#community-help
1 messages · Page 1 of 1 (latest)
yes i agree too, i have sold few AI generated NFT in past, it was clear IN their policy to create and sell but, i am bit concerned about the DALLE policy as it is really vague but, it does say that we are free to sell the right to the artwork as long it falls under their term like (gore, racism, inappropriate, etc). still a clear stand on its uses should be declare soon.
I AGREE, I see lot of terrible art on the superrare sold for insane amount of money but when you look at their buyers, its the same person all the time and even the bids are placed by the same person and even won by the same person. but still hope we could make something out of it.
@still ice just to be clear: i am specifically asking about GPT-3, their text generation AI, here. for Dall-E its very clear now that it can be used commercially, be it NFTs or whatever else.
but for GPT-3 its not clear to me yet
sorry, didn't know about GTP-3
I think these are the types of things we need to be doing more of. It’s really a shame we have to pay to do them now
Experimentation definitely a less casual enterprise now
I disagree. Here is the text for DALL-E:
"OpenAI owns all Generations (including Generations with Uploads but not the Uploads themselves), and you hereby make any necessary assignments for this. OpenAI grants you the exclusive rights to reproduce and display such Generations and will not resell Generations that you have created, or assert any copyright in such Generations against you or your end users, all provided that you comply with these terms and our Content Policy."
According to this, OpenAI owns all copyright on generations, and grants an exclusive license to us. It does not indicate that we can reassign that copyright, or create derivative works, or anything. Perhaps you can create stuff from it yourself, but then what? It's very unclear about chain of title on anything produced using it, other than they will not "assert copyright against you or your end users"
This is essentially the same as the overall OpenAI terms: "Copyright. OpenAI will not assert copyright over Content generated by the API for you or your end users."
Is DALL-E considering establishing an educational use tier, or a program that facilitates classroom use? I have been introducing art students to various AIs at a university in Japan, and I think it would be very worthwhile to find a way to give them access in this context.
Any news about interface / prompt available in other language like french ?
Any admins that can elaborate on what the plans are for the subsidized access program? Is roll out already underway? Or is the program still in current development?
That said, there is this, earlier: "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."
It's a bit clearer, but still seems a bit fuzzy for my tastes.
Hey now that I can use the images commercially... do I need to attribute dall-e 2 or can just use them?
You can just use them, they are entirely and completely yours
Anyone found a good way to download your entire collection?
Nope
I have a way to do it, but the images won’t have the prompts on them
ah same
Ok thank you very much so we can also sell them digitally as NFTs on the marketplace, that's right.
That is correct!
Can we sell the images as NFTs on Superrare or any other MarketPlace right now? Also the ones we did weeks ago?
Also, can we sell them in an art gallery?
Yes, you have full legal rights to sell them anywhere on any media, including your previous generations
Amazing! Should I include the watermark?
You are free to remove the watermark as well :)
There is no requirement to do so
Thanks!!
Man were there always this many NFTbros here? Really sad state of affairs tbh
Not an NFT fan myself, but I think it was inevitable that people wanted to find an easy way to “generate” art to sell. People will attempt to monetize anything they can :)
They really should blanket ban using this stuff on blockchain "tech"
@neat bridge Hi! sorry to bother, but I have a question regarding if we can upload generations we created with DALLE into Creative Commons for other people to use, considering this "OpenAI grants you the exclusive rights to reproduce and display such Generations and will not resell Generations that you have created, or assert any copyright in such Generations against you or your end users, all provided that you comply with these terms and our Content Policy. " and into this license: "CC BY-NC-ND: This license allows reusers to copy and distribute the material in any medium or format in unadapted form only, for noncommercial purposes only, and only so long as attribution is given to the creator.
CC BY-NC-ND includes the following elements:
BY – Credit must be given to the creator
NC – Only noncommercial uses of the work are permitted
ND – No derivatives or adaptations of the work are permitted
This is a really important question! Personally I already use Creative Commons licensing to distribute public good material and would love to be able to use it for DALLE creations, but the current copyright terms need to be clarified.
Most of my results are always "incomplete" always cutting some part of the object, I see some artwork here and it's always complete and displaying the whole object, how can I achieve this result?
@flat moon
Highly recommend checking out the DALL•E 2 Prompt book. It’s super informative.
Wow! This link should be the first step for anyone on Dall•E,
Thanks!
https://dallery.gallery/wp-content/uploads/2022/07/The-DALL·E-2-prompt-book-v1.01.pdf
Let me know if the results are better. I actually just read it myself. It really should be a requirement before starting.
for this prompt it hasn't changed much, just a better result.
Hm
The other other solutions would be to use the bot or image software to pad the image
And then run it through again
Can you send me one of the images.
I’ll give it a shot
Just post the full image here.
@flat moon
complete image
I took an image from one of the ones I generated, used #dall-e-bot to !pad 40 it
uploaded it to dalle, erased the watermark
This result is good! How did you do it?
and used the same prompt
so paste that image you just sent me into #dall-e-bot with the command "!pad 40" or whatever percentage you want
upload it to dalle, do a lil erase on the watermark, and run it through with the same prompt
Let me know how it works out @flat moon
@short salmon look this: https://www.reddit.com/r/dalle2/comments/w57pws/i_made_a_chrome_extension_to_make_editing_a/?utm_source=share&utm_medium=web2x&context=3
cool
When using the edit function, does the prompt only pertain to the erased area, or is it still using the full picture and prompt to fill it in?
The full pic!
I was wondering the same thing @weak cape thanks for asking 🙂
To piggy back on that question, I wonder if mentioning the main pic explicitly helps, for example, take a pic of earth and upscale, would prompting it “planet earth on a table” work better than “on a table”?
It may add extra earth elements where you don’t want them, hurts me to test these theories tho
I normally describe what I want the whole picture to be
About the 100 credits we get for being in the closed beta, do they expire in a month like the other free tokens or do they expire in a year like the purchased tokens? Ty
The 100 credits are paid. Paid credits have a year. Your free 50 will be used first and last a month
😂
Got a question thats probably been asked many time before but i couldnt find any answers. so why does every image dalle generates has square pixels at the bottom right corner?
that's the Dalle watermark
Hey guys, new to here, I want to ask if there's any other size for the picture that I can do?
DALL-E 2 features higher quality images than the original DALL-E, but there's no way to configure the size, no.
whats the link to use dall.e on mobile?
hello! quick question on public images; is it allowed to aggregate public links that have been shared (to facilitate searching through them)?
if you did want to, however, you could go through the work of manually stitching composites:
#❌┃tips-n-tricks message
that is very helful
thank you
@rare pumice I assume that’s entirely dependent on if you’re giving credit and using them to make a profit or commercial use
Or if it’s just personal use.
thanks!
just aggregating them on a website (non commercial) -- do you know if i would need to get permission and credit, or if just giving credit is enough?
and for credit, would linking to the original "labs.openai.com/s/fsdfdsfs" link suffice?
hmm or if there are existing collections of licensed-for-non-commercial use dall-e generations, i could use those
In general, it’s best to not use images without permission on a website. You should ask for submissions. And to credit someone, a link is not sufficient. You’d need to credit the creator.
helloooo!
I just got the invite, happy to be here 🙂 i had a question about login stuff. I signed up with my google account but now im wanting to change that, it seems like there's no options to change your login/account details at all. Am i overlooking something?
I have the same issue. Unclear how to access account details if you signed up using the Google account option.
Mm. I didn’t sign up with google.
Does anyone knows if the new rules of commercial rights apply to the previously generated images?
According to the announcement, yes:
Thanks I read that, but it says "starting today" so I don't know if it refers to the images created from that point forward, or to all the ones you created, and the rights to commercialize start that day "
this includes images you've created before today
Hmmm Can i like sell what ever they requested prompt is?
Right 🤦♀️
I'm recovering with covid and the mental fog is real 😅😂
I’m afraid not. We had this discussion a couple of hours ago in the mod chat
This works
cool, thanks!
is programmatic access allowed yet? 🤔
does credits reset every month by calendar or when you first got access
For free credits, it will reset/refill on the day you first got access every month.
For example, you gained access July 14th, it will refill the next month on August 14th.
More info here: https://help.openai.com/en/articles/6399305-how-dall-e-credits-work
is GPT3 integrated into dall-e 2?
Is there a limit for how many prompts we can make?
yeah there is
click your user button at top right
it will pop up with the number of credits
you start off with some freebies
and get 15 freeibes a month
and otherwise have to pay money for more
Is DALL-E api not released? They only grant access to their user interface?
no API yet, no
Hi! Found in #faq-and-resources-old :
If you have photoshop, use content aware, if not then try online photopea
Hello! I am unsure what the stance is on copyrighted content apart from the content policy that says "Do not upload images to which you do not hold appropriate usage rights.". I see images with popular characters (such as Homer), so it is allowed to refer to existing copyrighted material? Can you refer to popular culture characters? Known IPs? Visual styles tied to specific IPs? (I think the latter one is fair use). Or is it somewhat of a grey area?
It's definitely a grey area. For non-commercial use I'm sure you're not at risk, but I personally wouldn't be trying to make money off others' IP or style through DALL·E.
Oh, of course, that is out of the question. I'm only interested in seeing what popular styles it can represent successfully.
Hi, I would like to ask what are the commercial license terms of the generated images?
Hi guys, am I allowed to use "Hitler" in my prompt?
It's the perfect intersection of hate and politics, so I'd guess no.
I mean, will it just tell me "you can't" or will I get a block?
You'll probably get a content warning and consistent warnings can lead to account suspension.
Thank you 🙂
Hi 🙂 I have a follow up question, is it still required to credit OpenAI with the new monetizing system?
I've heard different answers to this question but to me this is the pertinent point in the Content Policy:
thank you!
It states 50 free credits until August 19 and 100 paid credits remaining. I was rewarded these 100 for being an early contributor, would I have to pay for these credits?
No, you don't have to pay for credits you already have on your account.
Joanne stated otherwise - it appears that page hasn't been updated.
