#community-help
1 messages · Page 11 of 1
And also some things that I make on Dall E are just variations on already done images
I don't think it's going to work but I'll try it.
Actually I changed my mind
I'll do another one
Truly amazing...
Yeah how the outpaint seems to work is you give it a prompt that would make the image from scratch, but then give it part of the image and it creates the rest to match. I've seen examples of people doing that with just a sliver of an image to give the colors pattern and it drew all of the rest of it matching that sliver.
Edit I think is a different thing. But I'm just starting too and saying what I see so far. 🙂
Where's the upload button for image file variations
If to share/publish, click on the image to expand it then it should be above it. Or to download, in the upper right of the image (either a down pointing arrow or three dot menu). If to upload your own image to work on, the upload button is on the same screen as where you enter a prompt/desc.
Hi! How do I use DALL-E to color in a drawing, colored pencil style?
I already have a black and white drawing and I'm trying to get dall-e to "color it in" with a specific theme
Is that possible?
is it legal to remove the watermark on the pictures? if it is how would i do that
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
It sounds like you may be asking if DE2 can do "style transfer" to apply a different art style to an existing image. It is not engineered to do that.
Good evening dear Openai team,
I recently got my permission for Dalle2 and already tried it out and got to know it…
My Question: My friends and have a discord, public, with the intention to inform about an NFT project we currently are creating. [Don’t worry, we are not planning to sell Dalle’s images 😉 ] I had the idea to create for entertainment purpose only some sort of comic using dalle2 and the created image. If we inform about dalle2 at the same time and don’t use it for marketing and advertising, can we publish these comic pages?
I hope you understood what I meant!
BW
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Yes.
is there some rules I'm not following? I tried to make something related to Queen Elizabeth, Joe Biden, Conan O'Brien and keep getting the "It looks like this request may not follow our content policy." can someone explain, what I'm doing wrong? 😦 just last week, making a "Jack Nicholson" looking man went through with no problems.
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
Public figures are not allowed.
Rest in peace Her Majesty Queen Elizabeth II
- Please be reminded of our Rule No. 6, no discussion of politics.
- At this moment, we will only allow comments regarding the Queen's passing in the #off-topic channel, anything more will be considered political discussion.
- Generations regarding past and current monarchs or any public figures are not allowed as per the Content Policy.
Thank you for your cooperation and understanding.
ok, thanks. i was just confused since it let me do a Jack Nicholson without a problem and I've seen some awesome Conan O'Brien ones. 👍🏽
was there a change to the rules or did those just slip through?
would have loved me some Queen wearing a Biden mask, to see how DALLE managed to crunch what up but, oh well.
Generating public figures is against policy, and repeated blocks by the content filter can lead to account suspension. Best to err on the side of caution.
now that I know what the issue was for not succeeding, will definitely try to not piss off the filter. thanks for the info. no more public figures for me.
why is dalle2 doing the discord verification so weird? no oauth?! I had to wait a month to get verified because I tried to join after running out
does anyone else get a lot of excess green in their generations?
The current Discord verification procedure is the only one we have available from the OAI interface currently. We are volunteers here on the Discord and do not have control of the verification policy. If you have a suggestion for changes to verification please mention them in #suggestions
Color effects can be side-effects of other terms in your prompts. Check your prompts for words that may have implied color associations that may be spilling over.
gotcha. I just realized it's probably pulling the green from another part of the image, it's using it like it's part of the color palette I'm asking for. Interesting
DE2 is not good at precise color allocation, particularly when there are multiple objects being rendered.
Hello, I was wondering if there was any open dataset for different images generated by DALLE?
Ah, gotcha. What's the difference between "extending an image" and filling in an image? Unclear what's possible / what isn't given an existing image
e.g. I've heard of dall-e removing objects within images, or rendering paintings as photographs and vice versa --- what makes coloring in a drawing different?
Here is one of our recommended videos on edits. Other good external training resources are listed in #📚┃resources https://www.youtube.com/watch?v=juZqvZGKVi0
• Website: https://labs.openai.com/editor
• Background-Music: https://www.youtube.com/watch?v=SLK3SCwwax0
• I hope this helps some of you out :)
Coloring in a drawing is an example of what is called style transfer, which DE2 is not engineered to do.
We currently don't support transforming the style of an image into another style. However, you can edit parts of a generated image and recreate them in a style you define in the prompt.
Discussions of other AIs are only permitted in the #ai-discussions channel.
As a community, we value dialog about the future of AI, including alternative technologies and their associated ethical issues.
ok I understand paying $15
but tax?
And you get no refills until october?
WTF
That's the worst system ever
might want to @ your government for that lol
yeah, it's your government's fault
and apparently you don't get free credits until October
beggars can't choose
lol send that over to #suggestions
No wonder people are using dalle clones
because no one wants to tackle this system of waiting weeks for a tiny amount of credits
and paying tax on top of a price
again, beggars cant choose. and tax is determined by your government.
either way
You may want to continue the discussion in #openai-chatter -- We try to focus on #community-help here. Suggestions can be made in #suggestions.
Hello there! New to the community!
I'm a professional artist for an educational company. I was wondering if it is possible to update a set of my personal creations and ask the AI to follow that style to create new images
yes
Sort of, you can extend that image using outpainting, but you can’t, say, upload an image and ask for DALLE to create a brand new image based on it
How to clear image gen history it's too big
Is there an ETA on other Payment methods like Paypal / Maestro bank cards, .... etc ?
There's no PayPal??
no
Can u clear gen history?
If you click "new" in the sidebar it'll clear your history and begin a new session, but if you want to keep access to your previous generations you should bookmark each prompt's web address
Thank you! I hope they add this feature in the future! Wondering how big should be the data set to produce good results
is there a way restore old session history?
If you started a new session, no. But everything should still be available through your browser history.
okay, tysm
i asked cuz once, my browser bugged, and i was able to see the whole history
btw, can you get credits back if i got things that doesnt match the description?
or not?
im just curious
is there any way i can recover an unsaved outpainting image?
Not the whole image, as is disclosed on the editor page. You can instead save the generations you made via your gen history and stitch them together manually to get the same end result
k
they dont show up in my recently genned
i just used 25-30+ of my credits creating a large image, and while i was zooming in and out, my mouse zoomed in all the way on the page instead of the editor and it wiped the image
is there any other way to get the images i generated
nevermind it just refreshed as soon as i tabbed back in and said that
my bad
i have a question, i got a warning for a prompt i used today that i didnt think would be a problem, but i guess it was interpreted in a way i didnt intend. should i worry about getting a warning?
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
You should be able to see everything you’ve created now!
Hey I've got a question, will there be more than one payment method? I don't have a credit card, only a mastercard debit card and I really want to buy more credits.
Can I now sell Dall-E 2 pictures as NFT`S, somehow I can't read that from the Terms of Usage etc. I think because my English is not that good.
ty, is this a new feature? im just curious
yes! check #announcements
yes you can
as of now, no. sorry 
so i technically kinda "predicted" it?
or it was more of a suggestion
idk
im curious
And did i have to declare the Pictures as Ai generatet or from Dall E " when i upload them as NFT´s?
Welp. Now that all generations are here to stay, is it possible that there will be a deletion function in the future? I have a big accident where it's the same generation with just slightly different woodgrain. 💀
i believe you will need to give credit to OpenAI in the description
ok thx
that doesn't sound right... check again in an hour or so and if it's still the same, email support at support@openai.com
No, it's not a glitch. It's something that's totally my fault because I was using the editor and not giving the AI enough empty space to work with so it backfired.
hope it's alright now!
Haha. I'm doomed to look it forever unless I'll be able to delete it someday. My disgrace should be wiped from DALL-E's memory. ;w;
Can I generate images higher than 1024x1024?
Preference 1920x1080 or anything else bigger than the former one
"You have 34 free credits to use until 20 September" does this mean they'll get reset the 20th?
Yes, though they'll get replaced by 15 monthly credits instead.
You can use outpainting to extend the canvas to the dimensions you'd like, but that will unfortunately cost an additional credit for each additional 1024x1024 square you use to generate it.
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Regarding the change to image-saving making it so all images created are saved (including retroactively):
Is there a way to delete images from history, in the event that some are particularly cursed, embarrassing, or similarly worthy of forgetting?
Is there any way to remove something from „All generations” list?
Any idea on how to type for the person being drawn be looking straight in the photo, and not sided/profile?
Maybe that is more a trick/english question rather than AI itself, but I hope you humans can understand me better xd
Ex: This guy is not
This woman is
That stuff looks cool as it is. What prompt are you using?
