#community-help
1 messages · Page 15 of 1
Thanks, that's a weird menu structure but I see it now 😅
no worries, have fun!
You can assume it does, but it is not shown in the interface.
How is this against policy?
I can't say cockatiel?
If so. If guessing cockatoo is banned too lol
Thought I posted an image in Promptober at around 12:40, but can't find it. Can't currently post in Promptober. ???
hi, apparently cockatiel, but cocktail for example is allowed. but e.g. Pissaro (the artist) isn't allowed.
Maybe the filter AI needs a bit of training 😂
I noticed "shooting" in any context is blocked too. Like "director shooting a movie"
how do i report a generation frame?
Go to “my collections” and select “All Generations” you should be able to see all the individual frames you’ve created there.
ok because "body" did something inappropriate
Hey
You can use the scientific names.
Hi! Did you have a question?
The AI is not engineered to work with text. A few common words may appear now and the, but if you need text add it later via Photoshop or another editing tool.
I tested a lot with dnd dragonborn today. i got a lot of false results (=Human faces). Should I flag them as wrong?
Or is this just unnessasairy work on my side?
@broken wedge seems like an issue that should be looked into, though. Something that works with context, not just "word bad, you no get picture"
Why didn’t this work?
Celebrities are filtered and restricted in text prompts.
Seems a bit excessive as DALL•E isn't advanced enough yet to make convincing human faces. You could easily tell a fake image from a real one
We appreciate the feedback - feel free to also leave comments via #placeholder-channel
The product is still in beta and they are still tuning the filters. Please drop a note to suggestions in #placeholder-channel 🙂
how can I get variations of images I upload, you used to be able to do it now I cant find the feature
water mark removal
Upload an image, select crop, and the generate variations button will appear.
Can I upload a photo of myself to make variations ?
I wish to make a halloween photo of myself 😎
yes
ah 🙂
does anyone know if it is possible to use the API to analyze an image?
• 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.
Click "Inspect" on the right-click menu for the image, and the tab should have a link to the raw image without the watermark you can download.
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.
Can someone reassure me that using dalles generations in a published work is an acceptable useage? Are there any stipulations to it?
Actually nevermind I saw the link above and it’s pretty clear. Thanks
Can I seed DALL-E with a collection of images by a specific painter, and then ask for an image in that style? With someone like Monet, there are lots of source images to work from… with (e.g.) “my brother-in-law”, not so much.
Not exactly. You can have D-E2 create a variation of any image which is allowed under the content filters
But it isn't learning/trained on that, it's only ever going to make similar results of that image while you're working with it and it won't be applied to text to image.
Is it possible to get credits with Paypal or is it only possible through visa card?
great question
Right now only with methods that stripe accepts. Sadly PayPal is not one.
Please do head over to #1021688737572327444 and vote for alternative methods of payments 🙂
Does anyone feel like the image quality and style for Dalle2 has changed over the past week or so?
Seems about the same to me
This was how the old generations look
these are the new generations, with the same prompt. I understand Dalle2 does not use seed numbers so Images will be different, but the style seems completely off. Even the design of the octopi looks wrong
How many times have you run the prompt
I'm not sure DALL E works the way you think it does
I could run it again
Tbh I don't know a whole lot about whats happening behind the scenes. But I do know that its not unusual to get vastly different results from the same prompt
In the past I would occasionally generate the same prompt a couple of times if I liked the results. The results were always pretty similar to each other stylistically. But the same prompts now do not even resemble the style of the older generations
it does seem odd. Most likely something has changed. As to what that is or why the change was made, I'm unsure
I tried the octopus prompt several times and got similar results to yours, and they do seem unusual
you may want to report those generations as not matching your description
The issue is that it's pretty much all of my prompts are slightly different than before. I'll send one more example sorry for all the spam
Prompt: "An oil painting of a melancholic ghost playing the guitar, in the style of Rembrandt"
I did three batches using the prompt of a melancholic ghost playing the guitar. All the results were relatively similar
Then these are the new ones using the same prompt. They are remarkably different
Yeah. Like I said, it does seem possible something has changed, but I dont have any information on it
ah no worries then. I was just hoping someone on this server would know what potentially changed. Also wanted to make sure I wasn't crazy about generations looking a little different now
@rain sky your series of comments made me curious... I reran dozens of my old prompts, many of them multiple times, and there's definitely a difference. These were prompts that I previously had run dozens of times and each run were consistent. As a whole, I think the changes have "refined" or made my images more detailed and still kept the coherence of the original prompts. In a few cases, the prompts completely broke style-wise, but I'm sure with slight tweaks I'd have better versions than the originals.