#community-help message
also: #community-help message someone specifically asked about the web page and Joanne said it will be edited
I realized that I can pass a prompt like "Pikachu riding a skateboard in Shibuya" but I got a content policy violation warning for "A photo of a yellow mouse shooting an electric spark to a monster.
Why is the former allowed and not the latter?
- I was curious to see what it considers as "content policy violation" but it seems a bit confusing from what I see above.
it would be the shooting part
OK, thanks. that said, would there be an official link that specifies all this.
while specific keywords aren't listed out, you can see the rules here: https://labs.openai.com/policies/content-policy
@lyric summit see Joanne’s announcement here: #announcements message
ooh i see
How do I know if I am using dall e or v2?
You're using V2. 100% sure.
what would happen if I uploaded an entire image, erased that entire image, and then entered a text prompt which was related to that image? such as "horse archer" ? Would the resulting generation have any remaining commonality with the image i first uploaded, or would it just start from scratch?
Quick question about commercial usage, is there a way to generate without the five colored squares in the bottom right? I'd like to create some seamless patterns to use in design projects
You'll have to take it to PS and crop
Or any other software, really
tested it just now. result has no relation.
how do you get the mobile app again
thanks
Hello. Can I sell the pictures I created on Dall E as NFT? Will I have any problems with copyright?
Sorry if it was asked already but are there different payment options available mostly for European countries that do not use credit cards? Or is it at least planned to add more option as time goes on?
Install Revolut
What's that?
Can I tranfer money via paypal to that?
From the looks of it I can, sounds good. Are there still other payment options planned? Just curious
Sometimes it is hard to understand the content policy.
The images the prompt produces is against the policy or the prompt is against the policy?
Something in the prompt is
so i inverted that and it didn't like it either
but I put in other random numbers and it doesnt care
I wanted to generate a package design for a fruit beer and i got a content policy warning. I read the content policy and it doesn't say anything about alcohol. it warns about "Illegal activity: drug use" but i doubt beer can be categorized as such. Also i have seen generated images of beer bottle here in discord 😕
I don't know if beer is the issue per se, but certainly you could get around it by using 'brown glass bottle' or things like 'drink label with hops and fruit' maybe.
It's definitely neither beer nor fruit, but maybe some weird combination of the two or another word. Apparently the content filter has a couple of layers and part of it is contextual
I have violated the content policy in my very first query
It was my like, 3rd that violated. Definitely worth reading over it.
i dislike the fact that one of my precious 50 generations had to be the discord verification
what are the uses ? Can I sell the artwork I generated ?
yep
I probably will use it just for creating profile pictures
what does the upload section do ?
my friend accidentally violated it by typing "death star"
I've shared my first in #new-user-creations Doesn't look very realistic. Any tips will be appreciated
Does anyone know if printmaking mediums/textures work? Like block print, screen print, collagraphs etc
Sun rays penetrating through clouds, penetrating word is violation, even ‘blood’ is violating the policy, I had to type Color red water for some horror effect.
Yes! I've had some great results with block printing, here was one for a hooded figure holding a scythe looking over his cornfield, "high contrast" was the key word to get it looking just right
Wow these are lovely thank you for sharing!
"Chiaroscuro" also works really well with traditional print techniques for those nice blocky areas
I bet this is because the classifier thought it was a phone number (we reject personally identifiable information)
oh whoa i just realized that i replied to a way-old question my bad (someone linked me to something written around that time)
It's fine, thanks for the response
this is likely due to a classifier thinking that it's a phone number, since we reject personally identifiable information (other things are physical addresses, social security number, email address, etc)
I've got a question!! Sometimes DE2 just flat out doesn't know of a cultural thing - whether it be an art style, mythological creature, food, clothing item - etc. Is there anything that we can do on our end to teach it things?
ex: it doesn't seem to know what a happi is, the kuchi-e art style, and a sizable chunk of old European artists my art history professor told me to read about (disclaimer: I did not, in fact, read about them)
great question -- let me mull over how best to do this and get back to you!
awesomesauce, thanks a bunch!! 
ok I just made this form and am going to add this to our FAQs in #faq-and-resources-old -- thank you so much for the inspo! Would be super helpful if you could fill it out too: https://forms.gle/kK2dSL4AzqQ2NnjR6
Are there concepts that you noticed that DALL·E doesn't know about, and you'd like for it to?
Please suggest them here -- and bonus points if you can tell us where we can find the data for them!
For example, a user gave us this helpful feedback and we'd love more like these: "it doesn't seem to know what a happi is, the kuchi-e art style, and ...
...wowzers! Gosh, that was super fast!!
I've been keeping a little list of things, I'll make sure to fill it out as best I can. Thank you so much!! ^^
thank you!!
how can you change your openai account pfp? planning on taking suggestions and recording them but the pfp includes my family and my face, which im not comfortable sharing online
It seems it pulls it from your email profile. I use a dedicated email on my domain for dall-e so I have no picture. If you use your gmail account, it will pull from there
thanks, ill change my gmail pfp then
are there any plans to accept crypto as a payment method ?
credit cards arent accesible in all countries
Install Revolut app
BTW Revolut if you're reading this - where's my check??? I've been promoting y'all way too much, and for free?!
lol
ill definetly check it out! thank man ^^
yeah :/ sadly cant make an account on revolut from my country
Please allow cryptocurrency payments! a lot of people live under degenerate governments with degenerate laws
Just wondering if we are allowed to generate images of cannabis plants? I assumed no and as such haven't bothered to generate any related imagery, but I would love to make some label art etc for my medical accessories 😄
Honestly, I don't know the answer. The content policy says no illegal activity like drug use, but I do understand it's legal in many places.
We can see if we can get some clarification though
Yeah for me it's something that is a very healthy and wholesome part of my life and I'd love to be able to generate relevant art ❤️
I can imagine if it's capable of making ciggs, beers, and pills (which I'm not sure if it can or not) then it probably should/could make pretty flowers 😄
Thank you for your assistance, feel free to @ me with any info, I'll make sure to keep some credits aside in case of good news
I'll let you know as soon as I get an answer 🙂
Can somebody tell me why the following picture description is not following content policy?
"A guy completely dressd in black with a torn and ripped cloak breaks the wall in front of him with a a black katana with red lightning around it in the middle of a corridor of an abandoned psychiatric ward. The environment and the characters are very detailed."
It can't be psychiatric ward cuz that worked many times before just fine
I don't quite understand what illegal means for dalle. I can't generate e.g. magic mushrooms? microdosing is a lifesaver for many people in different countries completely legal. The same story with other psychoactive substances or plants. So what should I do?!
@steady agate Try the latin name of the mushroom, something like "Vibrant Panaeolus cyanescens mushrooms + extra modifiers"
Probably due to content policy: drug use
But since its illegal.. you probably shouldnt :p
@vocal forum just tried katana for you - it isn't that. Not sure but I can raise this as a possible error in the system.
Yeah thought so
Might be the psychiatric ward // Nm, didnt read the last part
I just dont want my account get suspended due to errors like that
as I stated earlier, it is not that, worked fine many times before
Yes, I understand the concern on that one.
Maybe the mix of psychiatric ward and a weapon then
I wouldnt think that DALL-E recognizes certain not blacklisted words in combination and sees the combination as blacklisted
Very possible. I know there are multiple levels used. the system is still new though and working out the issues while still in beta.
Ive read it did, where certain words combined didnt work. Saw something on reddit where cannon by itself worked, but cannon + something else didnt
So quick question, am I allowed to utilize my generated pictures as profile picture, banners etc. on social media right?
yes. check the terms regarding the rules on content but you have full rights to do so
Hi, I bought credits so I went pretty prompt happy without realizing some of the images I created (and liked) were slowly disappearing from my prompt history. Is there a way to get these back? (Considering we are paying for these it makes no sense that they vanish) .
you can remove the watermark
you may be able to find them in your browser history, but yes, only the last 50 are kept in the system for now
Would have been nice that this was mentioned somewhere
Check your browser history, you can recover all of your photos.
Unlikely as I've been doing it on my phone
You can still do that 👍
how do I remove the watermark?
And many thx ❤️
any editor can do it. you can either crop or just paint over it. no need to preserve it, but you're welcome to 🙂
Aight, big thx ❤️
luckily I use chrome and it's tracking :p
was able to recover the important images. (Still, don't think a 50 image limit should be in place)
make sure to use save on images you really like. you currently have a 10000 image limit there (not 1000)
Hey, image limit has been lifted to 10,000
how did i miss that? lol - thanks!
I personally just save em locally ^^
if you save in your collection as well, it preserves the prompt you used
i do both, but when i am doing digital painting, i also save on my machine, and in two cloud locations
so, i'm a bit paranoid about losing things lol
Understandable tho, similar here xD
Can I force DALL-Es picture size to something else?
No, only way is trough inpainting. Which also uses more credits.
But can get great results
I read a few times, that uploading a picture with transparency (alpha channel) would work with the editor. Still the interface asks me to first erase an area before proceeding.
Hi, where is the best place to ask about the content policy?
Ask away! Let's see if we could help
ok Thanks! I just created 20+ images of Snoop Dogg,. and then received a warning when I tried a different person. should have read the content policy in more detail, concerning the creation of images of public figures.
also used a bunch of credits and now I'm flagged for attempting to use the prompt with a different person, no ill intent. either way, worth mentioning the warning, and the issue with the content filter
OpenAI is strictly not allowing generations of public figures including celebrities. As I am not a OpenAI employee, the pink names you see around here are volunteers, we cannot guarantee what will happen to your account.
If in case suspension happens to your account, follow the instructions given to submit an appeal.
ok, my question is less about that warning, and I clearly see that in the content policy now as I mentioned. What I mean is that I just spent the morning making many images which the content filter should have prevented? I but didn't. And then that severe warning when I modofied the prompt with a different person. isn't that clearly a filter/software issue?