Inspired on this:
#daily-theme message
keywords: ice, god/godess, flowered background, drink coffee, and mainly "art by alphonse mucha"
Have you tried plugging in "facing forward"?
Thanks.
I think bot might understand "straight" in a sexual way, instead of the point direction
I was also looking on how to do this. If it is not part of the feature set, can I ask why it is advertised on the demo page? Was it a removed feature?
We currently don't support transforming the style of an image into another style. However, you can edit parts of a generated image and recreate them in a style you define in the prompt.
will dall-e get a remaster option ?
Hello. Is there some way of batch download of my generated images? Thanks.
Is it me or does doing a variation completely get rid of the style?
Is there any way to delete a generation in the collection? I have a lot that clog it up which are useless
Tell me about it. 💀
I have some pics from when I went on a tourist submarine, there's coral and everything in there. Can DALLE give them teh correct color??
If so, what wording shall I use?
same (kidding they're all awesome)
if you mean using descriptions to alter existing images, there isn't really much you can do through dalle
unless you want to erase aspects and add new things to them
Hey at least yours look different.
Sounds more like a job for Photoshop.
😰 there's me trying to find an easy way to do it. Thanks you both
hopefully in the future though 
Could someone from OpenAI explain why "Your card was declined"...?
I tried a Bank of China Visa Credit card and an AMEX credit. The decline_code from Stripe (Payment Gateway) is always "generic_decline". Called Bank of China, they confirmed no transactions were issued.
Having double checked card information
I'm trying to get inspiration for an alternative history where humans reached space in the late 1700s, the main character would use that realities version of the remington new model army but it keeps telling me it violates tos?
I'm not even asking for anything violent
most probably an issue with Stripe or your bank's.. rather than OpenAI's
how do i create images with a specific resolution?
It may know that "a remington" is a weapon.
Yes it is a weapon but why is that bad I'm not asking for someone being shot or anything, just want it would look like to function in space
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
I cant find anything on the dall e apps existence, and there is no evidence of how i downloaded it. Its not in my downloads or anything. Im so confused. If asked me if i wanted to get the app version amd i was like yea, but even the site itself has 0 trace of it.
Now its here, me with an app that apparently doesnt exist
Hmmm. Does it do anything? It is on iOS?
that's an android navigation bar, so I guess it was on the play store
nah I don't think it was on the play store
through the site I saw an option to install the app, and it was downloaded from there
I find a 3rd party app of some sort in the Play Store.
it basically acts as a shortcut
ohhh, okay. My bad, I just assumed that it was on the Play Store since I didn't know that you could install apps through websites
I am told you can add a shortcut on iOS. Maybe that is what it is.
oh yeah true, but I think the one they have is is the proper one
ahh okay, no prob
Yes, it functions the same as the full site
Its on ios, yea
Interesting. Thing is i cant find it again oddly, cause i want it on ky new phone but the site doesnt want me to ig
It says downloaded from the play store when i scroll down, but when i press the button that would take me to the page it says it doesnt exist
Its odd, for sure
This is how it looks, it apso has a splash screen when you open it
How do i add an image to my favorites when it's in my collections? I can't seem to find a way to do it anymore
Me either, its a mystery ig
I believe that's an Android feature, similar to iOS. Where you can add a webpage and make it into an "app".
For now, you can select an Image then press 'Edit Image'. It will then take you to the Editor, now press the arrow at the left side. Finally, the "Favorite" button will show up at the right side.
Ok awesome thanks!!!
Worked like a charm, definitely appreciate you @verbal crag have a good night!
Rats. But I see you did a great job on the character you were working on. Maybe the sword is a problem?
is it?
the sword?
so Dall E violated its own policy? 🤣
I need a landscape of this badly, I can't let it go 🥲
I am honestly not sure. I don't work with this genre myself so I don't upload this type of character. Leave it up here and lets see if others have had similar problems.
okay, thanks bruh
I feel for you on this. You did a great job with the character. We have been asked to direct questions about the edit functions to this #community-help channel, so this an appropriate place to ask. Wish I had a better answer for you.
no problem
I hope this gets resolved 🥲
I love the image you got nonetheless!
I have 2 questions about the sharing policy, hoping to maybe get some help understanding a bit more in-depth.
-
Can you share the created/generated designs on all social media platforms without copyright issues?
-
Can you sell/monetize the designs that are created/generated? Such as sell them as prints, posters, clothing designs, or even NFT’s.
ah?
how would I go about correcting the ear, nose, and mouth of this image I generated using DALLE?
preferably would like to change to something like this (except for the eyes)
I'm not authorized to give you this information, i can only tell you what i heard/read online. As far is I know you can share the pictures on social media without a problem. Selling the Pictures is not allowed as this would be a copyright infringement. But as i said, this is only what i heard/read online and not a official answer.
Can you share the created/generated designs on all social media platforms without copyright issues?
Yes.
Can you sell/monetize the designs that are created/generated? Such as sell them as prints, posters, clothing designs, or even NFT’s.
Yes. You have the rights as per the terms but OpenAI still owns the generations.
Just to be safe, kindly read all the terms below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Is copy right stuff against the T&P? Things like pokemon?
As per the Content Policy:
Do not upload images to which you do not hold appropriate usage rights.
hey guys if i produce a photo is the rights to the picture mine?
I'm trying to get an image of Nikki Minaj with a horse face or as a centaur and Dale says it's against policy. Which bit is going wrong? 😦 "Nicki Minaj as a bubblegum centaur"
Please read the Content Policy to avoid warnings.
I did
Hmmm, what does it say at the bottom?
As per the terms of use:
Use of Images. Subject to your compliance with these terms and our Content Policy, you may use Generations for any legal purpose, including for commercial use. 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.
https://labs.openai.com/policies/terms
thanks!
If in doubt, please read the entirety of the terms & policies, linked below.
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Thank you. This was the question I came for. I'm about to comb the t's, but do you know if higher res images are possible?
I decided to use a different email and account name so I made a new account, do I need to close my original account or are you allowed more then one account at a time? I’m asking as I don’t want to get ban if your only allowed one account, if so how do I close my original one ?
That makes sense
You are only allowed one account. Please contact support@openai.com
Thanks I’ll do that
Is there an official (or popular unofficial) Discord server for OpenAI’s GPT-3 users? I’ve done some searching but haven’t been able to find one
I sent an email
Hey, is it allowed to use art generated with Dall.E as cover artwork for music and be uploaded on streaming platforms like Spotify? Thanks a lot 🙂
Yes.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
how did my art get removed twice just because it had an interesting crown design of what a crown of the king of the USA would look like? I specifically mentioned that I was only talking about the crown design, there where heads visible but of a sort of plastic general face, not any political characters and a us flag was visible in the background, is that to political?? I don't get it
or is this message also gonna get removed because of general keywords idk
Users from all over the world frequent this server, and there’s no way to have a nuanced discussion of conflicting world views on Discord. Use your best judgment in determining whether something might classify as political content — yes it’s murky, but we trust that you’ll be able to!
Rule 6 was applied by a mod in the other channel as well. Move on, please.
i don't get it, can i just get an explanation?
rule 6 is pretty free for interpretation and i thought this would be totally acceptable
damm
ill move on
Has anyone used the GFPGAN face fixer tool? Jw because google throws up warnings and wanted to see what the community thought before trying it
Would have posted a link but I got rule7'd 🤷♂️
What kind of warnings?
"It may request access to your data stored with Google, or read data and credentials from other sessions"
Guessing it's a generic warning but when things start talking about reading more data than I want to give I get concerned 🤷♂️
Is suggestions for discord for for dall e devs?
Both, really 🙂
Is it possible to remove something from "all generation" list?
Hey there, is there an API to call dall-e ?
DALL·E is not yet available through an API, and users are permitted to access DALL·E exclusively via our public-facing website. We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use. We will continue to evaluate how we release this technology as we learn more during the beta.
Not at this time. You can save your generations to favourites to reduce the noise though
Is there a way to download all our images at once?
Thanks for your answer
Just use selenium 😉
Not yet. They have seen this a few times in suggestions so it's probably on their roadmap 🙂
Is there a google doc/sheets available with artist prompts that work with Dall-e? I've tried several different artists with varying results and would like to not waste creds if they aren't apart of the training data set
So am I using the wrong prompt for artist here?
based on that i would expect something pretty more animated than what I got
I think sometimes re-iterating the artist helps as well, but generally the more complex prompt you have, the less the art style will really work for a specific artist. Here are some results I got:
https://labs.openai.com/s/h1JYqI0RQYC1bPXQGsJk9bpF
https://labs.openai.com/s/H50rvdSlPYU5fx63EXzj5T6m
Feel free to stop by #❌┃prompt-help ! That's a better place for these discussions :)
Great thanks!!
