#community-help
1 messages · Page 14 of 1
Alrighty ty
can I use the images for marketing purposes whether for an association or for a company.
Short answer is yes, but please have a read of the policy & the terms below.
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
Yes.
How to generate Heterocephalus glaber in dale 2? The scientific name does not work that well and I can't use the regular english name due to a random filter.
https://labs.openai.com/s/horHNd4xaZwiZQjGYLiMQkzw as close as I can get
not the exact thing but close enough.
Hi, guys is it normal to get generated letters?
that clear? rarely.
I just noticed I'm not getting my Promptober credits. Maybe I put my Discord name in wrong? I missed the window to fill out the form to rectify it. I think it ended yesterday, but didn't realize until today. Anybody know what to do?
Hi, what date did you sign up on?
Good question! September something?
I think I was too dumb to know how to input my Discord name, whether it was my username or the numbers after it or both.
heyy how can i buy credits with paypal?
don't wanna just throw my card info out there like that
if that's the only way then i guess i'll do it but wnated to make sure there was another way
sorry for the delayed response. please use the modmail feature to send this as an inquiry for review. thank you.
Thanks! I did. 🙂
I only just remembered that promptober is a thing that’s happening, and looking at my account I’ve also realized I haven’t been given any credits for the event (just the usual 25 you get at the beginning of each month). Should I do this as well?
Is it possible to change the frame size to 1080px by 1920px?
all outputs are 1024x1024, even when using frames for outpainting.
Hi, had you signed up for the event prior to 9/30?
Paypal is not currently offered as a payment method
Thank you. 🙂
ah, shucks
alright, no problem tho
how do i edit the size of the photo
head to #📚┃resources and check out the outpainting guide. Every generation is 1024 x 1024 (1 mp) though
Is there a way to delete generated images in my collection? A friend said I could make variations of their face but after I made them, they changed their mind and want me to delete them, even if they aren't public.
currently, you cannot delete images in your generation history. You can, however, remove images from your favorites
that needs to be a feature that's added because I feel like horrible crap as long as those images still exist
you can always suggest it in #placeholder-channel
well worth a shot
Yes, I filled out that Google form pretty soon after it was posted. I think I typed my username like “SaddleBag#3268”, but I don’t think adding the discord tag number to the end of my username would mess something up.
OK! Please send a message to the team via ModMail about this issue.
I have an update for you which I think you'll like! If you would like particular images deleted from your history, you can reach out to support@openai.com to get them removed. I hope this relieves some stress 🙂
thanks! That's good to hear. I'll do that when I get home.
Can you resize images on outpainting?
all generations are limited to 1024 x 1024 pixels (1 mp), but you can "expand" images via outpainting. #📚┃resources has some great guides on it
I mean for instance if I upload a picture I made on Dall-E 2 could I make it smaller so I can center it and then generate around the picture?
using the !pad command in #dall-e-bot is what you're looking for then 🙂
Ah thanks I'll give it a try!
Will there be more events like promptober? Like, one for Christmas maybe?
It sure would be fun. I hope so!
Why is Dall-e a prude?
is there any way to save all 4 versions of an image extension :(
?
Sounds like you have some feedback to provide. You can do that using the form in #placeholder-channel !
possibly "teenage boy"
Either remove teenage or change boy into male
Anybody got more of them credits?
How to get access to Dalle-2 API?
so whats with AI and cropping images? Like half of the head head is missing, etc. MJ did the same, Dall-e too.
there is a way to avoid that, by simply being specific in your prompt.
not yet
I know, but why does AI do that by default?
We’re developing a mobile app to socialize AI creators, and we’d love to access the DALL·E API. Maybe we could fill out some application form?
• We are currently testing a DALL·E API with several customers and are excited to soon offer it more broadly to developers and businesses so they can build apps on this powerful system.
• We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use.
that's how it was trained, mainly it's random, but in any case if you'd like to dig into it read research paper https://arxiv.org/pdf/2204.06125.pdf
Are the pictures that come out after typing the Prompts basically a Mash-Up of the 650 million + Pictures that are in the Training Database of DALL E ?
hi guys! i need an important question answered for clarification, if anyone can help me. are we allowed to remove dale’s watermark, because i’ve seen on many sites that we have the right to use these images for commercial use and i once saw that we could in fact remove 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.
thank you so much
for decoder training only dall-e data set was used approx 250m images, but for the encoder yes there were approx 650m. but it's not like these images what you're generating were already existing in the dataset
That is Super Awesome , Also This Project is Soooo Fun to Create things with , I just started today & now I'm Constantly thinking of different Explanatory Terms & Subjects to Create Images of , I have just 2 tokens left out of the 50 I was Given , this Project is Truly Amazing!
if you're just starting, I suggest first checking #📚┃resources #❌┃tips-n-tricks #top-tips-and-tricks
good luck!
Thank You Kindly !!!
I had a prompt happy dog smoking a pipe and drinking tea on a fall day in front of a barn, digital art which I generated about 20 - 30 images from, then I edited one, and from the results, tried to edit one of those. "Suddenly" this prompt violates guidelines. Like, seriously? Why'd it generate all the previous ones?
did you modify your prompt?
even if you slightly change it or change the order of the words it may cause violation, but you can always use synonyms, if you believe your content doesn't violate the policy
I don't quite recall, but I may have gotten rid of the double-space between the words dog and smoking, and made it a single space, but that's it
baffling
smoking has different connotation, you can check your history and see where it went wrong 🙂
a different connotation depending on the number of spaces? That seems bizarre - so like dog smoking, dog smoking, and dog smoking would all be interpreted differently? Not trying to be flippant, just want to understand what I'm dealing with
Am I one of the rare people that read the Terms of Use , Privacy Policy , Sharing Publication , & Guidelines ?
are you sure that's where your prompt went wrong?
Hopefully not. 🙂
I've read them several times, because I've had rejected prompts that clearly didn't violate anything. I don't remember anything about smoking though, have they changed?
I've tried myself the prompts you've mentioned, nothing seems wrong and there's no violation, so what I'm asking, are you sure smoking was the issue with your prompt
I guess I have no idea, it was literally the same prompt that I'd used before
I mean, I could try it again, but I don't remember seeing anything about how many strikes you get before you get booted.
If you get banned they will review your history, if there's nothing intentionally bad, you'll be ok. 🙂
Hi everyone. I'm new here. Is there a way to upscale the resolution from the default? thanks
The is not a way to do that currently
ah thanks
Is the Stock of the company behind Dall-E available? What's the abbreviation if so, I think it would be a good investment.
does anyone know how accurate the AI is on game related generated art?
dont think so. its both non profit & for profit as mentioned in the wiki. im confused af
What does this really mean in the terms of use page?
Like if I use outpaint and upload a photo of myself, does OpenAI have the rights to share the generation (with my face) and have full rights to do so?
If I create a generation using an upload of my face, and then after some days share the same photo of my face on Facebook, wouldn't OpenAI have the rights to take it down, as they have the full rights to generations containing the same image, and the image on Facebook is in them?
Hello, it’s probably to late now but I filled out the form on October the 2nd because I did not get the five credits. Is it still possible to get them?
Nevermind I may have used the wrong email
The last day to sign up was September 31st 😔
if you're not able to generate anything that goes against policy, and grants you full ownership of your own generations, what basis will openai have to take down your generation after couple of days?
but in any case I suggest directly contant support@openai.com to get more detailed info regarding licensing
What's with the stigma against mental disorders?
Why is "borderline personality" filtered?
Since when is having BPD automatically conducive towards self-harm?
I'd say "depression" falls in line with your limitations no?
Im with you on that one, but what I'm trying to know here is the rights that OpenAI has over images you upload
the terms state that you grant OpenAI ownership to all generations, which is all right
but if your entire upload is visible in the generation, doesn't OpenAI have the rights over the upload, technically?
regarding hypothetical matters, it's really best not to assume and directly have a conversation with support@openai.com
Is smoking cigarette/weed/cigar not allowed in prompts? I can't seem to add it to the edit prompt.
it is allowed
odd then I'm getting a content policy error
what's the full prompt?
Hello! How can i use the dark theme on Dall E 2?
hi, can you elaborate a bit? what do you mean by dark there?
@golden ruin sent you DM
I saw on the Tops, Tips and Tricks channel, some cropped images where it appears to be a dark theme on the site, if not a theme it must be some extension, idk
oh well just any verbal prompt of same meaning will work. use depicted, or prepositions like "on,at.."
is this what you mean?