Old (original) Oktoberfest gen:
New (current) Oktoberfest gen:
I've noticed the new generations struggle with surreal and abstract concepts. Do you have any generations that are not as realistic that you could rerun?
I have less of those and some of mine are not great examples in trying for consistent outputs. Usually when I go abstract/surreal, it can be very abstract conceptually where even the old runs you could have significant different interpretations in each run. The ones I did try -- big output differences. Like not even in the same league in terms of both interpretation and style of the prompt. Most cases I preferred the new ones, but in one case it was completely derailed where I'd probably need to attempt to re-write the prompt from scratch.
It also struggles a lot more with distinct art styles. Let me show an example
the above images were the old version
then the new ones
"A painting of Superman in the style of Alex Grey" for both
Also when starting a prompt with "A painting depicting", it will now show the "Frame" of the painting in the generation
occasionally
I ran those same prompts. Eww, the new results are not pleasant. I tried tweaking your prompt since I never was a fan of "in the style of" wording. I've always done "painting by", "art by", or just "by <artist>". Still subpar results. I have noticed on a few of the re-rolls on my own prompts the art got "flat" or less visually interesting. I will need to dig in and see if it's something with the artist or other element of the prompt. The two things that I dislike about the current application of Alex Grey is that color patterns are less interesting, but also it lost all the vibrancy of the original gens.
Please keep me updated on what you find. I noticed Alex Grey was much different. Same with Picasso which hurts my soul
two weeks ago - "A painting of the Batman, in the style of Edvard Munch"
versus: the new generations with the same prompt 😭
I'm in Spain without the S
I have a prompt where the ai did something I didn't think of and I don't know how to fix it
the prompt was to have a tree made of water with gold leaves
To be honest, you would be better off trying to generate a new image than attempting to clean up the existing one. Keep in mind that DALLE can only generate what it has "learned" from a database, and thus it will likely struggle with abstract and obscure concepts such as a gold-leaved tree made of water. (Your best bet is likely to generate the tree made entirely of water first and then inpaint the gold leaves on top, but that will also cost 2 credits instead of just one.)
oh okay
I haven't played with Picasso, but I ran some tests with Vincent van Gogh since he's covered in the DICTION·AI·RI. What I feel like I've learned with a few experiments is that results can drastically vary depending on how the prompt is written. Saying "in the style of" or "art by" or "painting by" or "oil on canvas by" can yield very different results. The best results seem to be "art by" or "<specific medium> by". I could see using "in the style of" in select cases, but it seems to lead to poor results. In some of my prompts, I use to just say "by <artist>". On the limited testing, it's not as good anymore.
Here's a better Edvard Munch that should give less Spain without the S. It still doesn't have the old feel, but it's a lot better than the outputs with the way your prompt was previously written. In this case, I went with "art by Edvard Munch".
I've had no luck with Alex Grey and your Superman. On other prompt subjects, DALLE got better if I included movements in the prompt like "visionary art" along with Alex Grey, but not happy with any of the results I tried with Superman.
These are certainly a bit better. What I miss about the old ones are that they are truly in the style of Munch, who created "The Scream." It looks like Batman is screaming, or at the very least Batman seems to be extremely uncomfortable. The newer generations make sense, but are they truly in Edvard Munch's style? Each image largely looks to be painted in a different style than the next.
I also highly recommend trying out Picasso
It feels like DALLE might be pulling in Munch's other works more losing that cool "The Scream" vibe.
I actually had a few generations in the past that specified "The Scream" as a part of the prompt. I used the prompt: A painting of the Joker, in the style of Edvard Munch’s “The Scream”. These generations were from a few weeks ago.
Then these are from today. They get the idea right, but in some way most of the images are flawed. For instance, the first generated image on the first batch looks like someone had taken a picture of the canvas from a distance. The same issue occurs with the third image from the second batch. Another issue is that lately, the background of the painting won't extend to the edge of the image. The 4th image of the second batch has a dark blue background, but the left and right edges of the painting are tan and look unfinished.