How can I invite a friend to use the DALL-E 2 website?
I think the only way in is by application and then OpenAI will let them in
Thanks!
Has anyone found an explanation for the random macaron generations? I've always found them kinda funny
I've seen a lot of generations similar to the third image
I edited a variation of DALL-E's image, and it said realistic faces aren't allowed lol
Is that normal?
Yes, if there's a face in it. DALL-E is more than capable of creating images it itself considers to be too realistic
Here's my theory about it from back in the day:
#ai-discussions message
oh, funny; definitely makes sense, thanks
i gave dall-e a prompt that it does not like any sugestions to get a similar result that are still within the guidelines the prompt was a shot of some travelers walking into a desert city built under the skeletal ribcage of a long dead titan, tarps are tied between the ribs to keep out the sun, trade in the city is bustling
Maybe it’s the word dead
There are many problematic words in this prompt, but the one most likely flagging your prompt is "shot".
Though I can't guarantee that's the only one
Why does the text never come out right. I wanted a hat that said “It’s Yeezin Time” and this is what I got
@rough tundra
@verbal crag
simply put: the ai is taught to create images, not text 
But I want text in my image bro. Is there any way I can request it to be able to make text?
I mean if it can create detailed images I seriously doubt you wouldn’t be able to make it do text as well
Only ping us when there's emergencies.
Regarding your concern:
• While model updates aren't frequent, we can definitely consider adding licensable data for it in the next training round.
• Please tell us more about it through this form.
Ok sorry I won’t do that again
The current model is yet to learn to be able to generate legible texts consistently. However, it can generate commonly used words/phrases like "Happy birthday!" or even "Stop" from a stop sign.
Google's research suggests you need about 30 times more training data before an aesthetic model like this can even begin to form words that aren't related to visual iconography (e.g. "stop" signs or common logos). Even then it will only do it if explicitly instructed - it will never understand language. So e.g. even the best model couldn't represent a legible "poem being presented in a PowerPoint slide"
you cant just program it to automatically pick up when someones discription says "it says" and then put whatever words are in the quotes onto the image? that doesnt sound like it would be very hard to program. pretty much any basic computer program can easily add text to an image
i mean discord is generating text rn as i type lol
AI is not a "program." machine learning develops its own set of cryptic rules that cannot be altered without complete retraining
Might want to read how the AI works
Not your typical "program"
oh okay... well if you ever figure out how to make it do that then be sure to let me know i guess. ill stay on this server.
i was unaware of this. i always thought it worked like a normal program lol. thanks for the insight. ig u learn something new every day lol
i don't work for OpenAI, but even if I did it probably wouldn't be a priority. there are other tools for adding text to images that work far more efficiently in terms of processing
fair enough. i guess ill just generate one without text and then add the text in with photoshop. thanks for the help though
I have a problem when trying to enter labs.openai.com from my cell phone, I just get a static white screen and that's it :'c
Sorry you're having trouble, could you contact support@openai.com?
They'll be able to help you with this issue
Ok thanks 🙏
What is the largest file we can download of the DALL E images? And how do you save the files the largest? I’d love to print some of these for my wall.
All images generated by DALL E 2 are 1024x1024
If you want to upscale them, there are many AI Upscalers out there.
Quick question, are dalle 2 pictures allowed to appear on blogs for a student project?
yes
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
There shouldnt be an issue, just make sure not to claim the images were created by you
Thanks. I’ve got a few I just fell in love with. Hah
Hey, guys, I'm new here. Just wanted to buy credits but JUST BECAUSE I'm from Kazakhstan it's not letting me to buy as an individual. Even if I wanted to buy from company name, there's no VAT ID selection for my country. I kinda feel discriminated.
that's most probably on their payment processor's fault, Stripe. maybe they do not yet support accepting payments from your country
Is there a way to send a generation from the collection to favorites?
Yep! Here's a temporary workaround: #community-help message
thanks. I think. Seems round-about for now. but it'll do!
i actually joined only to ask the same question ]
can you recommend one ?
I’ve heard lots of good things about Gigapixel AI, but I don’t have any hands on experience with it.
how do u get other aspect ratio?
Not available at this moment, but you can use outpainting. See #📚┃resources for a video on that
In its current version, DALL·E can only produce images in a 1024x1024 square.
What are the latest key-worlds to use in prompts to generate realistic images or photography?
How can I learn DALL-E something?
Have you looked at the photography guide?
https://destiny-sherbet-aa9.notion.site/Photography-and-photorealism-guide-045c50ded4c54fdfb52f55acaf613f1c
we have a ton of resources in #📚┃resources for users to check out :)
i suppose this answers your question as well @sullen swan
Ok. I’ll try to found an answer here
if you need help let me know!
Ok
Hi everyone... just wanted to know is it possible to generate DALL-E 2 art from Discord here ?
Or only from the website?
DALL E 2 is only accessible through labs.openai.com
ok thanks
guys, how come dalle is no longer putting its signature when I am outpainting via edit?
The signature will be placed once you download your generation 
What does outpainting mean exactly?
Take a look into the #📚┃resources channel for a full video tutorial :)
Essentially, outpainting allows you to stitch together many generations
to produce a larger image than 1024x1024px.
will do, thank you!
I downloaded it and its not there
Hi I'm really confused by something and the FAQ wasn't clear... when I was going to the page to purchase more credits, I saw a message that said credits are shared within the university where I work. Does that mean if I purchase credits any student/faculty/staff member can use them?
How do I use credits
Does outpainting cost credits? If so how many?
1 credit = 1 prompt generated (4 images) (by my question I’m not sure about outpainting)
Thank
hey so I like how the history of generations isnt closed off anymore but is there a way to delete generations from my collection? Mostly just to clean out the clutter. I know there's favorites, but there's some generations I just dont want to save even in the collection.
i believe there is no way to do this
Can I make a request the devs add it?
sure! every mod/dev lurks in #suggestions once in a while 
So, I got an invitation to DALLE2 about a month ago and used it without issues for a couple of weeks, then took a break for a bit and was going to log back into my account just now, but I'm now getting "user not allowed"? Was I somehow banned? I didn't get any notification about issues with my account and definitely didn't make prompts with improper content. I remember my last couple of prompts taking a while to process (there might have been server issues that day), but other than that I wasn't notified of any problems with the prompts themselves. Any help with this would be appreciated, thanks.
DALL E 2 appears to be down at the moment. Hold tight!
Is that why I can't log into my account?
Oh, that might be the reason then. I wasn't aware. Please disregard my post lol! 😓
Most likely, yes. The team is on it!
No need to panic! There appears to be an outage right now
Glad it's just a server error. I too thought I was somehow banned after seeing the "user not allowed" message 🙏
Phew! I'm glad too I almost had a panic attack as well haha
Came here thinking I was banned for using the word tentacle on a generation two days ago lmao
Gotta be careful when generating sea monsters
oh goodness, I'm glad I checked here. I haven't been to my account in a while and now it says I'm not allowed. but looks like they are working on it.
Same!
I wanted to make a rooster in a military uniform and now I have to wait T_T
HEY EVERYONE! No need to panic! There appears to be an outage right now, the team is working hard to fix it
Tysm! You guys are great
oh,i was about to say something about this "user not allowed" thing, i guess i'm not the only one
They could have set up a better outage screen 
Thanks I thought I was banned with no warning 😅
i got so scared i thought i was banned bc i got a strike last night
jesus christ i just had a small heart attack and thought i was banned. minutes after buying nrw credits too 😄
good to know its just an error
So as I read the User not allowed screen ia related to some sort of error?
Hold tight everyone!
There's currently an outage that appears to affecting everyone! Rest assured the OpenAI team has been made aware, and are working towards a fix
oh shoot I just sent an email to openAI support, now only see the chat here, same thought I was banned lol
Same lmao
Umm, have a couple questions, I've just joined the server ^^
what are they?
- I've seen some discord servers about other IAs (i.e. Midjourney, etc) where there was a channel for users to create images by sending prompts in a message. Is there something like that here?
you have to visit openai . com to make images
So that means that it's mandatory to spend credits in order to make images?
indeed
thanks
- I'm trying to get a specific outcome in Dall-E, but I don't seem to be using the right prompts. Where should I ask for help?
thanks!
🙂
Is planned some kind of image downloader? I want to archive all my generated images and it is impossible to do it "by hand" only. Or is there some other way how download all of generated images at once? Thanks .
No announcement as of now, stay tuned!
is there a more appropriate place for me to ask this?