Indeed I agree that it should have prevented it. It is not perfect since DALL·E has just entered beta. But for now, I advise that you refrain from trying to generate other celebrities/public figures. There are information in #faq-and-resources-old about warnings & suspension which you may want to have a look at.
I appreciate the help. No interest in celebrities, I read some DALL-E article on making esoteric art and somehow ended up with snoop Dogg stained art images ..
hey @severe tendon, so has your account been deactivated? I can go review if so!
Hey, just checked the account is active. Oh, my collection is all gone..
I understand though. And will avoid any public figures in my prompts
huh? this shouldn’t happen — can you try refreshing or contact support?
It was a refresh issue,. collection is all there after logging out and back in
Trying to understand the scope of privacy or anonymity when sharing Dall-E output: If sharing the image link to a generated image from the Dall-E website, can users see other images generated by me?
Nope, only the specific image will be shared
Got it, thank you!
Doesnt DALL-E regognize like brand logos like Discord?
I'd like to use DALL-E to create images that are rectangular in format, is there a way to do that? If not, is there a technical reason that makes this limitation? Thanks for any light you can shed on this!
basically Dalle starts with a 64x64 image and keeps upscaling it and making it more detailed until it ends up as 1024x1024
technically speaking making a different image size is hard because the entire structure is built for exactly one resolution
that said
you can cheat
and generate a 1024x1024 image
then use the 'inpainting' tool to extend it
basically shuffle half of your image off to the right or left
then tell Dalle to fill in the blank spot
Thank you! This is fascinating, I have a better sense of the why, and how to approach what I'm looking to accomplish. all my best– -paul
oh wow - this works fantastic. thank you!!!
no problem
ask away @woven quarry
I have been banned on reddit for posting 7 images like this one I attach
They say me that it infringes the fifth point of Dall-e rules (related to Sexual images), and I want to ask here if this is true for a photograph of a fully dressed woman, just for being seein from behind, I think it's insane
You even bragged there that you have produced even more sexual and NSFW images with bikini. I will repeat myself when I say that common sense is not that common.
Are you really surprised?
I did't say nothing about any bikini, I think you are confusing me with someone else, and yes, I'm really surprised.
And, just for knowing, what's wrong with the bikinis? I'm on the beach these days and I saw a lot.
🙄
I didn't know Reddit or Dall-e 2 was governed by Sharia law
Are you the only admin wich I can speak? Is there any other who doesn't ***** watching woman fully dressed?
I’m not an admin.
You deliberately tried to generate multiple NSFW images to the point Discords filter system won’t let me upload screenshots of your generations.
I’m definitely sure you mentioned even more sexual generations with bikini and “women picking stuff up from the ground”
Nothing personal, but I can’t comprehend how you have one of the most powerful visual AIs in your hand and your go-to is generating soft p*rn
DALL-Es terms are very, very clear and you are trying to bypass them. There’s nothing normal behind your generations and that’s a fact.
Don't know nothing about that bikinis and ground stuff, that wasn't me, but know what? For me doesn't matter, can't understand this limitations and I'll delete my Dall-e account, bye
Please just use common sense when using this technology, they obviously would like for you to avoid using it for anything that could be considered sexual or provocative, hence why they do their best to deter you from doing so. Please don't attempt to bypass it and use it for its intended purpose :)
Yes, generate Homer Vader eating burgers, B-Y-E
There are millions of interesting and fun things you can generate with this technology that no other modern models are capable of, if your sole purpose for using DALLE is to generate NSFW content, you are in the wrong place.
I have LITERALLY show that images IN MY WORK. Who are you to decide that this are Not Safe For Work?
Definetively this is not a technology for me
Im certainly not the judge on whether something is considered safe for work or not, but if a detection system is uncomfortable with it, then likely it could also make someone else uncomfortable
They ban me ON REDDIT, the Dall-e system said nothing about the images. I came here just to así if this is trully a Dall-e policy infringement
And reading both of you, I think It is
Thats because it's Content Policy of OpenAI, the reddit/discord mods are just upholding their sub/discord to the same standard. You can't fault the mods for that.
And boudoir pictures fall pretty close to sexual for some people, so they play it safe.
I want to know if what Reddit admins told me was true, that's why I came here
And I found one man from Reddit repeating same things, wtf haha
Definetively don't know how to use this technology, so I will just stop using it
is there a bot command that lets me add a watermark to this without turning it into a profile picture?
!help
HELP COMMANDS -
Unignores a previously ignored channel
Disables a command, even for mods
Re-enables a previously disabled command.
Marks a command as modonly, can only be used by those with manage server or the mod role
Removes the modonly flag on a command
Restricted commands are a way to prevent commands being spammed in the main channels. When a restricted command is used outside of a bot channel, it will ping the first channel in the bot channel list and print the response there
I have found the bikini user from Reddit, and no, this isn't me:
Yes, saw that. Still, terms of service exist here for dall-e usage. And Reddit mods there are probably trying to keep people from getting their accounts disabled.
I don’t know of one. You don’t have to have the watermark on it, by the way.
And the question is, can I get my Dall-e account deleted for making images of people dressed?
whut. last time i checked the sharing policy, you do have to have the watermark on it when you share it with other people. was this policy updated?
If you don’t follow the content rules, then yes.
That was indeed updated. You have full use and don’t need the watermark 🙂
yeah i know about the commercial use
but i didnt see anything about the watermark
youre right. searching for "watermark" with ctrl+f does not return anything.
wait no thats from march 23
The rules regarding disclosure are still being formulated to comply with ai disclosure laws, but as long as you aren’t misrepresenting where it came from you don’t need to mention dall-e
Joanne is an admin here for open ai.
That should take you to her q&a
yeah i know ive chatted with them before. i didnt see that message though.
thank you for clarifying it for me!
You’re welcome 🙂
never got an answer from them nor OpenAI regarding commercial use of GPT-3 coauthoted work though :(
i think ill send a new email to support today. maybe the first one didnt go through spam or something.
And if I can't ask here, where or who can I ask if this images follows the content rules?
Images considered sexy under normal circumstances are most likely not
(c) Copyright. OpenAI will not assert copyright over Content generated by the API for you or your end users.
I think that means you're fine
thats not what i mean
i am specifically talking about using GPT3 to help write your novel and then publishing it
the sharing terms merely mention that you can publish it. whether this is restricted to only fair use publishing or includes commercial publication too isnt clear
there do exist paid GPT3 generated novels on amazon but for all i know that could have been done illegally
@neat bridge the line:
You may not serve these image generations to others through a web application or through other means of third-parties initiating a request.
has been removed from the content policy.
Does this mean calls to the current API can now be made through a script or application such as a Discord bot? Or is it still prohibited?
if you can publish it, you can publish it, what in their terms makes you think they are restricting commercial use/
well i am not a native speaker but i think publishing doesnt always mean commercially right. like you can also publish something for free (e.g. fair use)?
its just that i wanna be really sure before i get into legal trouble
there really should be some kind of low quality version of the photos so we can try to get our prompts right. i just got accepted and burned through my 50 credits and even bought another $30 in credits... already down to 172
I have the same question here. Can't find any information about it
How do we add this onto our Dall-E 2 usage? https://twitter.com/clairesilver12/status/1551352823392047105?s=21&t=ZKtRGTDkCQ4R9rZ1ZCZ9gA
Hi I was wondering if anyone could help me answer this question please.
Since open ai is giving commercial rights from charging us, are we now allowed to create NFT's or is that still a no no?
Is it possible to share my entire collection with someone? Or would I have to share them individually?
It's not possible at the moment. OpenAI have heard feedback so this feature might come along in the future
Alrighty!
would be cool if they had a tagging feature, so you can share a collection of the same tag with someone. eg, someone can see all your #dogs
Oh yeah! Or just some sort of folders system too
It's fine
Not sure if this is the best place for this but does anyone have an idea what in this prompt is not okay? I even took out the word "dingy" thinking that was it, confused
it might be associating 'long shot' with the act of shooting someone?
guess that was it! i copied that from the dalle prompt book but maybe it's newly banned. thanks for the help!
ive noticed its pretty strict with anything that could refer to violence or weapons, i got a prompt flagged because it was pikachu 'attacking' squirtle with lightning. 'using' lighting was fine though
yeaa, i've noticed that for sure. wasn't even thinking about that cause i just assumed that was "camera stuff" and wouldn't get flagged. gotta be careful haha
Thanks Jeremy!
It's very difficult to produce video game art, or references to video games, which Dalle-Mini was very good at. Have they been filtered out of Dalle-2's Dataset?
does Dall-E understand dimensions? Like If I said "a five-foot-tall-woman standing inside a five-foot-diameter circle", would that make any sense to it?
From what I tested it will approximate the dimensions but sometimes that can fail rly horribly xD
hmmm.... which dimensions work well? feet? meters? cubits?
I tested it with swords around 2-3 meters and 2 meters-ish worked well but 3 meters was weird, DALL-E started putting the sword inside the characters chest and stuff xD
is there a way to see more than the history bar shows you on the website?
At the moment, that's the maximum you can see for the history sidebar. You can however check your browser history for the rest of your generations.
Question: Since people are being banned for using misspelling as filter bypasses, will people also get their access revoked over trying to get around 'neutral' banned words as well? (blood, fursona, fighting)? Thank you 🙂
None of those words are neutral - two relate to violence and one to sex. While I think bans are limited to particular edge cases who routinely try to circumvent the Content Policy, my guess is if you're circumventing those words you're in breach and opening yourself up for termination.
and one to sex
nah hold on there bud.
the word has been banned here for people mostly misusing it for sexy pics, but fursonas and furry fandom in general is NOT about sex inherently. thats a big misconstruction of what a fursona is.
Okay with regards to fursona, would that also apply to terms like “anthropomorphic animal”, if let’s say I’m generating an image that’s related to something in an Aesop Fairytale?
Nope, "anthropomorphic" is fine
Alright thanks
It's maybe analogous to the Philippe Caza issue. His art is not exclusively sexual, but because it's so often sexual DALL·E would create NSFW images on content tagged with Caza, so it's blocked. "Fursona" doesn't need to be an intrinsically sexual word for the model to have gotten that association.