I believe they mean that the website supports a dark mode/theme
no, but thank you Ene!
I believe it was not on the Dall E 2 website, but an extension, and now I've found one.
if you're referring to illustrations on dark background, words like "depicted or painted" will do a trick
I don't know if the website itself supports a dark theme. You can always get an extension for the browser 🙂
ahhh.. that's what you mean? 😄
I did it.
Treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.).
Sorry if it felt offensive, but it wasn't. I appreciate the help guys! @golden ruin @regal lynx bye.
Do the 15 free credits each month stack or are they gone if i didn't use all?
gone bro
Oh well that sucks
• 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.
Thanks
🙂
Hi, when i verified my account it used one of my credits. Was that intended?
Yes
Coolio, thanks
How can I get my hands on dall e 2? I pressed sign up on the dall e 2 page but it kept taking me to the dall e 1 page.
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/
How i can use outpaintig? I can't find it in google
Does anyone know why I'm not ever getting my 5 daily credits for promptover?
They go into paid credits. Open a mod mail with your dall-e email please and we’ll get it looked into
Alr, am I able to do that on mobile? 'Cause I don't have my computer on me rn
Ohhh icic, mb
Would I just send a dm to the bot? On mobile
Oh
How did I not see the thing above it lmao, alr
Well I dm'd it, hope it worked
I sent "help" at first bc for some reason I thought it'd have commands
Sorry, the pertinent part of the content policy would be
Public and personal health: the treatment, prevention, diagnosis, or transmission of diseases, or people experiencing health ailments.
This is in place to protect people from images that put the disorder in a bad context. Please go to #placeholder-channel and put in a suggestion to allow it. Apologies for the inconvenience. It’s beta and they’re trying to err on being protective. 🙂
How do I get DallE to generate a V shaped visor
Ask in #❌┃prompt-help 🙂
I hope this is the right forum for this question. I am a beginner with DALLE. I have a copyright question that confuses me. In the #gpt-models forum it mentions that OpenAI owns the copyright of all generations. Don't understand but OK. It also mentions you should never tell anyone that you handmade your picture if DALLE actually made it. I can understand that. My question is: if you generate a picture, then use it to handdraw/handpaint your own painting based on the AI image, could you legitimately tell people you handmade that image?
I don't understand why you say OpenAI owns the copyright
Otherwise, I can't see any issue with the situation you describe. It's not fishy to say that is an handmade image if you draw it by yourself, even if the generation is provided by DALL-E
are we able to tell if a generation was made with a free or paid credit?
Free credits will always be used first
Did the content policy change?
The content policy has changed on occasion when updates are announced, such as the ability to upload human faces again.
It was last updated on September 19 according to the site itself.
Is it possible to change the style of my own photo? For example if I have a photo of my face and I want it in anime style? Is that possible?
style transfer isn't currently possible
Can you take a real photo and outpaint it to add/subtract to it through a generation? In other words, take a photo of me, edit it to erase the background and then use a prompt to add a different background all while keeping your self in the photo.
As to my question about copywright ownership, here is what it says under terms of use #6. Ownership of Generations. To the extent allowed by law and as between you and OpenAI, you own your Prompts and Uploads, and you agree that OpenAI owns all Generations (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. If you violate our terms or Content Policy, you will lose rights to use Generations, but we will provide you written notice and a reasonable opportunity to fix your violation, unless it was clearly illegal or abusive. You understand and acknowledge that similar or identical Generations may be created by other people using their own Prompts, and your rights are only to the specific Generation that you have created.
As you can read, it says " . . . you agree that OpenAI owns all Generations (including Generations with Uploads but not the Uploads themselves), and you hereby make any necessary assignments for this. " It continues, " . . . 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. **If you violate our terms or Content Policy, you will lose rights to use Generations, . . . "
Image I posted to Promptober channel got flagged as explicit and refused to send? It was generated through DALL-E so idk what happened
I'm assuming it's something with the image because the prompt I can send just fine
"Anthropomorphic wolf wearing a devil costume, digital 2D art, sinister lighting"
I'm afraid we can't do anything about this as Discord itself is flagging your image/link.
I understand, just annoying that the cooldown still applies even when nothing gets sent
You can try to post the links in other channels and delete it if it gets sent through. And then post it in #🎃┃promptober when you're able.
Ok yea something with the image itself is the problem
Yep. As per your question about ownership, please direct your concerns to support@openai.com — they are the ones that will be able to clarify.
Discord's image moderation gets it wrong sometimes. Not sure why. Happened to me, tried posting a dog watching TV months ago and it got flagged.
I tried using a direct link to the image but it is refusing to embed it
Tried it as well. Discord is indeed refusing to embed it.
I generated a few prompts with the costume idea in mind so not like all is entirely lost tho
Even if not all of them are, spooky
Hi guys so I messaged DALL-E website form and website chat a couple of weeks ago regards how I give access to my team, or how we share a team account, and buy credit for the team, instead of buying credit for individual accounts, since sharing individual account is against T&C, still no response,
I wonder how much money DALL-E is losing because they don't have a sales strategy, other than a generic contact form and a website chat.
I've personally raised this to OpenAI.
I'm aware that support@ is currently strained because of the waitlist being removed.
I'll re-raise and hopefully you get a response.
Thanks, I do suggest they come up with a smart form that can delegate different support tickets, general enquiries, sales enquiries and then set priorities to urgent support and people want to give the business money
I am also not in their shoes so I can't imagine what is their workflow and if they are already doing that sort of segmentation.
Hopefully we hear back soon.
So if we share an individual account is a big no no at the moment correct?
Hopefully! I've re-raised your concern.
Last time I checked the terms, yes. Unsure if that was updated. You can find the link of the terms below. For now, your team can now individually sign up and access DALL-E as the waitlist has been removed.
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
Thanks
What is the "official media" prompt used for?
is there a list of past prompts on the day?
What are the keywords to create an object in full view or from far distance? I tried some words but did not work.
Have you tried adding the keyword 'isometric view' to your prompt? @zinc dune
@hollow dew tried now and it did not work either, I cannot get the full view of the oak tree, all creations are zoomed in
I prompt this: banknotes falling to an oak tree isometric view, water color style
I get below
I'm also curious to your question, please update me once you get answers
Hi, so I'm new to this and I would like to participate in the promptober, but I don't know if I need to do something before I can receive the 5 daily credits.
The sign up closed on the 30th of Sept. Sorry, you're too late
understood, thank you though
Isn't there a community page to view uploaded images? 
Not yet. Please vote in #1021688737572327444 to add your request 🙂
Do bought credits roll over to next month
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Alr cool
It's weird, I got 10 pumpkin reactions yesterday and I'm not in the hall of fame, or maybe I just did'nt look enouph but I'm not sure because I searched and it's not here
Major, I do not have any concerns. I am just reading what the Terms of Service say. They seem to say Open AI owns your generations. My only question is, the legal ramifications of that. IANAL, so I don't know.
Try your very best as to what you’re trying to create and I’ll do my best to help you out
wide shot, 24mm, large field of view, large fov, etc. might help
I will let you know.
hey i remember someone posting a tool that could make the background of images "shift" a bit more like a gif. like twinkling lights and stuff. does anyone remember that or have a link?
Hi I used to have around 40 credits but I logged in today and I only had 15. Is there any way for me to get these credits back?
• 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.
oh ok thanks 😦
The credit system is here to stay forever right?
Currently, you can only generate at 1024x1024
As far as we know, but the current system as we know it could change in the future
What is the outboard editor?
Will it be removed?
Cuz i aint jeff bezos
I have no idea, I have the same information you do
sad
Help, I need help. I wanna inpaint (not outpaint) an uploaded image. However, I can't get it to do what I want. I upload an image and it immediately bring me to the edit screen. It forced me to hit the checkmark to set where I want the photo. Then it automatically put in a blank generation space. I dont want a blank space - that is for adding to my existing picture. I wanna inpaint on the existing picture. Don't know how to get rid of the blank generation space.
go to labs.openai.com
and press upload an image
alternatively, you can go to the Outpainting editor and press the landscape icon with a + on it
this will allow you to add your image to the editor
Thank you. That is what I needed. You cant do it through upload. You have to do it through outpainting and "add image".