Yes, please contact support@openai.com for this, they will be able to help you regarding your concern.
someone help
how tf do I do outpainting
I was hoping the AI would do as it sees fit
but I have to actually write something
like extend the image as it would imagine it
based off the images contents
Please mind your language.
Here is a guide: https://help.openai.com/en/articles/6516417-dall-e-editor-guide
it's important that you have overlap between your new window and the image you're generating from
The outpainting process effectively uses the same generation process as normal image generation--i.e. it has no idea anything outside of the 1024x1024 box you give it exists--so no, it's not really possible to tell it to "extend the canvas" the way the bot is currently set up.
how would I word it tho
I wanna extend this image
I cannot think of a damn word to write down
like whattt
greasy hair
Please read the content policy to avoid further unnecessary warnings. Linked below:
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Hello guys !
How do you call this style or who is a well know illustrator of this style ?
Gotta say, would be really cool if I could resize images to be non-square in outpainting. Such as 1920x1080
If you outpaint with a specific amount of overlap, you could achieve specific aspect ratios in theory
You can also outpaint to a larger dimension than what you need and then crop to the size you need in a general editing tool.
In theory, in practice the moving of the generation frame is more than 1 pixel for some reason and so it creates a new 'side' to the image, making it no longer a rectangle.
This works, but in some cases I can see myself not wanting to lose any of the image.
the artist jamie mckelvie jumps to mind
Is there a list of animated characters that are recognized by Dall•e? (I.e Winnie the Pooh, rich Sanchez) or tv shows recognized?
(Im making a bizarre series of cartoon characters set in the handmaids tale and I’m running out of ideas that work)
At the moment there is no way to know for sure what is in the dataset
I have some google sheets from this discord of lists of verified keywords for styles and themes… so wanted to see if someone has been compiling character lists similarly
People might have, however I have not seen it yet. Maybe search around social medias to see characters people have used in their prompts
I think your best chance might either be to look into some of the guides in #📚┃resources, they might mention characters, or to search for specific characters via the discord search function.
It is indeed the case that there are specific characters that are recognized and replicated especially reliable, but I don't know of any such list.
I've seen people make like the infinite picture in dall e, how do I do it?
There are several ways to achieve that effect. The one I personally use can be found on YouTube by searching “Infinite Zoom DALL E 2”. There’s also a few tools out there that will do it for you after you generate all the images.
Oh alright, thanks!
So, is it only forbidden to use faces in uploads, or just people in general?
Is it okay if I just erase the face before generating the image?
Hey, new user here with a question regarding a sensor warning I got which I cannot understand why I recieved. I wrote "Jerry Seinfeld worried about big sausage catastrophe", and got a warning that I was requesting inappropriate content. I then tried "Jerry Seinfeld sitting in couch with Kramer coming in through the door", and got the same message. Is this due to me violating some community rule that I have missed, or is it a bug in the system?
That would be against the terms of service. No celebrities or real people
Content Policy: https://labs.openai.com/policies/content-policy
Terms of Service: https://labs.openai.com/policies/terms
API Terms & Policies: https://openai.com/api/policies/terms/
Thank you for the clarification @vocal forum !
What is the NEW amt of images that dall•e saves (keeps) now total number?
According to OpenAI it seems there is no limit.
I read an article about somebody doing that to design new clothes. Though just because somebody else did it I guess doesn't mean that's allowed
If depends. As long as you erase the face it’s fine, unless you are making an image with bad intent to deceive.
Do I have a limited amount of images i can create, because it said that I only have 15 left. I thought you could do as many as you want
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Unfortunately
I'm just starting out using the Dall-e program, and using an android mobile device/duck duck go- when I click the Download button, it does it's little thing but then I have no idea where the file went and no confirmation of downloads. It won't let me click it again. Any ideas? Perhaps I need to revisit previous generations on a different device or browser?
I would recommend downloading to a computer, so you can find the files in your file explorer or Apple equivalent of one
I have the same problem. I use Brave and its the same problem. I can only dowbload one image in one log in. 😦
would it be legal to sell dall-e images that i generate? like putting them on redbubble or making prints? i saw something about the copyrights in the terms of use, but i wanted to be sure and check here
Yes.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Yes 🙂
woah, awesome! thank you :)
thats the guy.... from.... a twix gin
?
aphen
aphex?
twins
fr
when outpainting, does dalle2 look at the style that the image currently has and apply that to new outpainting? or does it entirely rely on whatever prompt u put in?
to my knowledge, it completely relies on the prompt you put in
Does anyone know a website who’s TOS doesn’t ban uploading Art made with Ai as I am going to want to put any edits that I do and like enough online somewhere so I can get other peoples opinions on them too.
Artstation and Devianart allow AI art I believe. I've seen collections of others' generations on both sites. Make sure to tag your uploads as being AI generated as a courtesy on whatever site you go with
I was going to tag them as being made with DALL-E 2 and then edited by me in said other programs as well.
Is there a way to change the name displayed on my account/when I share images? I want to share things I make but don't want my real name attached.
I believe that's something you'll have to contact support@openai.com about
This is also affected by whether you opened the account via your linked Google account.
I have an image of a face i have generated, and it's very close to what i want, however i want to generate a teardrop in the persons face/cheek without having to redo the whole thing. I've tried deleting chunks of the face and rewording to specify tears/crying but not much has changed. any advice?
This can be tricky. But, if you haven't already, remove all language from the prompt that isn't "a tear drop" when inpainting on the cheek.
that helped, but still not perfect. added a tear drop emoji over a photorealistic image lol
i'm using up all my credits on this image
it's been a while since i used this so i'm out of practice with language specifics
getting a little detail like a tear drop can take a few generations, certainly. And then you have to ask yourself whether it is the right tear drop... Haha... The other little advice is I would say include the style, a painting of a teardrop is going to be easier than a photograph of one, since a teardrop in a photograph is basically invisible...
yeah i know, but the image i want to add it to is a photo. i Know it's gonna be difficult, i was just wondering if anyone had editing tips, or experience adding such small details
a teardrop on a bespoke cheek is going to be challenging... you might need to give it context of the eye and the "tear stream" down the cheek. Also, for future reference, this question is probably better suited to #❌┃prompt-help 🙂 anyhow, good luck!
well thanks for the advice anyway. i tried some of those things but none of them were quite what i was looking for. I gave up on such a small detail, and in case you were wondering, here's the image:
No worries. It can be tricky! It's a great image on its own, anyway.
Is it possible to have only part of an image being manipulated by AI? For example if I want to upload one of my fashion photographs, can I leave the model in the photo unaltered and use dall-e to edit just the background?
I believe dalle currently doesn't edit images, but rather creates to fill the space you have removed, so my best suggestion is just to manually erase the background, or at least the parts you want changing. That being said, there are far better people to ask. Just thought I'd give my thoughts while I'm here
I'm sure it's mentioned somewhere about how exactly to do it, but I believe it may be possible to upload a PNG with no background, so you could for example remove the background in photoshop before having dalle fill the space
You will have problems uploading photos of real people. Not allowed by yhe TOS.
Hi all -
What is against the TOS with this prompt? a very wide shot of a man playing a keyboard on the floor of a large, empty warehouse, dimly lit
I'm going to hedge a guess that it was the word "shot" getting flagged.
I was also once flagged for mis-typing "warhouse" instead of "warehouse" which is what came to mind first, haha, so things like that can cause a warning, too.
How can I edit my collection , as in remove images
You can’t remove images, but you can go through and favourite the ones you really like. Keeps it easier to find them 🙂
Ok it definitely changed but thanks
Are there any word on when eventually the current restrictions on motif will be lifted (no recognisable faces, etc). I am an artist and would love to experiment with creating cover art, so the current restrictions on uploaded pictures represent quite the obstacle/limitation.
No word yet. As an artist you could always combine faces locally
Yeah absolutely. It would just be amazing to see how the AI would interpret/alter/blend the chararcters in the picture.
I’m sorry if you answered this already. Is there going to be a bot for dalle, also, is there going to be a dolle 4
No plans that I'm aware of for a Discord bot that generates art for DALL E 2.
Is there a way to remove all the favorites and choose new ones since they were to save vs. the recode has made it very overwhelming?
I put the very same request in #suggestions - would be very helpful
I know right.
Fingers crossed.
When is there going to be another artists subsidy application program?
Why is there no expression date written to know when the credits were purchased etc., so they don't go to waste.
Would post in #suggestions yet am unable to due to timer.
I’m sorry you are frustrated with the timer in suggestions. We are doing our best to keep it nice and clean so openai can easily find the actual suggestions instead of conversations.
It does send an email for purchases, so that can be referenced until something like that was implemented
Is my gender identity still censored and if so why?