Well, my apologies if so. I have no interest in judging anyone.
Yeah more so just seems like an issue with the AI’s bias and weights for certain keywords.
Sometimes keywords get lost in the crossfire of content flagging, though it’s not too often.
What's unusual is that Gil Elvgren's style is perfectly logged by DE2 despite him being first and foremost a spicy pin-up artist!
it's good that he's not a banned term because his art style is awesome, but trying to render anything with a woman in it is a bit of a gamble because it might end up being racier than intended
Perhaps furry is banned because it has a higher rate of actually depicting sexual acts?
I try to stay in the realms of older European artists for artist-centric type prompts; though I can sometimes go for more post-modern artists
Could give Andy Warhol a shot
I'm not able to scroll down to my older generated prompts under history... is there a way to get those?
Yep! Your browser history
Thanks!
Hello! I'm part of a small team working on a hackathon and we have a cool idea to use DALL-E as a generator of personalized souvenirs. I read on your FAQ there's no API right now, but am wondering if we could talk with someone about perhaps getting access just to test this specific use case during this week.
What if the reason why there is no API right now is because, and get this, there is no API right now? So how could you get access to something that doesn't exist?
There's no public API.
I've found a couple of methods to access programmatically in a few minutes of googling.
It's not too far fetched to believe there is no private API either, specially designed to send a prompt and return back 4 image urls.
Probably the best you could do is a web scrapper library, though I can’t fully say whether it’s against DE2’s ToS to use it or not.
Volunteers here could correct me on that if I’m wrong.
Yeah, I think that's what I've found on GitHub.
And while I understand the answer to my question is probably "no", figured there's nothing to lose in asking.
Hello, programmatic access is currently not allowed. OpenAI has been terminating accounts which have been found to do so.
More info in FAQs: #faq-and-resources-old message
Alright thank you for the correction.
So yeah, no using programs to automate Dall-E 2 creations for the time being.
Exactly. That's why I'm asking for permission. And only to do it for one week.
Is it currently possible to change your display name (on labs site)? Just recently got my invite and in my haste failed to notice this wasn't editable. I like to keep my personal and professional creative separate. Thanks!
I believe it pulls from your email profile (I know that's where it pulls the profile picture from)
Interesting... my profile image is pulling the current one but the name is not.
You may find this resource from OpenAI helpful: https://help.openai.com/en/articles/6402865-is-dall-e-available-through-an-api
Programmatic access to DALL-E
Am I able to open a ticket to request a public name change? Seems like a pretty big oversight to not enable a display name in addition to anything required for legal / ToS reasons by the company (which are completely understandable and I take no issue with). Not mad, just constructive feedback :)
You can try support@openai.com 🙂
Comment: My experience with support@openai.com is that they ignore all emails and ignore all Tweets.
This is definitely not the case. They are however a small team and were recently on two weeks of vacation, so there could be a backlog or there may be some issues too common to deal with via individual emails.
Shot and the meat have triggered this, mine was something like sun rays penetrating through the clouds, I asked them what word triggered it, they said it might be the word penetrate, but they said not to worry if you get blocked, just contact us and we would lift the ban as we will know from the prompt that you didn’t mean it.
I’d say just wait and be patient with them.
I am already selling few just as experiment, the ceo of LinkedIn just launched few today as an experiment to see how it goes but, you won’t make any sales unless you are on premium marketplace like nifty gateway or superrare. There were some who were illegally selling it 3months ago on opensea but nobody did anything.
I have access to DALLE , how can switch to DALLE 2?
It is Dall-E 2 you’re using.
It’s just an umbrella term for it
can switch to Dalle to compare prompt results?
DALL·E 1 is not available and afaik never was outside of OpenAI
Well again the service you have access to is using Dall-E 2’s model. If you want the original Dall-E 1 there are various open source implantations out there.
Thanks
Isn’t Dall-E Mini essentially Dall-E 1?
I don’t think it’s trained on the same training data, just more so both use the same model training implementation.
Found it https://www.craiyon.com/
I don't think so - it's based on OpenAIs research but it's a different model with no connection. That's why they had to change the name.
Ah alright that makes sense.
I cackled when it said “this should not take long (2 minutes)” 🤣 DALL-E has definitely spoiled me with its 15sec generations, I instantly closed the Craiyon site after reading that
Appreciated. Email sent!
Hello. Can I sell the pictures I created on Dall E as NFT? Will I have any problems with copyright?
Can i make a prompt with a company logo in the sentence?
It’s been stated officially that NFTs are fine
Can you send a link?
Is there an official statement?
Joanne is an openai employee. The closest is the announcement that you have full rights to sell.
Understood. Thank you very much for your help. 🙏
Sorry I don’t have better to offer. Beta has been kind of frantic. We are trying to get some better statements devised by official channels
No problem. Thank you again.
Why is it that sometimes "black and white photograph" of whatever whatever returns what looks like a photograph, and sometimes it's like a rough painting? The same can go for paintings. Sometimes a painting by the same artist (take Canaletto) results in impressionistic work (!), sometimes in exactly what it's supposed to deliver.
That said, let's say I got an interesting result but it's rough and messy all over, more a drawing than a photograph. How could I refine it in Dall-E?
That’s a good question. Maybe try to ask in prompt-help 🙂
I've got another question! I'm admittedly not the most technologically literate - what exactly does this mean on the concept request form? As much as I would like to request "make DALLE-2 consume the entire Mario Wiki" so it can learn that Donkey Kong is not, in fact, a donkey... I'm not sure if that's simple!
hey! any way to reproduce characters very similar in whole different context? for example if im trying to do several illustrations of an re-appearing character in different scenes. how to do that best? any tips for me?
I'm not sure if DALL-E uses a similar library, but I have used this tool to test whole or parts of a prompt in order to get a sense of what images are returned and to see if they would be appropriate for my images. This is especially useful if you have terms that are close (like bendy 'straw' vs 'straw' grass) and you want to ensure you get the ...
The weird thing is that Donkey Kong proper does show up in the results here - yet all DE2 seems to give me is literal donkeys
Try “like the famous Donkey Kong” from XYZ
they seem to be filtering out copyrighted characters
https://twitter.com/waxpancake/status/1546939701231095808
I mean MAYBE it's just mickey because Disney is litigous
DALL-E 2's most popular posts tend to be that of Homer Simpson and Darth Vader though - two characters owned by Disney 🤔
Oh, and Kermit the Frog... all of which are (surprisingly) Disney properties!
It also seems to know what Mario, Hatsune Miku, and Waluigi look like for whatever reason
these rule by the way lol
It might help Waluigi in that he is the only thing in existence with that name. There is no other. He has no competition. He is the one and only.
Anyone know if there is a desktop version available? or if there will be one available in the future?
is it just me or are the results terrible today?
is there any way to prompt "no words"?
I shared this to an unofficial dalle server and I wasn’t expecting the screenshot to be spread on Reddit 💀💀💀 thanks for the quick response never the less I hope asking this didn’t cause any issue
I am trying to flag a bias, and the channel is locked. Where should I do it?
It's whatever, just annoying to have to do damage control before my morning coffee 😅 All good now 
Are there plans to make the older generations available in a more convenient way than browser history? Even an archive tab or a way to backscroll further.
You can save any gens you like
@elder dust Thanks is there a batch way to do that? Instead of one by one, a backup all?
Nope 😄
If I want to use an image generated by DALLE do i need to include the watermark in the corner?
Nope, OpenAI offers full rights to the generator
Hey guys. Probably asked before but. The testers as soon as we spend our courtesy credits will have to purchase more?
Or as testers we still have “unlimited” 50 prompts per day?
yeah they went paid
I thought there's no limit anymore.
so unlimited uses per day as long as you pay
for a few hours after they went paid the limit was still in place due to a bug
So no more allowance to the onboarded Artists'?
15 freebies a month
Wow, that didn't last long.
They ramped up recruiting pretty hard at the end
I only got a couple days of the free ride myself
Is there a support channel here?
I need to change the email address associated with my account.
Best to email support@openai.com
Can I get an answer on that question? #community-help message
still prohibited!
Then why remove it from their content policy?
I'd imagine a Content Policy shouldn't contain Terms of Use things, but just content things
actually i retract my statement and i'll check in with staff
Thank you!
no problem 
hi <@&998386466063396966> I was wondering if we can invite friends? I work for film and TV and would like to invite few of my team mates .
can i share the link i got for Dall-E registraion and give it to them as well ? or they need to wait
one user per account.
yes they will have their own accounts
You can certainly forward them the link to the waitlist, and hopefully they'll be on board soon!
in the meantime, i'll leave you this which is located in the #faq-and-resources-old channel
p.s the results are amazing..
sorry, I think something was lost in translation. your account is a single user account.
do not share it.
yup, openai is currently not allowing users to invite others, sorry!
got it. Any way to speed up having my work team in ? or invite other friends... with paid subscription or top up etc like midjourney ?
got it @rough tundra thank you
Have them apply for the waitlist, they are planning on adding many over the next few weeks :)
yes they already are @vivid mortar 🙂
Why does DALL E uses gibberish language ?
I can’t make it to use English in my generated images 🤷♂️
"It wasn't trained on text" is the short answer.
Some very common phrases work well, and occasionally you fluke into correct text. But the short answer is a visual model like this requires many billions of images in the training model until it can start to represent specific words given to it correctly, much less invent correctly spelled / phrased / meaningful text on the spot.
Does anyone know what the maximum size of an image is - is there a way to make the images higher resolution?
All images are 1024x1024
But, you can Composite multiple images together to make bigger ones
Question, did the verification pic count against my extremely limited total? D:
That's almost like if discord support made you buy a server boost to access it
Yeah, it used ~0.12c worth of credits... Unfortunately the only way until OpenAI provides a built in varification
I could literally build them one right now and it won't cost me twelve cents.... And I'm no wunderkind programmer or anything, not like these guys
Does anyone has a list of possible slots for "painting styles"?