Whoops. I spoke too soon. After uploading the image, I have to hit the checkmark to place the image. When I hit the check mark, it puts in a new blank space. I don't wanna new blank space. I wanna inpaint on my uploaded picture by using the eraser.
I don't think so. It's just an innovation thing for now and that's why it's limited for public use.
The day will come and we'll get DALL-E 3, 4, 5 while the supercomputers get stronger and stronger.
Will the first 50 free credits expire the next month if they're not all used?
[❓] — How do DALL·E Credits Work?
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits 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.
from #gpt-models (emphasis mine)
Just to be sure, dalle images can be used for basically anything right? Including pfps, and even selling stuff that uses dalle images, right?
"users get full usage rights to commercialize the images they create with DALL·E, including the right to reprint, sell, and merchandise"
I'm not good with law terms which is why i asked, but i guess that's a yes, thanks
I just quoted the site but as far as I can tell you legally own any images you create using Dall E
I think they want people to specify the source being AI though I'm not sure
Anytime Mr mattress
A couple hours ago, I posted a comment about how to do an inpaint of an uploaded picture but haven't gotten a response. The problem is that when you upload the picture, DALL automatically adds a gerneration frame. I don't want or need a generation frame; I will be inpainting the photo I uploaded.
Oddly, you can add frames, but once added, you cannot delete them.
Hi! What happens when you crop the image, then edit? The initial generation frame should be around your cropped image
And be able to be inpainted
Yes, apparently DALL doesn't know how to handle uncropped images. If you crop the image, DALL does NOT add a frame. If you upload it without cropping it, it adds a new frame. Weird. I wish you could upload a non-cropped image. The image I uploaded was 1600 X 1200. Soooooo, I had to crop it to 1024 X 1024.
can i use the watermark bot command on a non Dall-E image?
Is there any plan to make Dall E suscription based instead of pay per attemp?
• 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.
no plans other than what is available now
god no pls
I think when the wait list ended I lost my original credits and now only have 15 . Last time I used them I had 33....
What happened?
The change for rollover?
if those 33 were from your original 50, and you joined before September 7, probably yes?
idk if it worked differently before ending the wait list or if there were any changes when it happened - I joined after it ended
That would have been a useful email
• 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.
anyone else having issues with dalle rn or just me?
yea
Think its down
seems like their entire api is down
Yep xD
Ah it's back up
Hello, is there any good way to add text to the generated images? Most times you tell the AI to add a text such as "Hello, friends" it misspells all of the time.
As you said DALLE is horrific with text unless it is an extremely common word/phrase: I'd recommend just adding the text in Photoshop or whatnot.
@zinc marsh Stated it perfectly, Text does not work well, I would recommend trying some free online tools for adding text like Gimp!
Can I refund 1 token ^^? I kinda wasted it by accidentally inpainting a filled pic, so the AI did nothing.
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
screenshots are broken there
What does !composite command do with the dall-e bot. It doesn't stitch the images back together in the end, is it broken?
I uploaded images but want to see how it stitches them together
So been hearing a lot of news. How ethical is Dall-E in terms of sources
I know DallE uses some source images for Machine learning but with a competitor basically doing something really stupid (as in art thief) I want to know if there some transparency or ethical policy on how Dall-E operates.
We only know what's been shared in the OpenAI blogs, but you can reach out to support.
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
@patent rain may be able to answer that better than I can
the question is very sensitive but I think with a recent development I hope Dall-E would provide some transparency on where they are sourcing images from
and how
if competitor has to resort to using AI to steal art he is kinda bad competitor
I mean, both of them create entirely original images, it isn't like one is a search bar for art and pretends to generate new stuff #ai-discussions
Does anyone know how to get Dall-e to generate the type of art style used for lo-fi beats backgrounds? (e.g. Lo-Fi Girl) I can't seem to find any examples that really look like it...
perhaps upload lofi girl and ask for variations?
does seem to be anime-esque
I want to generate vastly different things using the same art style...
style transfer would be a great feature
Is there anyway to prompt dalle2 to produce an image at a lower resolution than 1024x1024?
or ability to somehow "teach" dale a few of your own pictures and generate something based on them
another idea: lowres previews. Generate say 128x128 thumbnails and use a credit to render those normally, after you know you'll like the results.
I wonder: would generating those previews use less computing than 1024x1024 ones?
I might submit these in #placeholder-channel ^^'
• 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.
Is there like a DALL•E 2 app or something? That'd make it easier to access.
Just the website at the moment
Treat others the way you would like to be treated, and assume best intentions. Don’t harass or attack others, and don’t engage in hateful or generally malicious behavior (e.g. sexism, racism, homophobia, etc.).
This channel is for questions. If you have feedback, please see #placeholder-channel
Nah I think I'll pass
Hey I got a question
Why do I have the app? Is it real or an illusion cuz I'm confused rn
That's most likely an icon that opens a browser to the DALL E 2 website
It's not a direct access it's an app
as far as I know there is no official DALL E 2 app
Wondering if anyone is familiar with any of the other AI picture drawing programs/websites out there. Just wondering what you thought about them.
I clicked the three dots on Google in dalle website and it downloaded that
Please see #ai-discussions to discuss other AIs
Yeah, as previously stated, thats most likely an icon that leads to the web interface
Basically a shortcut to a browser window
Oh ok
when I'm editing an image, do I describe what I want the whole picture to be, or do I describe what I want the little erased part to be
logically you have to describe what the erased part is meant for, but leave original image definers, like mediums and authors, so your generation matches
thank you
I'm having problems uploading an image. Nothing happens when I click on upload image on the first page and same in editor.
Anyone know whats up and can help me?
just tested, seems to be working fine, once you click on upload and choose your image, you can click on crop photo, crop it and edit it afterwards
Yes I know, I'm saying the menu doesn't open for me to choose an image no matter what I do
might be browser or network issue, or cookies and caches
Are you on mobile/tablet/PC?
PC
Which browser?
Edge
Strange. It should work on Edge. Do try Firefox/Chrome, I use those and it works on my end.
I'll open chrome and try just to see but chrome eats my RAM like nothing else
It works yeah
What can I do to make it work on edge you think?
Apologies, I have no clue. Maybe others can chime in.
Alright, thanks my guy!
there is on mobile. i forgot how i downloaded it but it is somewhere on the website
Ik it
And btw can i change my dalle accounts pfp?
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
Hey help me pls i reported an image by error
Image was ok
Am i gonna get banned
I wanted to report it doesn't match text
But I clicked it's inappropriate
:((((((((
You won't be banned for reporting an image
But I did it wrongly
Even if OpenAI knew it was a false report (which they can't unless they read this conversation), you wouldn't be banned for it
any idea why I can't prompt for cigar smoke without being threatened with a ban?
Could you share your exact prompt in #❌┃prompt-help please? 🙂
really, its just "cigar smoke"
The prompt "cigar smoke" is not being flagged when I enter it. Are you using DALL·E through labs.openai.com? If so, it might be worth reloading the page and trying again.
I reloaded, removed an image area, typed in "cigar smoke" and the threats continue
It looks like this request may not follow our content policy.
More requests like these may lead to an automatic suspension of your account, so we recommend trying something else.
If you're trying to do an outpainting/inpainting of an image with a cigar in it, that's almost certainly what DALLE is flagging, not the prompt.
The program will refuse to touch any image that contains a cigarette, cigar, pipe, etc.
there is a cigar in the image
That would do it.
No, it just refuses to edit in any way images containing cigarettes/cigars, even if it generated them itself.
thanks, seems pretty worthless especially if it created it
I think y'all capping
I think y'all capping.
Lmao.
And yes I wasted a token on this.
And btw can i change my dalle accounts pfp?
The issue Sven was facing was that DALLE can generate cigars/cigarettes, but refuses to edit the images containing them afterwards. (In his case he had an image containing a cigar he was trying to inpaint cigar smoke into, which was triggering the filter.) Also, no idea about whether you can change your profile picture on the website or not--I suspect the answer is no (especially if your account's linked via Google or the like), but you could try contacting the OpenAI Support email.
Mhm
Contact support for account related questions
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
And my account is not linked to Gmail, it's an @ martinstte.com email so I should have vip stuff
Are outpaintings usually downloaded without the watermark?
is it possible to get more credits via paypal ?
DALLE automatically applies the watermark when downloading an outpainting, but it's possible (and entirely allowed) to circumvent that by manually downloading the bottom rightmost generation from the Inspect tab and manually stitching it on. (That's the process I use, at least.)