• Our filter aims to detect generated text that could be sensitive or unsafe. The filter will make mistakes and we have currently built it to err on the side of caution, thus, resulting in more false positives. We're working on improving our filters, so this should become less of an issue in the future.
• While we're not sharing the exact thresholds, a few warnings here and there won't lead to an account suspension.
Will you add my request in there then please
The content filter isn’t perfect, and is being worked on.
The official channel is support@openai.com - I’m afraid we can’t do anything here on discord. OpenAI does read those two #suggestions and #bugs-and-issues channels.
Well it seems like the word femboy was intentionally blacklisted seeing as transgender is perfectly fine and it's in the same realm
I'm gonna be honest it doesn't feel great to be excluded like this
Intentionally or not it's not great
Also why is nudity banned in a bot that makes art? Is there any particular reason?
I would have overwritten my last message yet it got deleted so.................
The rules are set by OpenAI. This is for questions for help. It’s been 8 days since you posted in #bugs-and-issues. You need to allow a bit more time probably. As for nudity, This product is in beta and the internet being what it is they are being cautious.
@broken wedge
Fair enough although I would be nice to make Venus di Milo but wearing a 70s funk outfit
#❌┃prompt-help can help you with that
Can we get much higher?
I have posted for you. Please do note that paid credits have an expiration of a year, so this won't be an issue for many months. I would recommend you hold onto any emails for the dates for now. Thanks
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
🙏 I don't have the emails TBH
Rights question: If I generate an image with DALL-E, such as a company logo, am I legally able to call it my own and use it freely?
Like, they own the generations, but allow free use and reproduction including commercial.
Electrical person just now dabbling in art, unfamiliar and wary of complicated licensing questions, sorry for my ignorance, trying to learn. 💜
The terms of service are the best and definitive way to answer this. As far as I know, you have full commercial rights to your images. The only thing I’d caution is the way you said “call it my own”. You cannot mislead anyone into believing you created the artwork yourself.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
That's OpenAI's latest official statement to us.
I just verified and joined the discord server, but it still took 1 credit away from my account 
I thought it was free to verified
Is it safe to give prompts of couples hugging?
yes, I believe so
You can use any prompt to verify, as the instructions indicate. It just needs to have the code at the end.
Hey, how long does it usually take to recover password?
Hi, I'm a D&D player and I listed to use dall-e as a resource for fantasy maps and npc illustrations... used all my 50 free ones on my first day for 0 useable images 🙄 I wonder what it is I'm doing wrong?
I failed to make:
- a religious man holding an elevator model
- a cave map (it did give me a map with the word cave on it though)
- a blacksmith's workshop map
- a catgirl with no human ears and with 3rd eyelid type eyes (also tried to add whiskers later)
It seems dall-e refuse to place an elevator of any kind on a person's hand, do not understand the difference between a blacksmith's workshop and just a general crafter's workshop, don't know what a furnace is, unable to add schematics on the wall, refuse to omit human (or elf) ears on any type on a human-like head, refuse to give a person cat-like eyes, and have no idea what whiskers are.
Also, idk what went wrong with the cave, I only wasted one try on that one.
Edit: I tried to upload my sims4 avatar who is heavily moded, just to let it see how a head looks like with no ears, cat-like eyes, and with whiskers, but I don't think it helped.
#❌┃prompt-help might be of more use to your situation
oo ok, tnx
Is there way to prevent results including text? I tried to outpaint, with erased area, but it twice generated text only. And I don't want to use more credentials to battle with this 😅
Here's area which I'm trying to fill
and i'm getting this kind of results
why cant i use the word blood on the imput
I think it fall under "G-rated" which is against Dall-e Content Policy
https://labs.openai.com/policies/content-policy
communicating with dall-e can get very messy. There are several guides in #📚┃resources which you might find helpful. Additionally there is a guide for dnd characters here: https://fistful-of-spell-slots.tumblr.com/
omg is that pfp a dalle image
Yes!
You can get something similar by more or less by doing something similar to the guide: "[lizard type] with a human body, wearing a spacesuit, by [some scifi artists, probably stalenhag], digital art, trending on artstation,"
Oh waw, that looks super helpful. Tnx!
Sorry I guess "with the body of a human" is better
Wait!
If you're looking for fantasy
You should try just following the fantasy guide linked above
continuing (with two images) why it all the time trying to add text 
I'm losing my last credits because these text appears
and again, big empty area (bigger change to get text if area is small)
For humanoid animals there's a couple tricks you may need.
trick 1. Generally the more human things you give them to do, the more human they become. e.g. being a knight, wearing armor, holding a sword.
ty for the help
trick 2. "humanoid [animal]"
trick 3. "[animal] with the body of a human"
All will have varying degrees of success depending on the subject. Dall-e is SUPER good an animals. It has a hard time not making an animal an animal.
also, unrelated question
when using the edit feature for images already made with dalle, can i change the art style?
because im currently making some stuff for a jojo based rpg and i made something perfectly related to a specific stand, but its not in anime style
this guy here
The current beta of dall-e does not allow for seed images. So there's not a good way to change art styles right now.
Is there anyways of getting free credits? Etc watching a 30 second ad or something of the sort?
Waiting until a month after you first signed up, but otherwise, no.
Does verifying my account to join this server use a credit?
Yes, but as the instructions said, you can use any prompt you want to verify
Oh, alright. I did not realize that and just typed in the number
#daily-theme still exists for me...
...you mean red?
Have you considered crimson?
oh i forgor it existed
Yes, not a specific aspect ration though per se. Each generation frame is 1 mp, so you can do the math to figure out how many frames you need for a wallpaper
oh and then like stitch them together or smt?
If you outpaint, it will stitch together as you add
As long as the frame partially overlaps the old frame because outpainting only reads what's currently in the new generation frame
Hi!
Is it possible to restore video?
I know faces are against rules
But what if the image holder can show usage rights?
Even frame by frame I’d be curious.
Thank you!
Considering that video can be 24 frames a second or more, I can imagine using dalle2 for that would get very expensive very fast
That’s fair.
And in some use cases fine
Question is not related To personal artwork.
And if your goal is to restore and not to change the content of a frame, you'd probably be better off with AI that deal with quality enhancement/restoration
And arguably affordable
Are there such things?
Apologies, thank you for your reply
Just talking uprez and detail painting
Labor intense work in some spaces
Gigapixel/DeNoise/Sharpen AI can be bought and run independently on your system to restore quality to an image
Thanks! Will look into
Wouldn't be able to draw in new things like dalle2 though
Yeah that thing
Yea, I bet you all would do better
Dall-e please ‘recreate this image sequence in hyper real 4k modern photo quality’
🤷🏼♀️
Hilarious thing about it is it will always return a 1024x1024 image
what do you guys use as a substitute when out of dalle credits
And I think it would create a unique creative look as well
Mid journey
ty
Even then, dalle2 isn't the type of AI that would upscale the 240p image. It would just generate stuff around the 240p frame that it thinks is context appropriate even though none of that stuff appears in the video.
Gigapixel would be your best bet for turning a 240p frame to 4k, but I still question the efficiency of using it for an entire video
Even if you throw money at the problem and do something like 10 computers running upscaling AI, with each computer being given a 1 minute segment of a 10 minute video, you'd still have to run them all like 1500 times
But I don't work in VFX/video editing so I really can't say if that isn't actually a bargain deal lol
Yea, few seconds per clip in the end
And even at a lower frame rate, could relay information better , more clearly
Vs, painting the frames in some cases
And the AI would be a part of an efficiency solution
Not a whole one
But I hear you. And appreciate your thoughts greatly
How’d you guys find out about Dalle AI? What do you all think about the outpaint feature?
From a friend. Outpaint feature is pretty neat.
Interesting.
Yeah. The outpaint feature is really fun. I’ve just made my first one.
It's really just a slightly improved version of patch-e, but I do appreciate it being in one website now
Yeah.
Outpainting is ⭐magic⭐
Indeed. It’s really cool.
Hey everyone, this a channel for questions, not chatting. Please move these conversations to #openai-chatter, thank you!
I didn’t say it was a perfect solution 🤷🏼♀️
anyways, do you think dalle 2 will have unlimited use when openAI releases the "next version" if there will ever be one
wow
Discussions of other AIs are only permitted in the #ai-discussions channel.
As a community, we value dialog about the future of AI, including alternative technologies and their associated ethical issues.
In order to keep the conversations on our server focused and productive, we ask that all users to please stay on topic. When posting in a channel, be mindful of what other users in that channel might find helpful or interesting.
sory
Is it possible to upload a photo (for example of a person) and ask them to redesign it in a style of your choice?