There's no list. Some extensive resources have been put together but we're discovering new styles every day.
Basically, want to see something in the style of Johan Barthold Jongkind? Try it and see what happens (shrug)
Hi! So, I'm a game dev, can I use the images from Dall-E into my game?
Yes!
Yes!
Nice! Thanks 🙂
(these answers are both according to the latest use agreements!)
Keep in mind however there's a pending update to the terms of conditions and currently you may still need to disclose the role of AI in the production of the art. I believe this relates to e.g. EU law.
Hi, I read the content policy but have a clarifying question. Is it ok to create images of movie characters? Frodo or Neo for example. I can imagine it looking like particular celebrities, but that's not necessarily the intent.
Can we still use Dall-E logo on them?
It's commonly accepted practice, go crazy. I personally wouldn't try to commercialise any images clearly derived from others' IP though.
That makes sense! That's exciting, thanks.
Sure, why not?
I don't get taken down for using their logo
Nope, that's how they provide the image to you, I'm sure they'd love it if you kept the attribution.
Btw I used "Robot with futuristic super weapons" and it warned that I might get banned for violating terms
@past geyser
Thanks
Guess weapons is the word
The content policy makes it clear that DALL-E is for G-rated purposes and that violence and weapons are not intended to be created by the current model. "weapons" probably triggered the flag, but don't stress, it's a fairly benign prompt and warnings will only affect your account status if you repeatedly try to breach the content policy.
Thank you
any idea why this prompt counts as a content policy violation?
"the destruction of the earth in the style of zdzisław beksiński"
also on a related note how many violations before suspension
Destruction of Earth ig
"destruction of the earth", probably - it's a context-aware filter. there's no number, it's about patterns of use
It's also possible that DALL∙E flags zdzisław beksiński as a policy violation (NSFW). Judging by the results in a quick google image search.
nope, just tried it again without the destruction of earth and with the same artist and it worked fine, with admittedly not great results though
beksinski has been a topic of discussion, his style definitely works (in a limited manner)
Yeah, I got the sense that it is only trained on a few of his artworks and they aren't necessarily the 'iconic' ones
Ah, I see. Some artists get flagged though for sure.
Hi, I just started Dalle today and was loving it, and because of the strict guidelines I wanted to be extra careful I didn't violate them. About 100 images in I saw a "tip" when one of my images was loading - the tip was along the lines of "try using abstractions" and it had an example of a mug that looked like a face and other materials - in the prompt it use the phrase "looks like". So I tried "red head that looks like a shoe" and it flagged me with a warning. Was this a mistake?
Or is it actually something I can't say because it supposedly compares the group of 'red heads' to a 'shoe' or 'humans to shoe"?? Are bans permanent? If I get banned will I be refunded for all my credits? I purchased 300 dollars worth. I was really surprised when it warned me and was looking at the prompt for awhile dumb founded.
So after carefully reviewing the guidelines I think it gave me a warning because I was supposedly comparing either humans or red heads to shoes which I guess could be viewed as derogatory? I was just trying to make surrealist art and my intentions where not in the slightest towards that angle...I am not sure if it was automated warning from a bot, but I hope that warning doesn't tally against my account at all and OpenAI can get a human to review it and take that negative tally away from my account. I spent 300 dollars on credits and I absolutely do not want to violate anything intentionally, it was a mistake because of that "tip". Can anyone help me or reassure me I am good? I just do not want to worry about it. After the warning I just completely stopped my flow of making images out of worry it will automatically ban me. I wonder if I just said a "face that looks like a shoe" if that would have been ok, but its still comparing a face of either an animal or human or being to a shoe so, I don't know, any advice on what I did wrong and how to avoid it and can someone at OpenAI take that warning from tallying against me. I really don't want to lose access.
I was looking at one of the papers on limitations today and it specified how early users would get around the prohibited word "blood" by saying things like "red liquid." There's thus additional layers of content aware filtering that might have been tripped up here. Don't worry - your prompt was harmless and only repeat patterns of obvious attempts to circumvent the content policy would actually trigger an account suspension.
Has anyone been able to get midjourney styled artworks using dalle? I can’t even seem to get that level of stylization
My suggestion is to get a couple of example images and share them in #❌┃prompt-help. I've seen art here that imo blows Midjourney out of the water, but it might depend on exactly what you're going for.
Ok that makes me feel a little better, thanks. So you don't think it was the comparing of 'the person to shoe', it was more so the use of the color red? So what things can you compare humans to that are acceptable? If I get banned from a wrong automation is there someone I can contact to get it reviewed and potentially reversed? I just know I have no intentions to break the guidelines so I know any human review of my prompts would bring to light there is no wrong intentions - I am just scared of the bot banning me for something and making a mistake and then having no way to use the service again.
Also are bans permanent? And would my credits be refunded in that scenario? Would I get a human review instead of automatic bot ban? These are all my questions and concerns. I am assuming you are from OpenAI or are you just a member of this discord? Apologies for long message I just want to get this squared away so I know what to expect.
This was the prompt it warned me for: "high quality photo of red head that looks like a fancy shoe"
For sure! Don’t think MJ is better by any means; I just think it excels at one style. Will share some images 🙂
I'm just a Discord volunteer but I've been around for a while and these questions come up often!
I'm not 100% sure of the process but I believe account suspensions are all manual and human reviewed. I've definitely tallied up my fair share of warnings 😅 And don't worry about abstract comparisons of human features to non-human things, it's not a worry. Lemme see...
ok thanks
Subtle rephrase got it through, but I think "head" was too inspecific 😅 https://labs.openai.com/s/C7XBFKtwR0lGBesAju1rOZk6
So like saying a woman with red hair would have been ok, but red head implies a bloody head? I was referring to red heads like orange/redish haired humans haha but I get how the bot may see it as too unspecific.
Some questions:
1.Are bans permanent? 2.If banned can you get refunded for credits?
- If you're seen to be incontrovertibly and consistently breaking the content policy, yes. 2. Dunno. Doubt it.
Thanks for the info and bearing with me, is there a support email where I can contact OpenAI directly if I have further issues?
Yep! support@openai.com
Thank you.
Also was wondering about the words skull, skeleton, zombie, ogre, ork, and others along similar lines as these, are these all ok or no? I noticed monster and creature is accepted, I am not going to try these phrases unless I know they are ok, can someone let me know if these are safe? I am guessing zombie may not be safe word to use because of bodily fluids/gore?
how do i change my profile pic to something else other than default first initial
I was wondering if anyone could point me to what was flagged in this prompt “Hand drawn sketch of a Nigerian Dwarf Goat climbing a Jeep in the African Savanah” I was guessing maybe the dwarf part but I am not sure
does the capitalization of letters in prompts impact the generated image?
This normally pulls from your email
Nope. Punctuation can affect it though
I would report this in bugs, as that’s a breed name. This worked and not sure if it’s close enough. https://labs.openai.com/s/PxZXFjwXZMir0wfsqD4m4COc
Orc, ogre, and zombie I have used. Bear in mind that the system will analyse the whole prompt in a lot of cases so making those eating brains etc would trigger and also violate the terms of g-rated, no violence. Using the content guidelines will keep you safe. A warning won’t consume a credit and as long as you aren’t trying to circumvent the rules you’ll be ok.
Redhead all one word is fine 🙂
Hey, is there any way to change the aspect ratio through prompt or other method? Like Midjourney you att an --AR or whatever... Thanks in advance!
Right now it’s only 1024x1024. You can see the inpaint channel for examples of how to splice images
Glory, thank you kindly!
@rugged wing
Hi! I am having fun with Dall-E 2 and want to build applications on top of it. I noticed that its API isn't documented yet and the only way to generate images is using OpenAI's web interface. Is this because OpenAI doesn't want developers build on top of Dall-E 2? <@&998386466063396966>
Correct.
OpenAI allows commercializing the images according to the announcement though (https://openai.com/blog/dall-e-now-available-in-beta/). Is this limited to the images that are generated directly on OpenAI?
Commercializing the images = you generated it with your account following the ToS, and download it (again following the ToS)
I do not know what you mean by "generated directly on OAI"
For example, one could generate the images on a website other than openai.com
I'm curious why OpenAI wouldn't want developers to build on top of Dall-E 2. Is what you provided an official answer from OpenAI? Do you know where I can find the TOS that specifies that by any chance?
Feel free to reach out to @neat bridge directly with your question.
For terms >> https://labs.openai.com/policies/terms
I can speak anecdotally regarding why developers wouldn't be permitted. This is from my perspective, not OAI.
DALLE is still in beta, the OpenAI team is fine with you selling images that YOU create, but they do not want you to sell ACCESS to generate images for some sort of markup by using the API you find via web requests.
If you want to go ahead and create a product like that, just do not release it until the beta has concluded and the OpenAI team gives the green light for those sort or projects.
How does OAI ensure that the developer follow their standards? Can they charge for the service?
In time all of this may change. But it is a beta.
Thank you @vivid mortar and @lucid ravine 🙂
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our ToS.
@neat bridge 😛
So, are furries in general banned, or just trying to make sexual images of them? Also, what counts as a sexual image? Is an exposed chest sexual? I want to be sure not to break any rules
Hi, do we have the right to remove the watermarks for commercial use? EDIT: I think I found the answer using search. You are allowed to. Please correct me if I'm wrong though!
It's fine to create anthropomorphic animals. What counts as a sexual image is anything that's not G rated, as per the Content Policy. If it's too raunchy for a Pixar film, it's probably not how OpenAI want their technology represented.
When payments from Ukraine will work?
Hello there. I just started to use DALL-E today and I'm a little scared: I know it can't receive inputs with real faces and even in few some pics I've tried that were not accepted I didn't sent any pic from myself.
Now I was generating variations os a photography of a woman and one of the pics resulted in a very impressive lookalike face with myself 😲 . Would it be just a coincidence?
Hi, is there anyway to get images back from previous sessions?