@unreal gorge Maybe they are trying to update their API and it's experiencing some growing pains before it gets better? I also thought that there could a possibility of some copyright issues involved in using a specific artist's name or style, that forced OpenAI to change something with the way Dalle2 is trained or the way it functions in general? I'm pretty much grasping at straws at this point
Even the recommended prompts (Surprise Me option) come out looking kinda strange
This one used to come up with the coolest results 😕
Is there a way to organize 'My collection' into folders?
not at this time
reload couple of times
still giving me that error even though i reloaded 5 times
getting the same error
working fine here
is the image editing bugs or is it mine? sa me: Something went wrong, please try again in a moment.
(Don't worry, no credits were used by this request)

Working fine on my end. Try reloading the page or use another browser.
well the error only shows up when I try to inpaint/outpaint an image..
this
Hey, someone know how to fix this error?
omething went wrong, please try again in a moment.
(Don't worry, no credits were used by this request)
Ah, looks like I’m getting that error as well when inpainting.
It does not work
any way to contact the developers?
Pretty sure they’re alerted of this. Just need to give them some time to get it sorted out.
ok
I came here to check, I suppose I'm not the only one, whenever I try generate variations of an image I get the same error as you guys.
I am getting them too.
I have the issue as well
Happens to me when I try to edit an image
Yes, it’s having issues at the moment
ok
Hello guys! I have one question. Why my iamges is incomplite?
They do sometimes get cut off. Check out #📚┃resources and ask over in #❌┃prompt-help - you may need to use outpainting, which is having issues right now.
No this will cause my phone to refresh the page.
Ah clicking the edge corner works! I've been using this over a week and never figured that out... Thanks for the assistance.
I do think they need a close button though, I'm sure many have the same experience.
A world of difference that makes for the mobile experience. Thanks again!
Mobile experience need to be improved, first I need to learn how the code works 😂
Yeah, you have to carefully find a spot to tap
Ok @vocal forum just informed me I don’t know anything. Lol. Swipe from left to right.
All this time… Try that lol.
Cc: @indigo rune we probably still could use a close button. Lol
**you can swipe ⁉️ **
I think Major was right about you
I totally knew you could swipe from left to right. Mhm. 100%. Not just learning it now.
Has the problem been solved? With the error message?
Swipe left/right also has no effect when a collection item is open on mobile. At least not in my case.
This is currently on display in the DALL-E website. They’re working on it 🙂
Ah thank you
It works on iPhone for me. What are you using?
Is the variation broken atm?
Look up 4 messages 🙂
Oh yea, ic thx, I'm using my phone in split mode, so I couldn't see the above messages well
It works on android 12
If I generate a new image, and click to view it. Yes I can slide left to right to view other variants.
If I click 'My Collection' and the select an image to view from here; then the left/right swipe does not work.
It's all day that I am trying to edit a picture (generated itself with DALL-E) and I always get Something went wrong. Please try again in a few minutes. If the issue persists after trying again, please contact us through our help center at help.openai.com.
Thanks!
Weird. That seems to work here. Maybe it’s the version 😦
Still need a close button then. Sorry :/
Hello there, I accidentaly paid for credits by card from a different company as is listed on the invoice. I tried to solve with via chatbot, unfortunately, no one contacted me. Could you please help with this? All I need is to change the company ID on the invoice. Thanks in advance
You're almost certainly going to have to email the OpenAI Support team (support@openai.com) to get that fixed, unfortunately.
Thank You @zinc marsh
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 do 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.
Hi guys,
How would I look if I had a hair transplant? To see this, I upload photos, mark them and type commands like "he has black hair" or "add black hair". but the results are bad.
How would I look if I had hair? What should I write for the most accurate result?
And should I just mark the hair or the whole face?
Hey 👋 how does one prompt for text? For example “A wall with the words “go away” spray painted in red.”
Or “a can of spray paint labeled “johnson” as the brand name”
Hello , new user here, hello everyone!
I have a question - is it allowed to use a generated dall-e Image to be used as a Logo for a non profit organisation ( support association ). So it would be printed on letters and such.