Did they remove the 50+ point start? My points went from about 50+ to 15. I have only generated 16 images :/
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
when using the canvas and generating an extension it asks me which one to accept , cant i accept all 4 ? 😅
It will save all 4, regardless, in your collection, but it will only allow you to continue with one for the extension workflow.
if it includes his/her face, then no
i got 50 credits when I joined and used about 5 and then the credits disappeared?
Did they really?
Did your free credits reset to 15 one month after being granted your initial 50? Or did your credits become 0?
They disappear at the end of the month
What are the commands you can use in #dall-e-bot?
Free credits are reset to 15 on a monthly basis. You are granted 50 free credits on initial sign up, which will last until the first reset. Paid credits expire after 12 months and are not reset on a monthly basis.
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Why the expiration?
at least they could let us know the credits disappear, I was saving them and trying to check around what can I do instead of just doing random stuff 😦
Hi, do we know if Dall-E will ever get more function like the "text diffs" process demonstrated by Ramesh, img2img, image weight, ratio, steps sliders etc...?
is it ever planned to make an ability to earn or buy smaller amounts of credits?
what are the rules with the Dall E watermark?
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Is it possible to buy credits with paypal if you don't own a credit card?
no
you can get a pre-paid card though. from recharge.com that should work
a person functions with no plastic cards attached to their bank account. like, how do you buy anyhting
Easy; so you can buy more credits.
Question, is it possible to get 2 exactly the same images from the same text input?
Different generations ofc
debit card? cant use it online
Hi everyone, I am noticing that the Dalle 2@ editor, is not available for Mobile? only desktop? is this correct? thanks
That is correct. Editor is only on desktop
What? You're Dutch or Belgian? Stuck with Maestro
Dutch indeed
Yeah I moved here 2020
Banking system is so archiac.
Been using Bunq, had a new virtual MasterCard created in seconds linked it to my local (somehow) us account and paid from there took about two minutes.
Abn Amro. Closed my account without warning due to lack of use even though I was still paying for it. I have up entirely on the old banks at that moment
I would use Banq if I hadn't had SNS already
What's that
SNS is a bank
Open ai does have to upgrade their payment form, it's illegal in the EU right now because it doesn't show tax. But the Netherlands is unbelievably behind in finance. Like you said credit card. You can pay with a MasterCard or visa debit. Its just that the rest of the world left Maestro behind and everything that's not Maestro debit is called "credit card" it's a thing. Nightmare, until I got Bunq anyway
thanks for the reply 🙂
Do they not do similar things to Bunq then? So you can participate in the rest of the world?
well yea, you can get a credit card, but that costs even more money
I pay 9.90 I think (paid yearly last time for the steel card and not so many patients) and it gets me... 25 virtual cards (in seconds), to 9 physical ones. I forget how many sub accounts probably similar to virtual cards I can make (all with unique IBANS) which can be in certain countries local accounts.
You can change any card to point to any account at any time, have one point to two (different pins). Etc I'm not affiliated and this is essentially becoming an add but if you want to be able to interact with the outside world...
I mean, typically I work with either IDEAL or Paypal
I dont wanna pay for a credit card if I only use it once every few months
Oh no transfer trees, except one euro from transfer wise, saved me literally thousands paying off debt back in the UK
Then have a lower tier more like the 4 euros of Abn. Fewer accounts, I think one virtual card. But it's not just the ability to use normal MasterCards there's everything else. Check out their website, cause this is so off topic
And you still get a Maestro, in addition.
Yeah PayPal sucks and ideal is pretty okay. But Obv only in NL
#❌┃prompt-help should be able to help
Thank you
I seem to have been accepted into Dall-E 2, but I'm not sure. When I log in from https://openai.com/dall-e-2/, I go to https://labs.openai.com, which doesn't identify itself very clearly so far as I see. Are there distinct urls for creation pages for Dall-E and Dall-E 2?
The invite you received from OpenAI is for DALL·E 2. The first version of DALL·E, version 1, was never released to the public.
Read more here: https://openai.com/dall-e-2/
I use labs, i never knew about the other. ii never knew about. labs is where i end up anyway. beta is the text one. don't go there
yeah from the main dalle-2 page you click login and it goes to labs.openai.com/oauth/login
there's just the one account
What is your opinion on selling card of the dalle images with the rainbow dalle signature and without? I am actually proud to show and I write in my descriptions that the art is created with Dalle. But I find that the signature may distract from the image and disturb the artistic mood. In half the images I am keeping it and in others not. Which is not good to be incosistent. I would like to ask for constructive ideas and well meaning opinions on this by people who sell their art.
Personally, I remove the watermark, but always disclose the piece of work was made completely or in part with "DALL-E 2 by OpenAI"
You should feel very ok with removing the watermark, aesthetically, of course, you wouldn't want a watermark on an image, even if you are proud to have used dalle. OpenAI are not enforcing the watermark being present on images, but you should declare the image was made using dalle, which you seem to be doing, or very happy to do, so 👍
dalle specifically says it;s fine to remove
the bot chimes in with something about it every now and then i am not sure if it is triggered
Is it okay if i remove the watermark?
i gueess maybe not
Thank you that sounds very logical. I guess I will remove it with photoshop or just cut the edge 🙃. You helped decide to do it!
If you're sharing it in the discord or on reddit, keep the watermark, why not? But indeed, if you want to sell your image, make a product of it, of course, in my opinion, you would want to remove the watermark. If you're worried about cropping your image, you can always use "content aware" erase/replace tools in image editors, they usually do a good job of removing the watermark without compromising your image 🙂
Absolutely agre3. I will keep the watermarks on my instagram profile and remove it on products. Very happy about it.
If you go to your collection, click on the image you want to remove the watermark from, click the hamburger menu icon and then click "Edit image", when the edit screen is displayed you can right click on the image and you have the option "Save image as...", it saves the image without the watermark
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
It is triggered 🙂
why is the image quality gettung very poor, when editing an image, brought into Dalle? (when the prompt usually gives high res)?
It's hard to explain, but I have somewhat just gotten good at outpainting and avoiding this, which is a very real thing...
My theory is that is has to do with what context you give dalle to build from. If you just give the edge of the frame, dalle samples just those few elements and builds from those 'few' elements, often resulting in a 'lesser' image than the original image... But, if you are generous with what is inside the generation frame, more often than not, dalle will match the style and fidelity of what is in the frame as it builds outwards... Though, it's not an exact science, I can't even really explain my process, I just have some kind of instinct in which I can kind of mitigate dalles tendency to go a bit muddy sometimes... I try and avoid the fluorescent reds and purples and blues that sometimes emerge, and almost like trimming hedges, get rid of those as im building my larger image... kind of rambling a bit now, hope this is helpful somewhat.
This image for example was totally outpainted and is about 5000 x 4500 I think - while the fidelity is pretty consistent through the image, i was just very vigilant... you can see you can almost paint yourself into a low-fidelity corner if you're not conscious of what context you're giving dalle and what you're really trying to "Extend" and preserve from an image when you outpaint...
thanks for your answer. yes I do the overlapping thing with outpainting too. My question was more about one image brought into dalle, then parts erased. It's just not getting me the quality I used to get. I think this may be a helpful thought for what I'm doing. I could add somethings into my image, to give it context.
Great job keeping the quality with outpainting. It does burn more credits when you overlap, but it's necessary
Oh, I see... Well, I've found in general, you get better quality with InPainting if you give more room than you want your object to be in, if you can spare the space. Like, if I want a bowling ball, I erase an area much larger than a bowling ball. Otherwise you also just get dalle filling the erased part in and giving nothing.
The other suggestion, which I'm not sure even does anything, but it seems to, say "the quality matches the rest of the image" or "the aesthetics are maintained from the rest of the image". Also, specifying what you want. If you're adding something to a photograph, say "photography" etc
Also, thanks for the compliment! 🤠
great I'll give that a go, thanks
You're welcome, good luck!
what style is this
same w this
they're like soft 2.5D
im wondering a particular style
It looks like a blender render. There are some great things in the #📚┃resources channel. I recommend the DICTION•AI•RI by @rugged wing. It might help you find the style
ive tried reading through that
i cant find anything quite, idk
accurate?
more the second one tbh
Then #❌┃prompt-help will be the most helpful. The staff are usually lurking around, and they can help you pin point that style
thanks
Hello, I want to try "google diffusion-based super-resolution ai". Is there anywhere I can try that?
That's not affiliated with OpenAI or Dall-E 2, so not here anyways
im new, is there a way to get more credits evry month?