If you are having troubles with payments, I saw a conversation regarding a program that will exchange currencies for you to purchase credits, I will try to find it to link again
On desktop, the right side bar contains a number of your recent generations!
I accidentally started a "new session" and those were all cleared :/ is there any way to see them again?
check your browser history. old generations all come with a URL, if you just go to that URL again it'll still show up
Dang
Yep it was all there 🙏
Thanks 😄
Could even go back and re-save everything so that it'll come up in collection
When you ‘publish’ where does it get published to? I don’t see a feed to view other peoples images anywhere, so where is the best place to view other users images and prompts?
up to you to decide where it's published. you only go there from having the link I think
Gotcha, so its just like a share link? So its not literally publishing to like a ‘wall’ like you would see on Instagram? Is there a place besides discord where you can just scroll through Dalle images created by other people and see their prompts?
reddit.com/r/dalle2 is one place that comes to mind
although personally I feel like the channels in this discord are somewhat better
the reddit thing has a lot of submissions where people just don't post their prompt so I think it's somewhat useless
Ok I will check that out too, is New user creations the place to see most of images here on discord or is there a better tab? Yeah I def want to see the prompts, thats the fun part
not sure about that myself. haven't checked out most channels here since the credit system was introduced
Do commas vs. periods matter? I've seen it asked a few times but no official answer.
Punctuation does matter, to what extend I'm not sure.
Right, but I mean are periods and commas treated the same way?
there's a difference observable
Is it more different than randomly regenerating..?
apparently so
how it exactly works is hard to say. the thing with machine learning models is that most of it's decisionmaking is being made inside of a blackbox after all
Hey guys. Got a quick question. I generated a couple of images but I didn’t download each one of them and continued to move on to generating the others. How do I download those other ones?
You can flick back through your history on the right side of the screen
Browser history is the only way if it's not in your history.
Ahh alright, thanks!
Still would love to be able to browse full history, but I don't think it's OpenAI's priority.
yeah. searchable by prompt too 😍
Hi, I got access to DALL-E today and I clicked a button called "variations" in one of my result images, and dalle instantly spent 2 of my credits generating 8 more images that I didn't want to. I thought variations was to go back and show the other images I generated. Is there any way I can get my credits back? I don't care if the images are deleted, I just don't want them spent with something I didn't want to
hey! i read through the terms of use a couple times, but just wanted a quick clarification. the terms' section 2 says "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." would that commercialization extend to perhaps an album cover artwork as well?
I’m assuming so
I don't think you'll have much luck arguing for a two-credit refund; ostensibly the free credits for all new users covers learning the basics of the functions.
What’s breaching the policy here? 
"dead"
Wha? Weird. Alright. Would zombies work?
AFAIK yes
I removed 'kill' from the prompt, what's still preventing it?
DALL∙E does not only block words as is. It also uses contextual understanding. Maybe it still knows it as is -> "Berryman-Langford Memetic Kill Agent".
Also, this is definitely against the Content Policy as the output would be non-compliance to it.
Makes sense
Wait, why the Content Policy violation? Would it be counted as an (albeit fictitious) weapon?
Try changing agent
Nah I'll just skip it if it'd be an actual violation
As per a quick search, this came up:
And as per policy, non-G-rated generations are not allowed.
Aye, fair enough
What channel are people talking about pricing in the most?
#ai-discussions is a good place, as we brought it back specifically because it was overtaking some of the more formal feedback channels. Especially if you want to discuss e.g. the competition at length
just ask for a fractal parrot
Is there a way to see older generations? I have only made 49 total generations but I can only see the last 20 and some of the early ones were really terrific.
it should show you the last 50
if it doesn't you can use your browser history
all the stuff is saved on the server currently, but the UI just doens't let you access it, they're aware it's an issue but
Hi! Try refreshing your browser
no word on what's gonna be done
You genius, you saved the day. Thank you!
no problem!
Hey, quick question, how do I get to the dalle website, the one where you enter the promts, I've somehow managed to loose the link?
Thanks
question - is posting a photo of myself that includes my face counted as rule 5?
Hello, good question! Technically it is, but let me check with the team about this! Kindly refrain from posting it for now.
thanks! i personally dont mind sharing my appearance on the internet but figured it probably does fall under that margin.
thanks for checking!
Thus far I had 2 violation warnings with prompts. The first was for "A blooming orchard by Camille Pissarro". Surely your system didn't stumble over the first 4 letters of the artist's name? The second was "a detailed photo of a dystopian metropolis in hell". Was hell really the culprit? I'm a bit hesitant to experiment in fear of getting a ban for no apparant reason. More detailed feedback on the reason for a refused prompt would be most helpful.
The filter has a couple of layers, including contextual AI; it's not always an obvious answer to what triggered it, even to the AI itself. "Hell" was prohibited but theoretically has been 'unblocked' at the direct filter level. As usual, don't worry about a warning or two if your habits aren't genuinely attempting to circumvent policy.
Is it ok to make another Dall e account when you lost your account
"lost"?
I wouldn't think so.
Why tho
Access to the DALL·E beta is currently a limited privilege. If you've "lost" your account it suggests maybe you won't use the next one reliably...?
Eh
Probably not
Because if that is the user that lost a account will prob learn to treat their account more like
Good
Why am I getting a dialog box asking me to login when I am already logged in?
Hi! What's the link in the address bar?
No idea. I tried entering it to find out & Safari can's find the page. My best guess is that some internal DALLE process was timing out?
is on the prompt, :443
Probably, just making sure it's not phishing-related. Let us know if it's still happening!
I did not try filling in the boxes.
Is it still happening? Does it happen with another browser?
It's not happening now.
Cool. Let us know if it happens again, thanks!
Is there a way to get better than a 1024x1024 image?
Hello, any idea why this prompt might violate the policy?
Maximum for DALL∙E 2 at the moment is 1024x1024. But there are ways to get bigger results, check the #inpaint-outpaint & #edit-inpainting channel 🙂
Replied in #❌┃prompt-help!
Hey everyone, yesterday I got access to DallE but the only way I can access it is by signing in with my google account. and when I ask to send me a request to change the password to my email (Same as the google account one) I never received
You may want to change your password via Google.
As per: https://help.openai.com/en/articles/4936828-how-can-i-reset-my-account-password
how can i see what kind of art styles Dalle has
Hi, I'm contacting you in regards to a clarification about the TOS of Dalle, we're doing a law-tech project and are looking to use Dalle instead of stock photos for our presentation images. We know you're being strict against TOS violations due to the risk that not filtering powerful models such as DallE 2 create.
Would prompts such as this be fine? We don't see any issue with them, but just wanting to make sure:
"Gorgeous intelligent female lawyer in her 20s presenting tech project on her laptop, 35mm camera full body photo"
"High quality FS photo of a beautiful eastern-european female lawyer in her 20s with makeup presenting tech company on her laptop"
Prompts looks great! I recommend checking out the resources at #faq-and-resources-old. It will help you engineer your prompt to your liking.
Thanks!
Do you have any plans of using a finetuned GPT-3 model to "optimize" prompts for easier use by non-tech users in the future? I see large potential with DallE
I'm sorry I think you mistook me as an OpenAI employee. The pink names you see in the server are community volunteers. Yellow ones are OpenAI employees. I'm afraid I don't have an answer to your question 😅
So, are there any other payment methods planned? PayPal? iDEAL?
No heads-up from the team yet! But others have been purchasing with Revolut, heck, even crypto.
there's any risk of hitting the content police when trying to generate sword or knife, like ||a knife with the blade made out of fire opala, studio lighting, Sigma 85 mm f2.8||?
I was curious about that. I tried it for you and it did not get flagged.
thanks
Hi. I need the high quality images. I need at least 5000px without the color boxes at the bottom. I am happy to pay to download this ultra high quality images. Is it part of the roadmap?
I don't think we'll see anything generated without color boxes on the bottom. Maybe in the future for certain higher-tier payment plans, but I would think that's a long shot. (disclaimer: i'm just a DALLE user, not volunteer or an OpenAI employee)
I am happy to pay extra for high res images. min of 5000px. Just please don't make it too expensive. lol
Does anybody know if melodysheep is an artist DALLE knows?
@glad pilot I cant speak for the roadmap of the product as we are just volunteers, but these sort of features are highly requested and seem possible with the current model (upscaling may drastically increase cost as pixel count increases, this would be a question for joanne)
But I think the removal of the colored boxes will likely be an option in the future as you now have commercial rights to the artwork and its been stated that you are free to remove them. I would think this would be a toggle in the future
There is a user in this server who is named melodysheep, yes
is there a better way to emulate a style than just trying to emulate it as best as possible with a prompt? im trying to design some icons that look like these from google maps without burning through too many credits
does DALLE know different languages like dutch?
I've tried lithuanian and it doesn't exactly work lmao
Countries are generally not allowed i think thats why its blocked. (Its not because of the “mad” because i used that fine on other prompts)
No i tried dutch too you will get interesting results but not what you are looking for so i always make sure now my prompts are in english
If you put these that you have and leave some transparent space you can use the edit function and tap once in the transparent space. Type your prompt what it has to add/create and it will try to match that to your picture.
Nope guess you just have to try it. There is some people writing some things down that work and don’t but i dont remember where
Is there a way to output 16:9 images?
I am trying to get dall-e to generate a cyan 3d-rendered game controller above this stand-looking thing it generated but can't get it to work. Anyone know how to write the prompt?
why not upscale it yourself?
Do purchased credits carry over month to month?
they expire 1 year from purchase
hey, I just got approved and am experimenting to see how the AI responds to elaborate, surreal prompts--nothing inappropriate though, as far as I can tell. I put in the prompt "a cereal bowl containing thousands of tiny Tesla coils floating in milk. the tesla coils are shooting multicolored electricity all over the place and some of the milk is charred. there are little fires here and there. the spoon is floating over the bowl with some electricity reaching up and touching it." and got a message saying it might violate the content policy and that further violations might result in my account being suspended.
am I running afoul of some rule here? how worried should I be about getting that message? I don't want to get suspended when I just started--thanks!
