwords that might be used to blithely perpetuate lies. The world and its history is not off limits.
ok gotcha
AFAIK running a prompt for your friend is harmless, however be sure that you are not:
- Charging anything for each generation (against TOS)
- Sharing your account or enabling programmatic access to your friends in any way.
If I were you I would double check this via support@openai.com to confirm that it's okay, since this is a gray area.
Why is my thing stuck on this?
Got it thanks! I'll just use my own money then if I buy more credits. I appreciate it!
If you buy credits and then get the 15 free ones once the month rolls over, do you then go through the free ones before returning to your bought ones? Or are you forced to use all your bought ones to get to and make use of the free ones?
Free credits are used first
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
Yes! You have full commercial rights to your images
thanks š
I need help. I just logged into to Dalle 2 like normal but it said invite required and i think my account disappeared and i already paid 15 dollars for credits
Oh no! Sorry about the trouble. Do you mind emailing support@openai.com? That team will be able to help with any issues related to your account
ok thanks
wait nevermind i think i got it back
That's awesome! I'm so relieved for you!
I've had the "invite required" screen appear a few times before, it usually fixes itself with a page reload
I know Dalle has a policy against realistic faces as of right now. Is it circumventing the rules to remove the face from the image before adding it to a prompt? The Original image is a painting of mine and is a completely fictional person. I own the art, the person does not exist. If I were to completely remove the face it would not be circumventing the rule right? This kind of stuff is really iffy sometimes as an artist when you have a realistic art style. I just want to make sure I'm not breaking any rules.
I received a warning for discussing this in a different channel... I just wanted to ask to clear it up whether it was wrong or not
Yes!!
got a warning for typing a celebrity's name, is this normal?
Ok, so I just put in other celebrities and got no warnings, so what gives?
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/
From the content policy:
Ah well must have missed that part, thanks for the info
I thought it was just uploading images
Good thx
hi there, can I generate images in the server?
Not on discord, just the website
-I see, thanks. So users are just uploading the generated images direct, right?
Correct, or by posting the link to the shared creation
and can you explain about Credit thing, how to use it and what are they for?
thank you.
Q: How DALLĀ·E Credits Work
⢠You get 50 free credits your first month.
⢠15 free credits will replenish every month after that, on the same day of the month.
⢠For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
⢠If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
⢠Free credits donāt roll over to the next month, so they expire a month after they were granted ā but youāll get 15 new free credits.
Like so, to get the link https://labs.openai.com/s/Ll0HoHg1ky6zWfuJNJze8inB
Also, the faq didnt mention its 1 credit per generation, edit, or regeneration(4 results)
Sometimes when I write in the description "a photo of", the AI doesn't generate it as one (a picture for example), why?
omg I got the credit thing, m just left with the 36
idk, it also does not understand the "Golden Rain or Golden water Rain"
Hello! The trained dataset may include images where multiple photographs were in a single picture, like family photos, photo albums. D-E 2 may generate based on that interpretation of the data.
Thatās not a really common art image, but it is pretty cool. You can suggest it at #gpt-models message
Yes, but at the same time I would be very cautious with that phrasing.
I'm not judging your intent, @wind sundial but you should be aware that such phrasing is very similar to slang for an obscene act.
Very true there š
Thatās perfectly fine. It is quite sensitive to make sure you remove the whole face including visible ears. The safety mechanisms to prevent deepfakes is still pretty stringent at this phase š
Thank you š
Thanks š
i've noticed that when generating an image of a character that might be holding a gun (like a cowboy or something), the gun will be replaced by a black square with something that looks like a thing but isn't one, is that just censoring in the training data? just thought it was interesting
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 can't say if it's in the training data of course, but they are pretty firm in the anti-violence stance
Question: what is the time period per each dall-e training round (training and teaching dall-e 2) and what are the a acceptance rate of the form that decide what dall-e 2 train?