To my knowledge OpenAI doesn't currently support PayPal, though you could use another method.
are swimming pools/people on the beaches allowed?
I don't see why not... are you receiving a warning when you try to generate them?
no. I just want to post a shirtless clay guy in a beach setting and I'm extra careful.
https://labs.openai.com/s/AMIScRTzLDQhB9UaE3z5vCi6 - I mean stuff like this
what method are they offering beside visa cards? Cause I dont have one unfortunately. many ppl would be happy to be able to buy via paypal pretty sure
I tryed to generate a promt and a message on my screen popped up
"Incorrect API key provided: sess-8E9*********************************CQI7. You can find your API key at https://beta.openai.com."
What should I do?
Is that message repeatedly coming up?
I believe the only way to pay for credits atm is using the provided payment method, which to my knowledge is only via a credit/debit card of some kind
If I post a picture here from an original pencil drawing then enhanced with DALL-E 2 where do things like copyright stand? also if posted here does the picture become public property?. Newb so i gotta ask :).
yeah I know its a pitty :/ I wish they could add simplified methods like paypal, it would immediately help with ppl buying more credits etc
Hey FAM how can i get high resolution?
How do you do outpainting
@spring dock @sacred reef — Only way to get a higher resolution is through outpainting. More info here: https://help.openai.com/en/articles/6516417-dall-e-editor-guide
@sacred reef did you figure it out? 🙂
Are prompts related to violence such as blood, knife, chainsaw is allowed?
Absolutely nothing depicting blood and violence. A knife and a chainsaw may be all right depending on the subject.
But not depicting violence.
Every outpainting picture I make gets a white/translucent background. Is there a way to just get the image I make?
Do you have an example you can share?
It shows up white in some place, as dots in others and on discord it is translucent.
I'm currently manually trimming down the pictures, but I seem to be the only one with this issue
To get just the picture itself, you have to use the Download button just above the Generate button. (Note that that will also apply the DALLE watermark to your image, though; if you want to avoid that, you can just manually replace the bottom-rightmost part of the image with the corresponding raw generation afterward.)
I use it everytime
I use the download button but it doesn't apply a watermark either.
interesting.
Huh, it works just fine for me--are you sure you're not e.g. clicking "Save Image as" on the right-click menu or the like?
I wonder if you have a browser extension installed that is interacting in an unexpected way with the site.
Well, that is a clue right there.
It's almost certainly downloading the entire outpainting canvas--dots and all--then, instead of the image itself.
It didn't always use to have the extra padding around the image, it started a couple days ago
Do you mind trying on another browser?
Yeah I can try it on chrome in a bit later
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.
Check #📚┃resources under outpainting
Can confirm it is working on Chrome. So it must be bugged after the last Brave update.
I don't believe I ever got the watermark from outpainting even before this issue.
did yall get Corpse to do the video?
That’s our community volunteer @pine void
Terrific lol, makes me think of thatbassvoice guy on tiktok
i may have missed it, but i swear ive read the whole promptober post in #👻┃event-info . What is the topic for this week?
also how do i sign up for the 5credits for promptober?
Is there a way to not pay credits for dalle for a generation and do it with your own gpu?
I have a really good gpu and I am kinda broke so is it possible
Staffs in this server have extra prompts for free? Just wondering...
everyone comes up with their own prompts themselves and if there's something interesting or helpful we share it in #📚┃resources #❌┃tips-n-tricks #top-tips-and-tricks
Some of us buy extra credits
But we are just volunteers. I can’t speak for openai staff members
No, there is not
I keep getting this issue, tried different browsers, restarted my internet, tried vpn, nothing seems to work
I could contact support but idk what the wait time usually is, I was wondering if anyone here might have any solutions?
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
Same here.
Possibly DallE having issues
Can confirm, I'm receiving the same message on mine
awh man
Just give it a few. It happens 🙂
Same here.
Same boat.
You can subscribe at https://status.openai.com
Welcome to OpenAI's home for real-time and historical data on system performance.
also getting issues. No prompts have worked for like the past hour
hey, so I asked this question a few days ago, but nobody answered. I was wondering if I can use non-Dall E images to use the watermark bot command with?
I seem to be back up and running
Where do people get ideas and inspirations for prompt words? I'm trying to curate a list of a few hundred generic words that when you combine three or four of them in a sentence generate really awesome images.
I feel like I'm always gravitating to words like "Astronaut" and "Robot" 😅
That's so helpful thank you. I've requested access, I think it's private.
did you test how unreal engine 5 compares to unreal engine / unreal engine 4?
I have not
how do you accès the outpainting feature
Are you on PC or mobile?
A bunch of paid credits showed up on my friends account, but they haven’t purchased any. Anyone know why?
Was it promptober?
Did they sign up for Promptober? If so, they should be getting 5 credits per day for the duration of the event
Ah, yes they did. Thanks!
Will Dall ever be cheaper?
or a monthly payment based ai?
15 dollars is a lot for 100 uses
awesome
hi, how can I get anime images or images by video games?, if i typing an anime name or video game the DALL-E-2 don't works
#📚┃resources may be helpful!
Hi, I completed Promptober and filled out the form but don’t think I got the free credits.
why is the name obama banned
When did you fill out the form?
You can't generate celebrities or famous public figures
When it opened on September 28 or smth
Contact support@openai.com
is there any way to refund credits if the generated images turn out to be wrong? or maybe its still being developed?
No way at the moment, but be sure to add your vote in #1021688737572327444 for that 🙂
Is there any way to generate TEXT and Fonts? I do not see any examples as such.
DALLE was trained on images, not text, and thus it can generate images, but not text. (There are of course some exceptions: it can usually generate extremely common words/phrases like "Stop" or "Happy Birthday," and I have seen people use the program to create words in specialized typefonts by generating one letter and then outpainting each subsequent character, but in general you're better off adding the text manually then asking DALLE to do it for you.)
Hey there, is there a way to use the OpenAI API to generate images? In the docs I could only find answers to texts
• We are currently testing a DALL·E API with several customers and are excited to soon offer it more broadly to developers and businesses so they can build apps on this powerful system.
• We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use.
why
I believe it's because OpenAI wants to discourage deep fakes. I don't work or represent OAI, so I can't give a definitive answer
We ask all users to please stay on topic in order to keep conversations focused and productive. Please be mindful of what other users in a channel might find helpful or interesting when posting.
how do i delete a dall e account
You're going to need to contact OpenAI's Support for that one, I believe.
(Copy-and-pasted bot answer because I don't know the command to trigger it:)
"[❓] — Q: How can I contact support?
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com/."
hey so, just to be clear, are people who are long dead ok for us to generate images of?
hi,, can I make the images have another resolution?
Currently image creation is locked at 1024x1024. You can use outpainting to expand the image, but all generations will be 1024x1024
thanks :)
Hey! I recently got this email (attached below). How do I access Codex? Im looking to make a video game with it
This server is Dall-E 2 only. The OpenAI forums are where you'll get a better answer than here
is there someone who can help or a codex server
I don't believe there's a codex discord server. The OpenAI forums will be more helpful than here
is there someone here who could help
For the third time, this server is Dall-E 2 only. We do not do anything related to OAI's Codex. Please ask your question on the OpenAI forums.
is there no paypal method 😭
Hi guys, where do I go to use the Outpainting feature?
Currently, no
Thank you!
💀
indeed
I haven't been able to participate in the promptoctober or whatever it was called due to work, what do I do with the points I'm getting?
i mean, I love free points but I feel like I'm breaking terms of service or something lol
How do you make like enviroment concept art with this ai like with it being all blurry?
hi! can you be more specific?
Hi all, whe you created an image, others users can only see the images that you share, otherwise are they private?
Correct
HI everyone, after several tries I've been not able to find relevant and sure answer to this question : can you use images you generated with Dallee2 for a commercial use ? Like straight from dalle25 to a record cover for example ? thx
*from Dalle2 not 25 😅
Yes.
great, thanks
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.
Like I’m making a short film and need some concept art for some areas but when I do it even when I make extremely specific. It’s just all blurry.
can you show us the screenshot? what is blurry?
Like it’s visible, it’s just a bit blurry might be just trying it’s hardest
If I were to cite an image generated y DALL-E, how would I go about doing that? Also, is it ok to use DALL-E in personal projects because they are transformative works or is there osmething I'm missing?