Hey, how do I add a password to an account I created with Gmail
So that I don't have to sign in with my Gmail account each time
"shooting" is banned
ah thanks! I thought that might be the case but didn't want to keep trweaking something that could get me in trouble. I thought I had used "shooting" earlier re: a pinball machine shooting pinballs all over the room but I had ironically used the more violent-sounding "firing". it must be taking "shooting" as a noun.
"launching" worked. thanks again.
Hey guys, something i noticed and am wondering about: I’ve had a couple of images generated in which I’ve tried to add mushrooms and it seems difficult to get Dall-e to put them anywhere. Do you think there might be some restriction on the word ‘mushrooms’ because of psilocybin?
I got flagged using the word “weapon” but I think specific weapons are generally not flagged as long as they’re not being used for violent acts
Hi where can I get the content policy?
I've done like 100+ mushroom images with no trouble! There's definitely no censors around it. Do you mind sharing your prompt or what specifically you're doing?
It's in the dropdown: https://labs.openai.com/policies/content-policy
Yeah, I’m trying to generate things like bowls of mushrooms or add baskets of mushrooms in an image
I do see them sometimes but generally not when I try to use them in a locative sense — put mushrooms here, or have them growing from this
Yeah, it doesn't always go in the place you want it to. This is a bad example, but I've done a basket of mushrooms and it understood it well (low quality generation thought): https://labs.openai.com/s/mlfz7q3UPFFK2t1F5ZyYnydq
I know it sometimes has trouble putting something IN something else, like a bowl.
It sometimes helps to switch the order if it doesn't get it, like "a bowl of mushrooms" might work better than "mushrooms in a bowl". Maybe "a bowl filled with mushrooms" too -- not sure.
Is there a limit on how many photos can be saved in the 'My collection' section, either total number of saved images, or perhaps an expiry date?
https://labs.openai.com/policies/content-policy Please also check our #old-server-rules channel for the rules here.
Not that I'm aware of, but I'll double check
just checked, there is a limit of 10,000
Awesome appreciate it JPC!
How do I get better resolution?
dalle will only genarate 1024x1024 however you can composite images together to expand them (effective better rez)
do a pattern like this then edit them together
How do I do this? Is this on Dall-E?
Thank you for reply
no you use a program like photoshop or paint.net (some functions can be done in #dall-e-bot ) and leave some transparent area for it to fill
i said to someone else here #new-user-creations message
Thank you so much
How do you get Dall.e to put specific words in the image?
You can specify where you want the words, like a sign that says "CAT" (put the word in quotes), but Dall-E does not produce legible text. It will scramble your words. Sometimes it will get it right if it's a short word, but 99% of the time it will be scrambled.
You can also just say the word "CAT" and it will find a place to put it in your image.
Okay thanks.
Did anyone was able to create a shitzu dog using dalle is seen like it doesn’t recognise it?
If I buy credits while still having free credits, will the free credits be consumed first or the bought ones?
Are all weapons banned on dalle-2?
I generated a capybara dressed as a 1920s mobster, however I really wanted it to hold a tommy gun.... 😭
I kind of understand why it would be banned, but c'mon.... is their any real harm done by adding a tommy gun to this image? 😆
Their needs to be some kind of flexibility to the rule surely...
Are we allowed to generate images with a human show/movie character?
or is the rule no human-like characters/celebrities at all
It works for things like swords but I'm not sure about guns.
Guns are definitely not allowed. Though I would like to see capymobster with a gun.
Me too @astral geyser.... Me too..... 🤣
Yeah, characters are allowed. Humans can be generated, but you can't upload them.
ohh ok
so then what's this about?
I was gonna use gordon ramsay in a prompt but i think it violates this one
Oh I mean, you can't try to generate real celebrities. Gordon Ramsay wouldn't be allowed.
Like I think Walter White would technically be okay since he has a specific look/style, but Bryan Cranston would definitely not be. Stuff like that probably gets a bit gray. Not sure.
That's how I've interpreted it. I think the intention of the rule is to not use it to generate fakes of real people or to use it to manipulate/defame people.
Hello, I'm wondering if there is a way to share access to generation capacity to another account? Like setting up an owner/reader system for an organization.
Once again with the weapon question, would brass knuckles be okay to use on my mobster capybara or would that be a "no no" too? 🤣
I just want the image to illustrate that this is one mean capybara not to be messed with... 😆
I feel like wearing brass knuckles might be okay as long as it’s not being violent with them? But idk. 🤣 Like it seems similar to a sword, which Dall-E does not restrict.
No, they just have to wait for access.
I just gave it a go, it allowed the generation, but alas... no brass knuckles were generated...
I guess the world just isn't ready for a threatening looking 1920's capybara mobster....
But one can dream though..... one can dream.... 😔
looks off into the distance
What happens if you just use an emoji in the text for generation?
@astral geyser Thanks for answering my question!
there's actually a youtube video about it!
https://youtu.be/HyIqzMswEJc
In this video I challenges OpenAI's DALL-E 2 to turn simple Emoji Prompts into beautiful artworks - and it WORKS?! - let me know what you think in the comments!
➨ Get my APP, Courses, eBooks, Brushes and more:
http://www.jazzastudios.com/Shop.php
➨ MERCH - Shirts, Hoodies and more:
http://jazzastudios.com/Links/shirts.php
➨ ART PRINTS and PO...
why does this trigger sensitive content
“first person view through a republic commando lenses of the battle of Geonosis, looking down from a Low Altitude Assault Transport, Star Wars, 4K”
the word battle?
I'm guessing battle, maybe assault too...
First Person view through a Republic Commando Helmet HUD of the planet Geonosis, looking down from a LAAT gunship, Star Wars, 4K
this one worked but results were aweful
Hello! I just received my long awaited invite to test out Dall-E and im wondering if what I'm testing is Dall-E 1 or 2?
Also does anyone have any links to a guide to using it? With different tips and instructions with all the possible tags like "3d render or digital art"?
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
Great! Thanks so much.
A tip that worked well for me is to stack art styles, like for example "digital drawing fantasy realism detailed art"
I've got some awesome looking results, just be creative! 🙂
what is dall e bot?
Do check the resources at #faq-and-resources-old 🙂
Also what url can I use to get to the website to generate? I can only seem to get back to it by saving a link I generated and then clicking "back" button within Dall-E to be able to make promps again. For example www.labs.openai.com just takes me to the Dall-E website talking about it where you can explore and sign up for the waitlist. Sorry for all the questions! 😅
Oh, neat! Thank you 🙂 I wonder if there is a list of all possible styles/tags that work or such.
if there is, I'd be interested in looking at that list too! 😀
The official website is labs.openai.com — you may not be signed in if it's asking you to sign up.
No official list, just community compiled ones at #faq-and-resources-old 🙂
I have a question, with #suggestions, is that just for the discord, or is that suggestions for the development team?
Every suggestion I've seen in there has been for dalle2 and the web app, not the discord
Ok cool! Now for a follow up question, do you think they actually read them? 🤣
I'm just having a bit of a joke, but it'd be good to know if these suggestions are actually getting across to the right people. 😀
Suggestions that get 10 ⭐ reactions and are added to #starred-ideas are more likely to be read. Since the influx of new members. See Joanne's message in #announcements
#announcements message
They have read them, and generally do read them
ohh that's awesome! thanks for the reply! 😀
Feel free to suggest anything DALL∙E 2 related, whether it's for the dev team, or for the discord!
And laze is correct! To add, if we (volunteers) think that a suggestion is crucial and will add great value, we tend to re-send them directly to a channel that will surely get read by Joanne / any OpenAI staff. Else, the community can voice out their opinions normally through #suggestions and better yet, getting their suggestions starred that would then be added to #starred-ideas as laze mentioned.
That's great! This seems to be a pretty solid community! I've been having a blast, not only with Dalle2 but with the discord as well. Thanks for the replies!
You're welcome! Make sure to scroll through the channels for some amazing DALL∙E 2 art. I recommend the #daily-theme channel for a start 🙂
I'll defiantly check that out! Thanks!
I didn't see anything about attribution in the terms of service for Dall-E but I noticed that they do have their logo in the bottom right corner.
Since we have full commercial rights, can this be removed?
It would be nice to have a way to remove it within the tool as well.
is there a way to add words to a prompt for reference purposes that dont have an effect on the image output?
As per an OpenAI staff: #community-help message
I believe they have plans to improve on the Edit feature of DALL∙E 2, stay tuned!
This thread might help: https://twitter.com/davidad/status/1551143240065228800?s=20&t=-KXmPGeEVbGrguH_baw8ug
Specifically the second tweet.
Someone else asked 🙄. I should've known lol.
Thanks.
Why did this prompt get flagged? I'm pretty sure it doesn't break any rules
I don’t think I’d actually do this if it were allowed, but could someone theoretically sell access to prompting through their account? I was thinking that there are plenty of people who want access but don’t have it, and might be willing to pay to have someone else who does have access input the prompt they want. I’d assume it’s against the rules, but seems worth asking
No, this will get you banned from the platform if you are found providing this service
Wow, super weird. I wonder if there's any association with the war.
that's what I think too, but it's weird.
No need to fear about being banned for these sorts of things, if you obviously mean no mal-intent with your prompts, you wont be banned for false flags like this :)
thank you. I didn't even think about the current state of things, just wanted to see some landscapes of a city most familiar to me
Hey, nice to meet y'all I was wondering if there's any way to change the canvas size of the AI, i mean not the 1:1 size, like a horizontal one
not without external compositing. there are some helpful functions in #dall-e-bot however
oh, ok, thank you for your kind and fast answer
Someone just tried to sign into my account
how do I remove my card from it
<@&998386466063396966>
@rough pumice That's a serious issue. Unfortunately the volunteers don't have access to OAI's tools or backend.
yes, thank you lucas.
was just about to paste that.
no problem!
bless
Hope everything turns out okay!