You get 15 free ones every month
Hi, try adding mediums like "3d casual life illustration, trending/found on dribbble"
not sure if it's what you're aiming for, but adding illustration type from icons8 library and then finishing with found on dribbble, will give you some nice examples.
thanks
unfortunately that saves it in a lower resolution than the max....
When I try to log in it says I don't have an invite
Sorry you're having trouble logging in, please contact support@openai.com for issues logging into your account
Ok
It worked, thank you.
Hey there, not sure if this was asked earlier but what happened to the Collections? It now stores ALL my search results with me saving them. 1) was storage enchanced? 2) will it overwrite the images I did save earlier?
Next to the generations tab in your collection is "Favorites". That has all the ones saved previously and where saved ones go
Thanks Ninja! So whats the difference between the collection and the resulting images in the right side margins?
If you click the right side it shows your history like it did in the past. The "All generations" just shows all of them so it's easier to scroll through
The size
The history used to stop after a few generations and literally the image size
Awesome, thanks for the reoly!
can you change the size of a generation frame
Currently, no
you have to use the editor
Are you allowed to use dall e images as cover art on spotify etc?
Yes.
what should the prompt be to expand this into the torso?
i tried neck (made the head), torso (also made another head)
Have you tried expanding it less?
Yeah, that way it'll know the top of the image is a head
Can just try "complete" or "expand"
in the discord server, is there a way i can turn an image into a discord profile picture?
thanks 😄
I have a kind of really creepy question this is more of a staff question?
ask in @agile karma
Ok thank you
Please do not encourage people to violate the terms of service.
sorry i had no idea
it was just a joke, is that even possible?
Is it prohibited to generate / share artwork that displays the use of psychedelics? (ie. a tab on a persons tongue). As well as smoking weed since it's legal in my state is that alright on the AI?
hello is the max output 1024x 1024? can we get more high quality file?
Help
With Outpainting, you can. But the generation window itself is currently locked to 1024 x 1024
I'm trying to make a funny picture of myself using outpainting to change the background to Chichen Itza
the problem might be that there is a face on the picture
but it is not in the generation frame
but in the entire image
I have the same issue. Something went wrong and i cant generate anything
goddd
try replacing your face offline by something else, upload it, do the changes, download, reinstate your beauty on the changed image
no face in the picture?
okay, i'm doing that, thanks mate
hm... @molten reef and @formal cedar could be a server issue. give it a sec and try again. its not a content warning, so i don't think its your photo...
ok
sorry, i meant to ping 366
Now it works. I was unable to generate anything.
removed the face and it's working
oh well, thats good news 👍
no biggie, was still reading along
try again with the original picture, though. It might work again in general, now, if the server was causing the problems.
alright, gotta go now
I'll try that. Thank you
Pardon me but what’s out painting
Oh my, let me tell you a tale of a most wonderful feature of dalle... Outpainting is the process of 'extending' or (painting outwards) of a dalle image using the dalle editor. There's a quick little tutorial at the bottom of #📚┃resources if you'd like more info.
Is there a reason that folks don’t post the prompts that generate certain images? Sometimes I can extract it from the file name, but far from frequently…It just seems like given the nature of DALL-E, we’d want to share prompts as widely as possible to minimize credit waste.
To encourage collaboration within the community, please be respectful to others. If you enjoy someone's work and would like to know the prompt used, consider asking politely if they're willing to share the keywords they used. Keep in mind that not everyone is comfortable sharing this information, so please respect their decision if they decline.
i personally always share prompts, and not only because i'm a volunteer! but nobody's under any obligation. theoretically, prompt crafting skills are valuable and if there's anything to the premise that AI art could be commercially viable there are some reasons users might not want to share. the only reason that it could be an issue from my perspective is if users were masking "innocent" images that were created with malicious intent by not revealing the prompt (which has happened)
I am so lost in questions about yhis technology and i have lot of doubts especially that sometimes images generated looks like stock photos, i would really appreciate a detailed explanation about how dall generate the images and want to know if this technology will ever generate an artwork that might be similar to someone else's work especially that I as a 3d artist am considering using dall as a "concept art generator" and thank you in advance. I am very excited about this "boom" in ai technology. A new history will be written.
thank u ser
Hey there, how can adjust the image weight of the uploaded image? So I want to create “variations” but want to control influence of original image. Thanks!
There’s no way to do this currently
Okay. Thanks!
new user question, can you change the aspect ratio of the initial results without having to use the expand canvas?
you can also just u now, download the4 image and make it your pfp, the bot jut makes it round nd puts the watermark around the edge
no
thanks!
read the research paper, i's linked on the main dall-e 2 page. essentially, your prompt goes to an AI like, the one that AWS uses called Amazon Lex when you speak to an alexa, to understand what the words say, then these word are fed to an ML trained to turn word into images, and those images are fed to an ML trained to make images better. and then u get ur output.
(roughly, not exactly, but close enough to hopefully get the point across, again, read the paper if u want more info)
Thanks for your reply i will surely do 👍
Hi, I had like 30 left from the first 50 credits, I opened it up today and saw it had only the 15, will I be able to use the other 30 or they are gone forever?
I was saving the first ones because I thought the 15 would be add to the ones I had, not just replace them.
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Hi! for more information you can reach out for the help here support@openai.com.
Can you do outpainting on the mobile site?
You can, but it’s really only tested and meant for desktop at the moment
Couldn't find it
It’s not fully supported last I heard. https://labs.openai.com/editor
heyy..does DALL-E produde aesthetic images which would be similar to which is done manually done in stable diffusion ?
That totally depends on your prompts, but yes, it's possible. Might want to check the guides in #📚┃resources.
Discussions of other AIs are only permitted in the #ai-discussions channel.
As a community, we value dialog about the future of AI, including alternative technologies and their associated ethical issues.
much thanks.
Has anyone worked with t-shirt designs, I would like to see the results
which words to I put to make it look clearer, I know "Digital Art" is one.
try with 8k
and photorealistic
because I use that, its all blurry in every area of the image.
then go with 'Detailed'
or you could also try '15mm wide-angle lens' or 'Award-Winning'
hm
do i do like this (New york city street view, night time with god rays, 5mm wide-angle lens, Award Winning, Detailed, Digital art)
you might get similar imgs (ps. i used STREETS OF NEW YORK)
thats def alot better then what i get
i don't have much idea about it
ok
cool
i wanted to try to get some concept ideas for my story but it just gets all blurry 😦
i highly recommend checking out #📚┃resources before using more credits
like i need a outside view of a comic store based in the 80's ish
im just waiting for my credits to go up soon
adjectives such: sharp, clear, photorealistic, real life
mediums: found on unsplash/Pinterest/Pexels/Tumblr;
Lens:
How can I get 4K resolution in all images ?
outpaint ?
For pc
Outpainting is the process of adding more content to an existing image with DALL∙E. Padded empty space is added around the image, and DALL∙E fills it in!
(There's much more you can do with outpainting, this is the most basic and commonly used. Check the #inpaint-outpaint channel for more ways to outpaint.)
Generate or select an image and use the !pad command in the #dall-e-bot channel.
Upload your image to DALL∙E and press 'Edit'. Do not forget to erase the watermark.
Generate it with a prompt that describes what you want to see in the new image as a whole.
Is there any way to tell DOLL-E that the renditions it gave were good/bad (👍 /👎) for the words that you provided? I wonder if this would help the AI to improve?
Or if there was a way to identify what parts of the picture are good/bad. Say you typed in "the cow jumped over the moon". Perhaps the cow is great, but the moon is not what you were looking for. It would be great to let the AI know what parts you did or didn't like.
Instead of us all having to learn how to communicate with the AI, perhaps the AI could better learn to communicate with us.
There’s a button to report images that don’t match at all the prompt you provided, but that’s it.
So it says you can upload faces
but I am still getting a lot of "against policy" responses
how am i able to put myself in a new outfit? the image editor is only letting me put down boxed? i thought i would be able to draw on top of what i want to see different
https://www.youtube.com/watch?v=juZqvZGKVi0 this is a quick demo. upload your face and erase around it, and enter a description of what you want
• Website: https://labs.openai.com/editor
• Background-Music: https://www.youtube.com/watch?v=SLK3SCwwax0
• I hope this helps some of you out :)
what are you trying to generate?
I'm trying to "photoshop" one face on top of a body builder.
and at what time you're getting against policy warning?
Initial image upload of body builder.
Before I could do something similar by blocking out the face
that may be a nudity issue not face issue
hmmm
possibly.
I'll test that
Did the error message change?
I swear it used to say it was graphic conent or something
this is the one
Hi ... when I click on Help under the menu, some text appears briefly then it goes blank. Chrome on Win 10.
did you try deleting cache?
With all these new features being released for dalle, why can't we get cheaper credits?