I sent the following prompt: round logo "cocktailbilder", modern
And the prompt got flagged, any idea why?
Using the editor, I was trying to follow this article, near the bottom he has a step: Crop the image to the left, erase the blank part, and click Generate.
I am need to learn from step 4 through 7 to make different aspect ratios of art
Crop the image to the left, erase the blank part, and click Generate.
Save your favorite version of the inpainting.
Repeat steps 4 and 5 for the right side.
Bring both halves into Photoshop and merge them together.```
Is this relatively accurate information, I am new to DALL-E
I will continue to find good information about the editor and hopefully more good articles, could not link the one above, but if you have good edit tips for panoramic art, or tall art etc, please let me know and thank you.
I have a Photoshop sub so might be I just elaborate on ideas in there? ty
If I share a watermarked image will others see the word seed in the metadata?
Hello, I can't seem to find an option to change the password of my DALL-E account on the site. Where can I find it?
• Logout or open an incognito browser
• Go to https://beta.openai.com
• Navigate to the login page
• Click ‘forgot password’
Still not working? It’s likely you’ve authenticated using your Google or Microsoft account. In this case, your password can only be managed through those accounts settings.
is there a way I can take an image I've already generated and have the AI give it a new theme? I got a picture I really like but I want to see about getting a Halloween style to it, even taking the same input and adding Halloween to the end of it gives me new stuff
Why are people cropping their images in #dall-e-bot ? I dont really see the fuss
Hi! Cropped images can be used in D-E2 to outpaint - that is add more to the image and create a bigger scene that features the original image.
Hello. You can edit an image and change parts of it through inpainting. D-E2 does not do style transfer for a whole image based on another image.
Hey! The PNG file does not have prompt metadata in it. If you'd like to check you can use what is known as a PNG chunk viewer.
Hmm, But cant you just already do that in the editor with the Outpaint selection tool?
Hello! Cocktailbilder may be throwing it off. Please drop a comment about this in #placeholder-channel
Yes, now you can but that is a fairly new addition to the generator. Also, some folks just want to stay in the 1024x1024 diameter
thank you my friend
almost like the images are not meant to be squarish
Hiya - yes, this can happen from time to time.
Oh so people also just use the "erase tool" to basically outpaint?
*Erase I mean the cirle tool one to change a specific part of an image
You can try using photography terms and portrait terms to see if that improves it. OR
You can outpaint the ones you like to add missing parts
#❌┃prompt-help is a place to ask for more info on that
Yes.
it seems constant on my last prompt, that's why i'm surprised
Ah Gotcha! Just to sum it up its just another way to outpaint
Bad luck? Every seed is different.
Anyone here in this server that works with OpenAI's engineering team?
It'll be helpful
Do you have a specific question?
Ahh yeah, I'm using the OpenAI's API that uses GPT3,
It has one feature that can convert Movie title to emoji.
Im using that API in my app
That api works perfetly on the website
But in my App the emojis are not rendered
Example: -
This server is focused OpenAI’s DALL E 2
I think you’d have better luck contacting support@openai.com
thanks for this email
ill contact
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
Thanks for this email once again bro. I sent them an email.
Is it possible for the ai to create artificial texts? I tried asking it to make a painting of a poem but it was all gibberish.
Hi, D-E 2 cannot consistently create images with cohesive text/words.
You may find limited success with one or two words.
Okay thanks
Is DALL-E 2 available throught API?
from: #gpt-models
[❓] — Is DALL·E available through an API?
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.
Thanks
When will we get access to the net trained on 20B imnages?
This one seems to not even be on par with Stable Diffusion 
I'm guessing it's still too compute intensive?
The bigger net can do it... but that's not the one that's shared here.
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.
Hey! Will the credits I am getting from Promptober be transferred over to next month?
Or will it reset to 15?
I believe the credits from Promptober count as paid credits, but I'm waiting for an official response to clarify.
I signed up before the deadline for promptober, still no credits
please send a modmail with the email address you signed up with.
They will expire September 30, 2023.
Is there any way to retrieve an out painting job or put existing generations together? Construction workers accidentally severed the cable.
you could download all the individual generations and use the editor to overlap them
I can use the upload button to add back a lost image?
all generated images will be in your history. If you want to overlap them again creating an outpaint, you could upload all of them back into DE2 and overlap them in the editor
So I'd have to download onto my computer then reupload or no?
correct
Okay. Thanks!
Working on expanding this. Afraid to do anything yet in case I do it wrong and lose a credit. What do I need to do in order to complete the top of the head, and do I need to type in details in order to keep the art style consistent?
I need help. I never got my Promptober credits. Everyday i check and it's always the same
And yes, I did fill out the google doc/sheet
#❌┃prompt-help will likely offer better results for this
dm @agile karma about this issue please
I keep getting this error no matter the prompt or image
did u read the error?
Celebrities or other people who have lived within the past 250 (I think) years are prohibited to prevent deepfakes.
Ok sorry to bother
Ok my account is deactivated, can a mod please investigate. What I put in did not break any rules.
The prompt I put in before I was banned was beautiful woman. Does that violate anything???
Copy-pasted from the DALLE bot:
"[❓] — My account got deactivated. What now?
• Please follow the instructions in the Help Center article linked in the deactivation note to appeal.
• We have been and will continue to reinstate accounts when we review your prompt history and don't see ill intent with attempts of bypassing our filters for policy-violating prompts."
Hi there, unfortunately, we do not handle account related issues here. Please contact support@openai.com for your concern.
Do you know any art generator that has a public API?
Let me look into that, I dont even know what is a unicode
short version, there are multiple ways to represent characters. The old one is ASCII, which is limited to 256 characters, and can't expand at all. Unicode is the modern standard, which aims to support every possible character anyone has ever physically written, including this super rare 'o' which showed up in like, one cyrllic translation of the bible: ꙮ. It also includes emojis and other symbols that might be useful
Because of this, unicode characters require more bytes to encode - if your program reads them as ascii instead, you get groups of multiple garbage ascii characters for each unicode character
this is a question for #ai-discussions
ohh i see
Can you make it to watch a ad to get more tokens
Your best bet is to suggest that in #placeholder-channel, though I am sure the OpenAI team has received that request multiple times before.
Getting a content policy warning for some song lyrics, would it be that it's interpreting it as spam?
Most likely something in the lyrics. Have you read the content policy?
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
Oh, lovely ;)
Yep, nothing in the lyrics even get close to what should flag on the content policy
Post the lyrics?
So what do you wanna do, what's your point-of-view?
There's a party soon, do you wanna go?
A handshake with you, what's your point-of-view?
I'm on top of you, I don't wanna go
'Cause I really wanna stay at your house
And I hopе this works out
But you know how much you broke me apart
I'm done with you, I'm ignoring you
I don't wanna know
Ah, maybe the "I'm on top of you" is enough to flag it
that would likely be it. 🙂
How do I delete a source image I’ve uploaded? There doesn’t seem to be a delete function anywhere?
At the moment you can't
Can someone manually remove an image for me?
you can contact support
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
Thank you
you're welcome
I have sent an email to support and have not received a reply.
What should I do?
Help me.
it will take some time
Thanks for your reply
I was worried because it already took 15 days
I'll wait a little longer
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
@subtle shell
Thank you, I'll check it :>
How do I tell DALL-E to please keep my main pieces I may really REALLY want, like the swan variations, completely visible so I can use them?
love this stuff, ty for inventing for us!!!
i am not sure if I understand, you want to edit your generation without changing swan or?
I was hoping the variations would not have clipped swans
is there a way to prevent that?
you mean cropped, you can verbally prevent by adding words like "fit in the frame", full body, full shot, in the middle etc... like logical cues, if not, you can simply edit in outpainting tool and expand the area, hence show the whole image of the swan
Okay awesome, will ramp up in edit/outpainting tools asap. This tip should really round out my word seed experience!! ty
make sure once you add a new frame, include the original piece, in this case part of swan, so the system understands that it should follow the style. don't repeat the prompt when giving new prompt, it confuses the system
Sounds like your saying, with DALL-E skills this piece is savable. nice!
or labor photoshop time 😄
Love this tool, feels like a revolution
is it possible to get a image with transparent background?
or at least that green background to remove later
in any case you're able to remove background from any image without having specific green background
as a general process question, would be better to photoshop/massage my original image before clicking variations? (and then re-upload for more variations). -- or just let variations do the work now that I have centered? What is the more experienced process of the two? ... actually might be 3 routes.. if I include the Edit flow in DALL-E.. 🤔
lots of homework.. try them all I am thinking..
through commands or manually?