If you ever want to change your email just email support
That might help a lot too
I've got a fair few warnings (stuff like laser guns, a fight between cartoon characters, and stuff for friends like "Kanye East", ethernet cables attacking people) that have all triggered the warnings. Nothing seriously against the policy but I do sometimes forget to word stuff right. Am I in any danger of being banned for prompts like that, especially if it's automatic?
Nope! as long as its not intentionally abusing (generating NSFW content, typo bypassing, etc) you're completely in the green
It's probably way too early for that but are there any plans for a dall-e API to develop apps with? Or is that not going to happen because of how hard it would make content-policy policing?
Q: Is there a publicly accessible API for DALL·E?
OpenAI: We do not at this time, but do have a longer-term plan to offer this. Please also do not attempt programmatic access of DALL·E, which is against our Terms of Service.
Hi, is the current model Dall.E 1 or Dall.E 2?
Hi, for doing verification for discord, my one credit got consumed. Can that be refunded ?
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
nope. one credit is well worth it for the generous display of art and the combined knowledge this community offers, imo 😎
Hey all, I'm having a bit of difficulty generating an image and keep getting warnings that my prompts potentially against the content policies.. I've checked and can't see anything I'd be in violation of, but it seems really consistent.
Is there something I'm missing with trying to generate a Fantasy portrait of a humanoid Tiger holding a sword in the desert?
To add to Jeremy's reply, manual verification is offered in #verify for those who did not want to spend 1 credit for instant access to the community. 🙂
Technically DALL·E is to be used for G-rated, nonviolent content, and swords are weapons. The contextual filter does block the word "sword" in most contexts I believe
That would make sense - I figured as things like Disneys Sword and the stone and Aladdin were G-rated and feature swords, it would just be the violence part that was against the policy. Thanks 🙂
Hey guys, do you perhaps know what Dalle's issue would be with the word 'fitting' in this context? "A high quality photo of a yellow agapornis roseicolli wearing fitting sunglasses”
agapornis may be the word being flagged.
i got access to dall-e 2 a few hours ago so this question may seem stupid but is the 50 generation limit each day or is it a different time period
The first sentence I tried was the same, but without the word fitting. That worked fine. It only gave me an error when I added the word fitting
50 free the first month, 15 per month afterwards
Then in that case, previously the word fitting was abused to circumvent the Content Policy. OpenAI might have blocked it now.
Would there be a list of words we can't use. It seems like such a normal word to use, but it flagged me as using the tool inappropriate.
Unfortunately there is no official list of banned words.
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.
Ok, thanks for the help!
ooft. brutal.
Hey everyone! I've been using DALLE for a bit now and I was wondering if there is a way to check all of your prompt history? I didn't realize it was keeping it from session to session and now I lost some of the cool images I generated 😕
For now, you can access your previous generations via your browser history. (Hope you didn't delete them!)
That worked - you're a lifesaver, thanks a bunch!
You're welcome! If you're not aware, you can save it to your collections! Up to 10,000 images.
Are there plans for increasing the resolution or adding the ability to buy a higher resolution version? i would love to have a 4096x4096 version as a 3d artist to use in my art
Not in the short term afaik - see my latest comment in #ai-discussions
ah fair enough. hmm would almost be interesting to somehow generate clusters where it generates 4 corners of the same image to combine it into 4k
You can absolutely create 4k images through advanced compositing techniques - hard work to make it cohesive, but it can be done!
I will give it a try. I am more after generating virtual textures or even skyboxes. but i am already loving it to get inspiration for concept art rather then needing to pay 1k for a single concept art image. it is not perfect yet but good enough to get a sense of what i want to re-create in 3d
Hello, I just realized I verified the wrong Discord account and was wondering if there was a way to reverify on another one.
Try to verify your correct account with the bot.
Will do 👌 , I just didn't want to accidentally trip something
You can't use prompt with variants, right?
Like I upload a picture and I want it transformed in a certain style.
Yep, no prompts when generating variations
Anyone knows whether this is planned in the future?
If you don't use the free 50 credits for the first month, in the next month, will you have 65 credits or is 50 the limit?
No heads-up from OpenAI 😦
I remember watching the Dall E 2 video from OpenAI, where the girl uploaded a picture of Mona Lisa and typed 'Give her a Mohawk'
Your wallet's the limit 😬
I'm not sure if the 50 credits expire (you can check your balance here https://labs.openai.com/account)
But when you've finished the 50, the maximum you'll get per month is 15 credits that expires and automatically replenishes after 1 month. It does not stack.
If you bought credits, will you still get the 15 free credits on top of your paid ones?
Every month, yes.
Where do I find all my old generations? The tab on the right does not scroll back that far.
Browser history
1 credit = 1 request
Ty
Where is that? There is no place that says ‘browser history’ that I can see, I only see the last 50 generations or so on the right side of the dalle website interface
It's not in the DALL∙E website 🙂 It's on your browser. What browser are you using?
Oh I see, firefox
If you're on a Mac, access the Browser History by pressing CMD + H, if on Windows, Ctrl + H.
Ok thanks for the help
Hi! Do the first 50 credits ever expire?
Yes, they expire within 31 days upon receiving
But believe me, that's plenty of time to use them 😉
Is it true we get 15 credits free per month?
Your credits will replenish to 15 each month indeed. They do no stack
So if you end up with 4 credits at the end of your period, you will receive only 11 credits
4 + 11 = 15
Oh okay. Thank you so much 🫶
Invite teammates to join
Hey can the bot in #dall-e-bot remove backgrounds? Also, where do I find a list of all the commands?
the commands are automatically reposted in that channel on a schedule - may just need to scroll up a bit. No, there's no function to erase backgrounds
That would make a great pinned message, <@&998386466063396966>
Pinned! 🙂
How do I get to the DALLE2 web app again? Cant find it on the site
Thanks
Why tho?
Not quite there yet but is there some division I can speak to about commercial use and terms, etc?
@rugged wing terms are here: https://labs.openai.com/policies/terms
I was wondering the same, but it looks like you can use them fully, commercially, but you need to state in the relevant places that it was generated with OpenAI
OpenAI own the generation but allow you full exclusive rights to sell it and use it etc
yea, looks pretty good for a beta/apha rollout t&c page, just ya know for business you need uptime agreements and bleh bleh
oh so you don't mean in relation to the actual images, you mean in relation to their service?
yea, I assume the images are yours as you paid for the credits
i mean the nitty gfritty stuff my legal dept always wants me to have signed
yeah there still does seem to be caveats there even
yea, there will be a whole mess of issues for a few years while the legal system gets it's head around deep fakes and related face usage issues
logo's and all that
Will my initial 50 credit prompts combine with the subsequent credits gained each month?
Or do they reset to 15 credits each month and the previous in the account get deleted?
no, free credits expire after a month
so any old ones would be replaced with the new ones yea
I see
i bit the bullet and just bought some because i guess have to learn how to make the right type of prompts for it too to get the most out of it!
take a look at some example images you like and look at the prompts that generated them, I seem to like the images that specify the camera lens type and details about lighting etc
that's a great idea, thanks :)
I'm sure that's more of an offshoot due to pro photographers having that sort of metadata in the high quality images
"a riot in the city in the style of the Bayeux Tapestry" why is this a violation? riot?
Hey I’ve got a rectangular image that I wanted to use #dall-e-bot to pad out to a square. This doesn’t currently seem to be possible though
Post your image, I’ll square it for you
Yeah. Rioting. Some locations (ie china, Pakistan—anywhere there’s unrest or violent history) can also trigger it
So are you just wanting more blank space below these characters to see if it will generate more?
Yeah, basically some transparent space above and below
It’s partly going to be an uncrop, and then I’m also going to generate variations. It’s an experiment on how it interprets some characters that are there but might not be recognizable as text. And hey, if it fails and just makes a funky image, that’s a plus too cause it’s a free cool looking image
Yea I think the detail for that one will be a bit too fine grained for it to generate anything new or recognizable unfortunately
Something like this may give more legible results? Not entirely sure
It looks like the original symbols were too small to register as text, possibly because of the circumflexes, so I’ll definitely try that one next
You can definitely see where the original ends and the inpainting begins, but I believe that’s because shader normal maps were mostly culled from the training data. But I could fix it up pretty easily to be coherently readable, at least by my graphics program. Overall a decent alien language
I do kind of like these crazier ones too though as sort of psychedelic abstract art
Hi guys ! Total noob here but having a blast. I thought my "recent" list would show all of my prompt history..
But I lost some of the pictures I created at the beginning because I didnt save them and its no longer in the list ! Any way to retrieve them ? ❤️
Unfortunately, no. Luckily they bumped the max collection from a thousand (which you used to be able to fill in a week) to 10k, and reduced the amount generated from 6 to 4, so it’s a lot less likely to get overfilled quickly. In the future save everything, even if it’s not what you were hoping for. I’ve lost a few images from my first week that way and it sucked
Sorry, didn’t mean to hijack the thread there. I probably should have put this in #suggestions
Have you reached out to the team at support@openai.com? That would be the best place to get the answers you seek!
When buying new credits it says + tax. How much is the tax?
Is it different depending on what country you are in?
@mossy drum for a 15$ USD purchase it had $2.25 tax. And I live in Canada. hope that helps
AWww, sad news, I was proud of those prompts. Thanks for replying.
Thanks. I live in Australia, so I wanted to be sure before purchasing it
it will be something similar for sure
yes, go into your search history they will be there
out of an abundance of caution does anyone know if long-dead historical figures would fall under the rule against creating images of celebrities?
I've seen people doing historical figures like george washington, da vinci, cleopatra, etc. I think the rule is more to prevent deep fakes / defamation, so I assumed that they are okay.
How does one change a password
Any plans to make OpenAIs dalle-2 work with an API?
Would make for a great logo generator, using GPT to check if a sentence is going to low or high quality results
Any @openai guys here for some questions about productizing Dalle-2? Thanks
how can I use the DALL-E 1 from github from the official website?