If you upload an image is it available by the engine for all future creations ?
yes
Didn't quite understand the question
I saw an option ( have not used it) to upload your own image to dalle
DALLE is absolutely horrific at text, so you're honestly better off just generating a blank wall/spray can first and then adding the text in an image editor.
I want dalle to draw like a children's painting with crayons is there a phrase that is suitable?
Like crayon painting? Not sure xD
There was actually a daily theme about crayons only a week ago (#daily-theme message); you can scroll through the posts there and look at the file names of ones you like to get an idea of good prompt structures for that medium.
Discord is the easiest way to communicate over voice, video, and text. Chat, hang out, and stay close with your friends and communities.
Hi, I know that free credits do not roll over to the next month, but do the 5 daily credits for participating in promptober roll over or are deleted? I have "You have 0 free credits to use until 29 October,
and ?? paid credits remaining." These "paid credits" are the free promptober credits. Thanks for help.
The Promptober credits acts as paid credits and will expire after a year.
Thanks Major, that is excellent, cheers for the info.
@tacit coral - #🌄┃general-art message
As far as I know, the model has been static and no training data updates were done. DALL-E have been always able to do realistic faces. The community’s approach in wording the prompts improved that’s why you’re seeing a significant improvement. Check the #📚┃resources tab, there’s a guide on how to prompt realistic faces.
@verbal crag #openai-chatter message I think something has definitely changed, at least a little bit.
This was the beginning of the discussion: #community-help message
@verbal crag your thoughts on this? ^^^
I unfortunately can't comment on this as we volunteers do not know what's under the hood of DALL∙E. But what I can say is that there's been no announcement from any OpenAI staff about any changes to the model.
ah okay. I just thought it was strange. Hopefully whatever is going on gets resolved soon. Most of my favorite artists' styles took a hit in quality 😕
We can get credits from participating promptober ?
You would have had to have signed up for them by September 30th, unfortunately.
I'd like to try dalle2. I'm using https://labs.openai.com/ and wondering if this is dalle or dalle2. If this is dalle, where can I find page for dalle2?
You are using DallE 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/
@vocal forumThanks. I was confused as I did not see anything written about ver2 on the page I was using.
can i have the server invite link
👍
I used outpainting to generate a large image and now I have all the square parts of the image that I generated into my gallery but not the large one I ended with, how can I get access to the large one and download it ?
you need to download it after you finish generating, there's download icon on top right corner. you won't be able to see it in the collection
so if I closed my page before downloading I won't be able to get it back ?
You’ll have to piece it together manually using krita, clip studio, photoshop etc
Is the a way to download the images without the watermark at the bottom ? Because I don't want to have 3 of them onto the final picture
If you right click and inspect you should see the link to the image that doesn’t have the watermark
thanks
When I’m doing a big outpainting I download every step, just in case. Sorry you got bitten by this. It’s beta still.
Hi, how can i make it understand a certain word, for exemple i want my pseudo "PowerSmite" to be stylized ?
DALL·E 2 is not engineered to properly generate text. While it may get a few simple words right now and then, that is the exception rather than the rule. If you need to add text to your image, consider using an external image-editing software.
@rain sky Moxi from the OpenAI team said yesterday in #openai-chatter that the feedback you and I have provided here will be brought up with the team.
I want to try redesigning my pfp to still be a floating island, but different, anyone have any tips on what to put for the prompt, I’ve tried a few times and I have not really liked any of the results
This is a great question for #❌┃prompt-help
I'll ping you there.
Are there multiple DALL-E Discord servers? I’m trying to connect with a friend, but he’s not here.
I suppose someone could have theoretically created an unofficial one, but this is (to my knowledge) the only one affiliated with OpenAI.
How large can you output an image? Most of the images I've created are small. But some I really like and would like to make them background wallpaper on my computer. (ie; 2560x1080 (ish)
I've outpainted images beyond 4k resolution. I think there is a theoretical limit due to the cache and RAM allocated to your browser (I've heard some people mention browser crashes), but I don't think you'll run into any issues unless you just keep expanding
@clever cliff forgot to reply 😅
do you have to input something extra to encourage the size?
Every generation frame is locked to 1024 x 1024 pixels. You'll have to outpaint multiple images. There are guides in #📚┃resources
As @zinc marsh already mentioned, this is the only Dall·E 2 Discord server that is directly affiliated with OpenAI,
so if you can't find your friend here, they are probably in one of the unofficial servers. :)
ahh... I got mentions of this. In other words... you download what you get and fill in the spaces yourself though another program?