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.
I mean manually
Anyone know how to get free credit ?
oooh, okay!
also, i'm trying to get an oil painting of tulips, but with the whole plant inside the drawing, but "complete drawing" and "full image" didn't work... any suggestion?
you get 15 free credits every month
not the first day of the month, but 1 month (and 2, 3 4...) after your registration date
oooooh, gorgeous!
but i was thinking about something more or less like this
but with a tulip instead
i apologise for not expressing myself correctly
oh I see. something like this i suppose
yes!!
anyone else getting this am?
What happens if I report something that didn't generate related to the prompt
good, if you see there's a glitch or something, report
prob network issue, try to refresh the page
did that but same message
when did you get this message? it works fine so far
lik 15 mins aago
still getting it now
im using it right now, works fine, try different browser or clear cookies and caches, it may help
I just tried and it still the same problem 😦
Ahhh i think i found out the problem, trying to recreate a film scene with a gun so i guess the word triggers it?
you should have specified where you were getting error message. while some guns are allowed, it depends what you're asking in your prompt.
ohhh nope i guess even after i removed the"gun" it still gives me the same message. weird
what is the full prompt?
is there a way to get a picture without the watermark or are we aloud to remove it?
I also got a warning for "It looks like this request may not follow our content policy."
full prompt here: Film still of Tony Leung Chin Wai in leather jacket talking on the phone sitting on the edge of a rooftop building, view of Hong Kong Victoria harbor morning in the background, wide shot, from action-crime film “INFERNAL AFFAIRS” (2002)
Hi is there a DALLE 2 API?
I gotta believe in this concept and run with it 🙂 time freeing up soon...
I sure hope so! I have some serious GCP (google cloud platform) ideas...
Hello, does any have an good links to guides on how to use outpainting? In particular insterested in how I can reflect uploaded images in the original image (if that makes sense!)
• We are currently testing a DALL·E API with several customers and are excited to soon offer it more broadly to developers and businesses so they can build apps on this powerful system.
• We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use.
@sour grail Please refer to the message above.
There are some guides in #📚┃resources plus an official guide from OpenAI: https://help.openai.com/en/articles/6516417-dall-e-editor-guide
You are allowed to remove it, 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.
No celebrities allowed as per the content policy. You may refer to the links below to avoid unnecessary warnings.
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
I did find, that if I mention a prominent artist style, that works okay
Yep! Those usually work. There are compiled artists/styles/lenses in #📚┃resources , be sure to check it out!
perfect!! Style templates of others too learn from!
I like the DnD one for sure, gonna try it
Should I report a result if it is a pig with an arrow in its head? my DnD....
this one is borderline...
Is there a way to test Dalle-2 API?
I have a seed phrase reproducing the same results over and over, is this possible?
are u using the editor?
mmmm maybe I was mixed up in Edit mode..
Hello! There is no public access to any D-E 2 API at this time.
stuck, so if I erase a section in the editor, and I am done erasing, what is the next button I click? I uploaded an image so there is no text in this case (can share screenshot if that helps)
do I bring back the seed text from original, and erase, then click generate?
oh I see. I will say flower the birds are sucking, then generate and should fill the space I erased
Is it possible to shrink the outpainting generation frame?
Hey guys, I still can't upload images for edit
I'm trying to press this button but nothing happens
it is always 1024x1024.
hello, are you still having this issue?
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
??
just a shortcut to the home page
fixed. macbook lagging
Reboot helped
i'm 10 days late to promptober lol does anyone have a listing of what the "word of the day" was for the first 10 days?
You can check the pinned messages in #🎃┃promptober.
(After a brief check: Harvest, Lantern, Oktoberfest, Cozy, Barn, Costume, Basket, Cinnamon, Quilt, Hay, and Pumpkin Spice.)
Nice thanks!!
Yo guys, I'm not able to log back into my account for some reason, looks like I forgot my password, and when I ask for a recovery email I don't receive any email... I've already checked the mail I inserted, and it's correct
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
how to get dall-e 2 access?
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/
How do I change the format size when outpainting?
all generations are limited to 1024 x 1024 pixels (1 mp)
I feel like there was something lost in translation there. @slender cave @regal lynx
How so
1024x1024 is the size of a standard generation, and a generation frame in the outpaint tool
Are you saying you'd want to outpaint a larger or smaller section?
I wanted to make it taller
so that it would be outpainting like, 1024x1536 for example?
or you want the entire image to be taller?
if the former, you can just overlap the generation frame. I realize it's expensive in credits to do it that way, but it will work.
but for the later, you can just add generation frames where you want them, and then save it to fix up in Ps.
Alright
I sent an inquiry from chat cube about 10 hours ago and waited for 4 hours but did not get through. I also sent a new email but no reply so far.
I'll wait a little longer or help me
I do receive bills from them
Their replies are not in my spam box
Hi there, unfortunately we volunteers do not offer account support here in Discord as we do not have access to OpenAI's systems.
Is the artwork used for Dalle2 stolen? I'm worried about the ethics of like using anything made from the ai if like its a trace over another image like others are claiming
Is there a way to cross link a post from one Discord category to another?
Thank you for your reply.
Hello, you may refer to the DALL∙E 2 GitHub preview linked below, scroll down to "Model training data", the first paragraph should answer your question. https://github.com/openai/dalle-2-preview/blob/main/system-card.md#model
Thank you for directing me to that!
Unfortunately Discord does not offer such functions.
That's unfortunate.
Thank you.
Hi. are we not allowed to post any videos or is my discord acting up? :)
Can you not post the discord link?
No, sorry. No external links
It’s not a link. Just a video
But ok, that explains it. Was just confused and thought my discord is broken
I've had the following message at https : // labs .api .openai .org / for several days now. Any tips on what I need to do?
Our services aren't available right now
We're working to restore all services as soon as possible. Please check back soon.
0tZ5GYwAAAAAudQ0fXcMrRopfXnNi2rUiR1ZBMzBFREdFMDIyMAA5Ym...
• We are currently testing a DALL·E API with several customers and are excited to soon offer it more broadly to developers and businesses so they can build apps on this powerful system.
• We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use.
yeah, it's weird that it says API in the URL, but that seems to be the URL I have always used to access DALL·E. What URL should I use instead? openai dot org says there is no website at that address...
I'm in...
it was COM, not ORG. Thanks for the tip!
Try labs.openai.com
Thanks it worked
hi all, I have a dalle-2 account but I am unable to sign up to use gpt-3 with the same account. Help!
If you can’t log in using openai.com you’ll need to contact support
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
so the bot created a sign in my one image that said "EKloenijoes", I generate with that word and i consistently get bowls of fruit/food. Whats up with that
now it just gave me "ELUO EDLBE" which is cooked food, mainly meat..
Seems like the word edible could be interpreted from the bottom one? Also it has information from different languages, so maybe that?
probably gives you the closest to what you might have mistyped
oh im just inputting what it specifically gave me in a sign over a store. lol
okay after reading the FAQ, I wanted to see if the image completion tool is available, like if i have image and I want it to complete a portion of it? I had seen this before
If you're talking about outpainting, there are guides in #📚┃resources
okay perfect
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.
oo thank you Dall-E!
i accidentally typed into the promptober chat and got refrain from participating for an hour welp
Was wondering after I edit an upload, it will gen my 4 images after i phrase and generate, what does the Accept button do?, in my exp. last time it generated again, I don't want that cost so now I just click away and get prompted if I really want to click away..
, how do I operate in this area correctly using the navigation arrows ( i get I can preview my 4 generated pieces, but not sure why Cancel/Accept button) ty
The accept button confirms which of the 4 generated images you want to remain in the generation area.
Besides outpainting, is there an option to let the AI "redraw" an uploaded image?
Variations
Where can i find that?
Ohhh, i see the option for generated images. But is there an option to upload an image from my PC and then generate variations?
i think you can't generate variations of uploaded image form pc
i was looking in editor for generate variations and can't find
I'm trying to learn different styles, so I can best describe it when generating images. Need some help on what kind of "art style" this would be.
I tried digital painting but its not matching
Ask in #❌┃prompt-help ! But I would consider looking through #📚┃resources which has a lot of different art styles. This one appears to be a children's book illustration of a haunted house. Possibly air brush?