I feel that is the most requested but is being ignored
I highly doubt it's being ignored. There is more than likely someone (a team?) coming up with models and ideas.
I'll try that thanks. Seems that on Brave browser, the Help menu doesn't appear at all ...
Hello guys. can i delete generated images?
nope, stays in your history
that's your personal collection, it's not public
It seems like with some testing DALLE doesn't understand a concept of foreground and background. I have tried different iterations of the same prompt "A ballerina in the corps de ballet crying as she watches the prima dance" and it doesn't understand. Has anyone tried this type of prompt? X looking at Y or X behind Z....?
you could try inpainting
woohoo's idea is a good one to help solve for the issue. Sometimes it is helpful to understand the perspective of the training.
concepts like "x is in the background" "y is in the foreground" or even "background" and "foreground" aren't always available in every trained caption.
I find that background can work for some artistic creations.
But I agree with you on foreground. One thing you can try - incorporate terms for camera/photography/art techniques which use that type of positioning.
am I unable to inpaint anymore? just the generation Frame method?
im worried not being able to draw exactly what i need inpainted will mess it up
I'd say it depends, I've done foreground-background prompting but overall scale has to be taken into consideration, I've noticed that sometimes the "space" isn't enough for the system to fit everything in, unless its supported by the specific prompting
I've been trying to get my account deleted for well over a month now only to receive nothing from OpenAI and DALL-E. It would be great if we could have some way to close and delete our account and personal data.
Did you email the support? support@openai.com ?
Yep six times now
Make it seven times.
Hello! I just wanted to understand how to tell DALLE to write a specific WORD
at what stage do you get errors?
I get absolutely no response from support at all.
Hi everyone, i have few questions 🙂 If you can help me with some of them, i'd be glad!
- why does this discord seems way less active than midjourney? Isn't Dall-e more popular?
- what can can i find more content about the new add-on accepting faces of people (still nothing on youtube)? I would like to improve the quality of my new creations.
- Is there a website with galleries of thousands of creations? If not, that should exist!
- Is it allowed to talk about Stable-Diffusion here? I wonder about their work also.
- https://generrated.com/ is a nice gallery
Thanks. 🙂
depends the prompt, one of them could be "word "love" depicted" / I've noticed that if you add "found on behance or dribbble" in the end, the chances are high to get the word
- this discord might be less active than midjourney since D-E 2 doesn't actually use discord as the image generator. this is more of a place to share stuff and ask questions
oh really, sound curious, I'll try that out
- you need to use their discord to generate something, of course it will be more active.
- check photorealism guide in #📚┃resources
- #📚┃resources
- only in #ai-discussions
Hi! for the first questions, I don't have any specific answer. for the rest, visit #📚┃resources you will find lot of interesting tips and tricks how to improve your generations. for the last part, you can use #openai-chatter or #off-topic or #ai-discussions but I think it will be better to discuss them at the respective discords, since the users there can educate you better
#daily-theme message for example
am I unable to inpaint anymore? just the generation Frame method?
im worried not being able to draw exactly what i need inpainted will mess it up
ohh wait i understand i think
am i able to generate multiple frames at once?
I see.But still playing around to let DALLE understand to write down my nickname
It catched the words and letter changing the owrder
Thank you everyone for your answers, I am checking the different #
Hello im new, is there a place/forum to find helpful keywords?
try letters separately and maybe use outpainting if its necessary
ok
#📚┃resources , but maybe if you can tell specifically what keywords do you have in your mind? I can give you direct link?
hmm, i guess i want to make a digital art
avatar, full body, thats sort of thing
Is Dall-e better with short prompts (3 to 6 words)? I feel like I should write as less as possible to have good results but it might also be because i'm afraid to give too many information and loose my credits by wanting so many things haha.
yes, well, if you're not sure yet, visit #📚┃resources check how prompting works, what are the best orders for specific art style and then you can come up with the prompt that will work best for that specific generation
no , can't still make it write a specific word
yes, thanks, looking into it
you're not paying credits for detail prompts, you can be as detailed as you wish
I'm very curious about what it looks like when you add your face and ask to generate something around. If you want to share your creation, i'm very curious :))
What i mean is if i give too many details, i'm afraid that it will not understand what i exactly want.
you need proper order, avoid unnecessary explanations, and it will understand. for example this order, you can try generating traditional art with it
hardly matters but i'd advise you to use commas
thanks for all!
Attribution question. The ToS mentions Dalle must be stated when sharing an image. What happens if I sell artwork that uses Dalle, either in full or in part, to someone else? What happens if I license work for print or decide to use it in print within my work works How does this dalle attribution get maintained? This angle of Dalle is super off putting compared to MJ. I need to track and stipulate which parts of an image are AI? That's just mental.
Yes. Avoiding deception is our primary intention. We're not enforcing that outputs are attributed to OpenAI / DALL•E 2. We'll update the content policy to make this clearer in the coming days!
You can read more about FAQs here.
Yes.
Oh and btw, is there a function of upscaling here like on midjourney? Or should i use another tool?
you can generate variations, will give you three different variations of the same generation, you can try outplainting tool as well
Outpainting is the process of adding more content to an existing image with DALL∙E. Padded empty space is added around the image, and DALL∙E fills it in!
(There's much more you can do with outpainting, this is the most basic and commonly used. Check the #inpaint-outpaint channel for more ways to outpaint.)
Generate or select an image and use the !pad command in the #dall-e-bot channel.
Upload your image to DALL∙E and press 'Edit'. Do not forget to erase the watermark.
Generate it with a prompt that describes what you want to see in the new image as a whole.
The question isn't whether we can sell, it's about the over the top attribution requirement. Are people attributing all Dalle outputs? Unlikely.
sorry Ene but that's not my question 😄 I wonder if i can upscale like having the same photo but improved. I think?
I want the same painting with more details, not variations
variations are the similar to what you refer by upscalling.
hey guys, i am pretty new here. When editing photos, does it feel like dall e sometimes miscalculates depth of the element i am trying to add to the photo?
like for example, im trying to add penguins in the background, but dall e adds them in an oversized manner that makes them appear as if they are in the foreground
How is this feasible? How does this relate to compositions of artwork used in print or sold for print? Are there details on attribution for print or when you sell work to folks who have zero obligation to follow these attribution rules?
Hi, visit #❌┃prompt-help I think you can find some tips there
ok ill ask there, thanks
I'm not sure. Let me show you the difference to me. The 4 are variations. The last one is an upscale/improved version of another painting, with a better quality. This is MJ.
I am not very familiar with MJ upscalling, but variations here give you different variants of the same image, out of which some of them may be improved in prompt-wise, quality wise etc.
Yes, it is a little bit different 🙂
Thanks for your time.
By being a volunteer here, do you have some credits?
It's pretty expensive for a student.
you can always apply for a subsidiary program here https://docs.google.com/forms/d/e/1FAIpQLSfJzXwRXbyk3vwUtkoIVe17d4YiZ8gstikDUPCaERPqMZ8GFQ/viewform
thanks! I will definitely ask.
is there a trick to get images that are in the same style as the original when creating a generation frame. Want to create a series of art compositions that look
like they are from the same collection
hi, did you check #📚┃resources ?
i will check it
Hi Uxe, Mj's upscaling is different than D-E2's variations.
Variations will be whole new images.
Mj upscaling is a literal upscale to higher resolution and may also add additional detail depending on the mode
regular versus uplight
DE-2 does not have any option like that. You can use third party upscaling apps like gigapixel
voilà!
or if you are familiar with notebooks, ESRGAN
aha, so that's what MJ upscalling is 😅
Hi, there are a variety of ways to fulfill this. For example, if you are sharing images on social media you can leave the D-E2 "watermark" in the corner.
How does this relate to print. This seems to be the only ref to attribution in the FAQ. What if I sell artwork in composite form to a company for print in a commercial capacity? They must stipulate that it's partially Dalle? Where are clear and concise attribution rules for those selling Dalle works commercially?
You should consult with the company you are selling it to that the tool you used has these stipulations, and how they will acknowledge that.
That is not a functional nor practical way to operate.
If you have additional questions, you should contact support@openai.com
But I encourage you to also ask the company you're selling to what their position is based on those terms and conditions.
I believe that's the old Content Policy. Please refer to the updated one here: https://labs.openai.com/policies/content-policy
For further questions regarding the terms & policies, as WereSloth mentioned, feel free to contact support@openai.com
support@openai.com hasn’t responded to my seven emails in over a month. Any tricks to catch their attention?
Thanks, Major! It is similar. Basically the intent is - don't deceive your audience.
how about "a photograph of a cool monkey"