You might want to have a read of the OpenAI blog as they have a lot of info there eg: https://openai.com/blog/dall-e-2-pre-training-mitigations/
In order to share the magic of DALLĀ·E 2 with a broad audience, we needed to reduce the risks associated with powerful image generation models. To this end, we put various guardrails in place to prevent generated images from violating our content policy. This post focuses on pre-training mitigations,
So can you summarize it? Please?
Or is there a summarize version it
sorry, their blog is quite huge and i can't summarize it. maybe google can help?
Ok
I will use a summarizer AI
maybe ask in #ai-discussions - someone there has probably looked into what is publicly available.
Ok thanks
Hey. is possible to change aspect ratio in the prompt of does have to be done in Post-edit ?
in Mid Journey it's simply done typing : --ar 16:9 for instance
I am just done with some work and summarize it
But still no info on "training round happened every...."
So yeah
Question: Will this ever be free?
That's not something the community has answers to. I suggest you email support@openai.com with any further questions about their training and pricing
Ok thanks
Forgive the noob question, but is there anywhere I can go to get more clarity over the content policy, and more specifically, how that policy is automatically enforced? I've been getting...............at least moderately contradictory results at times. I'd like to understand them, or at least call them out
how do you mean, "contradictory results"? As in, you created things against policy? You can report that - it shouldn't be a problem unless you share it
Well, so for example, take the content policy surrounding violence. "Gun car" is rejected, however images of tanks, which one could arguably refer to as "gun cars", are not. In fact, "Tank firing main cannon" sounds rather extremely violent, but that prompt goes right through. Just created like 5 of them. Maybe this doesn't violate policy, but I can't see how. I'll report and people can do something with it if they wish to
well "tank" is a more ambiguous word than "gun", so it's harder to wholesale block. if i were you i wouldn't attempt to push the filter's limits, as that has gotten users banned and it could make a difference if your account / prompt history ever comes under review
For sure. I'm not trying to push the envelope necessarily, but I've been getting a kick out of just entering random stuff and seeing what comes up. I've had a number of rejections, but haven't been able to figure out why, that's why I'm curious about gaining a broader understanding of the policy and its enforcement
Q: I just got a content warning! What does that mean? Will I get banned?
⢠Please don't worry if your prompt clearly doesn't violate the content policy, but has been flagged by our system!
⢠We chose to err towards our classifiers/block list being aggressive to start, but have been constantly iterating on them as we learn more. Please share prompts you don't think should've been banned in suggestions so that we can improve.
⢠While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
If i want dalle to outpaint an image, just with the AI thinking about how the rest of it should go, should I do inpainting with no prompt or does that break it?
Hi is there a way I can change my email address as I am currently using google sign in. This is so my team can access and I do not have to give my personal details? Thanks
Can I sell merchandising with my generations without the Dall-e logo?
send an email to support@openai.com
You do not have to include the logo. For other rules please read https://labs.openai.com/policies/content-policy
might be better for the #edit-inpainting chat
is removing the watermark via the inspect element still a thing or is it best to just crop it out? i plan on using dall-e for spotify playlist covers (bet u have not heard of that one yet) and other stuff in the future
Yep! But youāll need an image converter.
ok great thank you
Random question: is there a way to see the history on the mobile version on phones?
Yes there is! Try rotating your phone and using landscape mode.
Does violation of policy takes away your credit?
How do you actually remove a payment method?
Hm, I tried that. Am I missing something obvious?
Hey, can I legally create an Instagram page with my ideas and the images created by DALL E? I've seen this a few times now. Thanks in advance
Sorry about that, I got pulled away. Hereās what I see in Safari when I go into landscape mode
Hi, good question. If you follow the rules of use in #gpt-models and #old-server-rules then yes, you can share your results publicly on websites.
Thank you
Does DALL-E has an API?
Q: Is there a publicly accessible API for DALLĀ·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALLĀ·E, which is against our Terms of Service.
Let me guess, Pro Max? It doesnāt show on the smaller Pro size.
Their fault, not yours. No need to apologize š
Switch to desktop mode on your browser
has anyone every printed a dalle image at a decent size (like 3ft by 3ft)? i assume it doesnt turn out super great? any good upscalers that basically leave the image the exact same but increase dpi?