It could also be digital art
A good strategy might be to find a similar media/artist
You absolutely can generate variations on an original image. You need to upload it, crop it, then choose generate variations.
The crop can be just clicking "crop" if it is already a 1:1 image
Outpainting doesn’t seem to work on my iPad, in any browser, even with desktop view on. I don’t own an desktop computer, has anyone had success getting it to work on mobile? It seems to not be able to use the image select and accept menu that pops up
it's not been announced as supported yet, and i haven't been able to get it to work on mine either 😦
This has been asked before but I cant find any clear answer. Is there a description addition you can add to prevent the generation of words/letters?
couldn't quite get it.. so you want to prevent accidentally generating words?
Yes. for example if I try to make Halloween themed buildings, it keeps trying to write some form of word that resembles Halloween on the side
What is the prompt you are using? Sometimes tweaking it can help
Looking back I think it has to do with the word "Poster" or the fact that I was generating variants of a photo that has existing wording in the background. (I was trying to use part of a building to generate a haunted house)
@short salmon Can I use DallE 2 for commercial purposes or is that only something allowed for DallE 1? Thx 🙂
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/
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
please read the content policy and terms of use. That will answer your question 🙂
Thanks, since it just specifies Dall-E without mentioning a number, shall I assume it's talking about DallE2 as well?
I believe so, yes
Thanks, very helpful. By the way - do you know if openAI still grants new access to DallE1 or is all new access granted just to DallE2?
Dall-E 1 was never released outside OAI. We can only access DE-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/
why am I using up credits to fine tune an image that is not what I want
everytime you generate something, you use a credit
I get that its just not a good business model since its taken me all my trial credits and still have not been able to generate the image I want. I uploaded an existing image and want to add an object to it took like 60 attmepts
#❌┃prompt-help and/or #📚┃resources might have helped. Also, you can always leave feedback in #placeholder-channel and vote on what you'd like OAI to do in the future in #1021688737572327444
Thanks will look into it.
Hi, Is there a way to generate an image that is larger than 1024 x 1024?
The outpainting feature allows you to stitch together many generations to create an image lager than 1024x1024px.
Take a look into the #📚┃resources channel for a full video tutorial :)
okay thanks.
I patched reddit but it isn't opening
How long does it take to get my forgotten password email?
I am trying to re-draw some succulents by uploading real photos. Dall-e is missing a lot of important details recognizable by succulent lovers. it is obvious that network has insufficient experience in this domain. Is there a way to add a batch of succulent photos in next AI train cycle for better succulent understanding? I've could create a dataset for you.
• 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.
Guys, I've contacted support, not even them reply to meee :/
It's been over two days I sent the message :/
Can the AI draw humans?
Can we get multiple aspect ratios with the same image? Say I want 4/5 and 3/4 same time.
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
@scarlet birch ⬆️
@manic umbra you can do that using outpainting
Thank you! Is it the same thing with adding generation frame?
Yes.
Sorry, what are you asking about?
I patched reddit using revanced manager and after installation when I try to open the app it isn't opening
Hello. Is there anyway to make sure that the produced image is not cropped (cut)?
Sir, this a OpenAI Discord
For DALL-E 2
Oh no
Sorry, they had a similar channel.
I sent it here by mistake 😅
I feel DALL-E is a bit unsatisfactory
You seem to be saying this in multiple channels. Did you have a question?
why?
DALLE is helping me a lot for my photography
I always get low quality
#📚┃resources and #❌┃prompt-help might help!
No
don't be a fudder my friend. You can use another AI to increase resolution
Read through the #📚┃resources if you want better images, or ask in #❌┃prompt-help
Thank you, I've already done that, but they didn't reply to me... Although, I've tried logging in with google and it worked ;D Thanks btw
Does anyone know how to use Dall-e 2 to interpolate between two uploaded images?
are .odt files not allowed in the #ai-discussions channel?
can someone find me who created this? or what the prompt for this
Given the signature on the man's shoulder cuff I'm fairly certain it's handmade, in which case the artist who presumably made it is clearly labelled.
can someone help me to make a Pochita Cat?
how can I get a cat similar to this dong from Chainsaw Man?
I am reasonably certain that almost none of the images DALLE was trained on contained a cat with a chainsaw through its head (and I would imagine DALLE would be quick to flag any attempts to generate "a cat with a chainsaw cutting its head" or the like as policy violations), so you honestly will probably struggle quite a bit in doing this. Your best bet would probably be to generate the cartoon cat first, and then attempt to inpaint the chainsaw over its head.
i've seen a cat like that on twitter so i wanted to create mine
that cat is very cute lol, how can you ban this?
basically chainsaw man is the most epic anime in human history, dalle must embrace it
Is there a way to add a team member to the DALLE Beta, or do they need to sign up through the waitlist?
One would normally need to sign up through the waitlist, but OpenAI previously announced that they had cancelled their waitlist, meaning they should be able to sign up directly.
are .odt files not allowed in #ai-discussions ?
Why does the AI stink? I did the prompt "a horse with wheels instead of legs pulling a cart with legs instead of wheels" and I just got a picture of a horse pulling a cart.
Awesome, thanks. Didn’t realize that
Hiya, DE-2 is very strong at creating coherent images. This means that more conceptual, abstract and surreal images require a lot more prompt engineering and experimentation.
Stating that the AI stinks because of a single prompt idea not working well is a bit like saying a hammer sucks at putting a screw through a piece of metal.
How to expand generation frame
If you want to expand in one direction while keeping the image resolution the same (at the cost of more credits): https://labs.openai.com/editor
If you want to expand in all directions while shrinking the image resolution (fewer credits, but smaller resultant image): !pad [#] in #dall-e-bot, and then upload the result as your image to edit.
In your case (continuing the conversation from #❌┃prompt-help) since you already have the head and just want to generate the rest of the body, I'd recommend the former option. Upload the image, perhaps erase the bottom part of the image if it fades out or the like, and then generate with whatever prompt you'd like. Make sure to overlap the frames somewhat so DALLE knows what characteristics and general art style to base the outpaint off of.
Just asking out of curiosity, but what was up with the sudden Christmas topic change in #🎃┃promptober ?
it's what Lucas and OpenAI chose! Parts of the world are getting colder this time of year 🙂
Funny enough, it was actually snowing earlier for me lol
that's crazy!
yeah, canada is the place to go if you ever wanna experience snow from like October all the way to like April - May
It’s a cold place indeed 😭
it was only 85 degrees today where I am 😎
👁️👄👁️
Hello, I have tested the edit function of Dall-E, specifically trying to replace an element with another, for example a door featuring a symbol on it with a door featuring a different symbol.
It would seem however that whenever I use the eraser tool Dall-E just removes these elements from the scene painting what it thinks is behind the door for example.
Is there anyway to replace elements with other elements rather than removing them?
hi, since the generations aren't layered as smart objects, erasing objects separately isn't possible, however if you want to change let's say a car with a bike, you have to delete a considerable amount of space so the new object fits in perfectly, otherwise the system won't have enough space to make a new object, hence will complete the generation with whatever it thinks is coherent and right 🙂
does dalle support using sentences/words
How can I make a panorama images like these:
when ever try rendering Dall-e comes with 1:1 Ratio image only
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.
what do you mean exactly? obviously you need verbal prompt to make a generation
Not sure I understand: you mean you can replace one door but not multiple doors?
How would you go about doing that?
you can replace anything you wish, you just have to make enough space so that the new object will fit
mb, i mean words/sentences in the generatioon
In my tests I had a few doors and a few floating tiles, I erased them in the edit tool and then described the doors and the tiles the result was that all doors and tiles were removed completely
In another example I erased an arcade and described an arcade with different characteristics, the result was that the arcade was removed
So far all erasing results in actual removal from the image with inpainting of whatever the ai thinks is behind
to problem might be either you're not erasing enough space or leaving the original prompt while describing the new generation
There was no original prompt, it was my own image , regarding the space i just erased the doors and the tiles. Would posting some examples help get a more precise assessment of the issue?
Also: when describing the elements to change what does the ai take into account?
- image around the erased areas
- image within the erased area
And does the prompt need to include the context or just what needs to be placed in the erased areas?
ai always tries to generate coherent images, so if you for example erase mouse and ask to generate an elephant, you won't get it right. regarding the description, do it one by one, and simply describe what you want. e.g. a red door. a black car and etc, you don't have to specify image within or stuff like that
Thank you, I'll try that
What is generally regarded as a good % to shrink down an image fot outpainting purposes in order to mantain coherence in style? 30-40%?
depends what you have in your mind
Let's say desktop size, but for the purpose of the outpaint the important part is to mantain the style. The generation in question is a plane, but I'd rather dictate the magnitude of things surrounding it according to the amount of shrinking I can do on the single original 1024x1024 square rather than the opposite
So I was wondering if there had been some findings like "oh if you shrink it lower than 20% you'll generally lose the style entirely"
the only general 'rule', without using any verbal commands, is to stay as close to your original image as possible to maintain the style.
As usual thanks for the huge help Ene! I'd say it's turning out pretty well 🙂 (I'll share when it's done)
quick question, I made a piece inspired by dialogue around AI art and used the dalle watermark with the multiple different colored squares. would I be able to sell that piece? I'm assuming it would fall under the same general licensing rules dalle usually has but I wanted to double check
I can send the piece as well to clarify, it has the watermark in the usual spot but I added the watermark into the art piece itself too
hey can i delete my creations?
That is not a feature at this time. If you have a concern about a particular creation, please contact support@openai.com
ok, can i delete my account though?
Anyone would like to answer this?
Contact support@openai.com about that
K
There are two ways to contact our support team:
• If you already have an account, simply login at https://openai.com/api/ and use the "Help" button to start a conversation.
• If you don't have an account or can't login, you can still reach us by selecting the chat bubble icon in the bottom right of https://help.openai.com.
Is there a way to put some art in an transform it to a different art style? If not where should I go to do that?
Dall-E 2 doesn't currently support style transfer but you can suggest it in #placeholder-channel !
Thanks, is there another AI that does style transfer, I'm really curious to see how this cartoon art would look realistic
How do you use DALL E2?
type in what you want to generate a photo of at https://labs.openai.com/ . Make sure to follow the content policy though!
OpenAI Terms of Use - https://openai.com/api/policies/terms/
DALL∙E Content Policy - https://labs.openai.com/policies/content-policy
Additional DALL∙E Terms of Use - https://labs.openai.com/policies/terms
The AI works with an underlying language analysis technology that attempts to extract linguistic features from the input prompt. The language features are encoded in a way that allows for matching to tagged images in a subsequent step. The process of extracting language features is not the same as computer translation, in the sense that a set of input words may wind up as a collection of attributes, all of which become goals for the art generation process. Grammatical relationships are very important in the process.
Is there a Dall-e api I can work with?
And is there a pricing model, rather than the 500 free images
Hello. Please review #gpt-models for details.
[❓] — Is DALL·E available through an API?
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.
The current pricing model is 115 credits for $15. A credit is one generation (four images)
ohhh
Hi, is it possible to generate 2000*1500 ratio images?
Not in a single generation, but you can use the Outpainting Editor to extend an image to that size (at the cost of more credits, unfortunately).
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.
How do I remove all generated imaged from "My Collection" ?
You can't remove images from your collection at the moment, but you can remove images from your favorites
Is there a way to upload an image and ask dall-e to make a 3d render of it?
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.
Would this be based on an existing image you provide?
Making the whole image distorted can be done in a variety of Ps / illustrator type programs
You don't need AI generators for that
Adding details can be done through inpainting using the DE-2 editor.
You would upload the image, crop it, then choose edit.
From there you can apply a mask using the tools to cover areas you want DE-2 to generate
Type a prompt, and watch the magic happen.
Is there an API available?
• We are currently testing a DALL·E API with several customers and are excited to soon offer it more broadly to developers and businesses so they can build apps on this powerful system.
• We do not permit programmatic access via web scraping or data extraction, as outlined in our Terms of Use.
Hey, is there any way to trace back an image to a prompt for DallE? Someone in another server is trying to pass off clearly ai generated art for money and I’d like to try and make people aware of their behaviour
I would understand if not however as it would be fairly complicated
you can find prompts in your generation history, and it's in the image file if the file name hasn't been modified. You could also tell them that passing of AI Art as their own breaks the TOS they signed and contact support@openai.com to see if you can get the removed from the service.
Aha, I’ll check the image file name now, this could be quite easy for me to tell. As I said, I’m not too sure, it might not have been made with the app but if they have been using it then I wanna deal with it obv, thank you
where can i find this bot on the server? github .com/openai/dalle-discord-bot 🙂
please @ me in your response ❤️
@storm marlin I don't think this Discord bot is available to the public. I believe it's OpenAI's proprietary bot (I could be wrong though 🤷 )
ah okay crazy
@storm marlin this is an ongoing project. The code is in the project. The project is for helping users on the server.
Is there something specific you were looking to do?
I love this image EXCEPT FOR the fact that I cannot get the Ai to put a teacher in the image. I have wasted several credits trying to edit this imags!
When inpainting you often want to go simple
@ruby canopy what prompts have you tried to inpaint ?
@lucid ravine I tried leaving original prompt, but adding "teacher seated at desk" after deleting that section from the original.
I have also tried deleting the area where I want the teacher and putting only "teacher seated at desk" in prompt.
I will have to go see what other prompts I have used, but each time I just get the original seemingly unchanged.
@lucid ravine my success with editing existing images have had a very poor rate of success. Any good tutorials you can recommend?
I will ping some of the other prompt leads to see if someone has advice.
@lucid ravine Thank you!
Hello,
I have a question for generating interiors.
How can I have consistency in the interior design of the same room, seen from different angles?
Let’s say I am standing in a room and looking at the wall in front of me with one table in it. This would be the first image generation.
Now I would like to go to the right side of the table and generate the image for the same room with the same table, but looking at another wall and a bit of the first wall.
How can I achieve this?
Thanks
With inpainting it is often necessary to remove way more than you need to. This can be painful as you like a lot of the picture, however there is often something about some image that dall-e thinks is just mathematically perfect compared to the rest of the image and it just doesn't want to move away from it.
Getting the teacher to sit may also be difficult
@arctic topaz That actually makes sense to me because the edits that have given me the best results have been the ones where I have removed a large selection. I will give that a try. Thank you!
Good luck. Also, don't say "chair". Just say "teacher sitting"
Don't prompt anything else
Saying the word chair may make it again focus on a chair
@arctic topaz Thanks!
Let me know if that works. I'd like to share the advice to other users later if that proves to work.
@arctic topaz That method got me my teacher. Lost other aspects I liked, but I will keep working on it. Thank you for your help, removing more definitely helped.
@arctic topaz just used my last credit. 🙄
Have to go get another prepaid card before I can buy more... makes me control my spending... this is very addictive!
I will try that. I took out pretty much everything but the little girl and the base of the window near her; leaving the desk should help.
@arctic topaz that's great! Anxious to work on this more myself.
It's not a stable art. I think one of the issues here is that the prompt interprets the little girl as the teacher sometimes so it leaves her out.
@arctic topaz I love the three windows.
You may sometimes have to roll again with this technique
multiple figures are tough for dall-e. When possible try to hack them out of your efforts.
@arctic topaz I did a few image edits yesterday using just a head that I liked, so most of the canvas was blank. I like those results.
If you want to keep one of the subjects there, it may be good to describe the next subject as "next to" the other.
Again though, I don't recommend doing multiple subjects.
@arctic topaz the only thing in this image was the woman's head. Prompt was something like "digital art painting artist self portrait 1600's drawing room" (not exact -- would have to go back and find it).
@arctic topaz then I flipped the image of the head and added it to my working image and got this.
Can I generate text and how
If you want to generate any text: Use a prompt associated with a medium that uses whatever style of text you're aiming for (movie poster, magazine cover... maybe even just prompting for something with a caption might work, but I can't say I've ever tried that).
If you want to generate specific or legible text: Manually add it using an image editor. The chances of DALLE adding coherent text to an image is effectively zero, unless the text you're aiming for is as extremely common as "Hello" or "Happy Birthday" (and even then it's liable to misspell it at times).
does dalle 2 use seeds
hello, can the generation frame be larger than 1024 x 1024? i would like to print my picture in a larger size
You can expand your images beyond 1024x1024 using outpainting
https://openai.com/blog/dall-e-introducing-outpainting
https://help.openai.com/en/articles/6516417-dall-e-editor-guide
on the website, how do you just generate variations of an uploaded image? can't find an option to do so, need to edit first and use outpainting which kinda wastes a generation