You can do it all using the Dall-E2 editor in the website! I recommend looking up a guide as it's a little tricky to describe without an example to demonstrate
For instance, this is 3 generation frames (all slightly overlapped to keep cohesion). The current resolution is 2944 x 1024. If I added a row of generations on top, it would be around 2944 x 2000 pixels which starts to approach a 16 : 9 aspect ratio (cropping down would achieve it)
just catching how this works. very cool. I'll give it a try. thank for pointing that out
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/
Ah okay thanks. Hadn't seen those messages.
Would asking for “seductive ceiling fan, manga” get me a warning?
Just got here- Warning?
Like, sensitive content or whatever.
DALL•E doesn’t let you generate NSFW content for obvious reasons
I'm not sure but I imagine so 🤔
The artist I used to create cute dog paintings doesn't work anymore :(
It definitely did change
I made this for the DALLE DictionAIri a while back
exact same prompt now
@indigo rune I know you were asking about examples of this earlier ⬆️
the artist (John William Waterhouse) used to work consistently to make a simialr style
-_-, I'd put the imgur if I could
now it's a lot more realistic and inconsistent with the painting aspect
I have a question, it's about language. I speak spanish, and I got an idea, how to traduce prompt in my language and in another ones, if there's a properly traduction, of course. If I can say it, I guess is 'sugerencia' in my case. What about it?
I think prompts in other languages already work, if that's what you're asking (or at least I dimly remember seeing some before). Otherwise if you'd like to make a suggestion to OpenAI, there's actually a Google Forms you can send them in in #placeholder-channel.
Can you tell the ai that it should dream and draw what it thinks?
nvm can it create NFTs?
read through the ToS & 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
Yes.
Ok 👌
Hmm So does anyone know how i can tell the AI to create an NFT? Should I just name the "NFT"
I'm not sure how much help you will get for NFT's here, try looking in #📚┃resources or asking in #❌┃prompt-help
you can turn anything into nft if that's what you're asking
well you can always just think up neat little ideas on your own, if not maybe just make a prompt generator
yeah but what prompt could help to make an NFT?
I don't think you fully understand what NFT is 🙂
what would be the difference between an "NFT" and a normal image?
NFTs are just images that people list on blockchain marketplaces like OpenSea. I could list any one of my dalle generations as an NFT. Is it that you want to recreate the art style of a particular NFT? @spare shoal
Yeah new to nfts
Isn’t there a difference in formats?
Or am I just lost
There are no image format requirements for something to be an NFT
hmm ok i see
Ah ok so it doesnt matter which size ok
Thanks guys :) Much appreciated
Hello, I understand that I can create and sell art I generate using DALL-E, but does that mean my clients can use the art commercially? For example, they purchase an image from me and would like to put it on a t-shirt, mug, etc to sell. Is that allowed? Thank you so much for your time and help!
#community-help Hello, I understand that I can create and sell art I generate using DALL-E, but does that mean my clients can use the art commercially? For example, they purchase an image from me and would like to put it on a t-shirt, mug, etc to sell. Is that allowed? Thank you so much for your time and help!
Hi, please refer to the terms linked below. For further questions about the terms, you may contact support@openai.com
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 expand the own art that I made?
If you’re trying to outpainting your image, upload it to labs.OpenAI.com/editor
how do i increase the size of my generation frame?
Unfortunately generation frames are restricted to 1024x1024
but mine is lower than that
Shouldn’t be
Are you using dall-e via openai?
there’s a guide to outpainting in #📚┃resources that might be helpful
Anyone else getting this issue when outpainting? "Something went wrong. Please try again in a few minutes. If the issue persists after trying again, please contact us through our help center at help.openai.com"
same
connection issue
looks like it logged everyone out
@indigo rune fyi
Should be ok now, try again?
Hello. Is it possible to imbed the Dalle ai search and display section into a 3rd party website?
Would you mind giving the full text of the prompt you used for this test? Every word in the prompt affects the results. I would like to try a parallel test.
Regarding other languages, I think the issue is that the language model used to parse the prompt for language features was designed for English, but training data exists that is tagged in various languages and writing systems.
॰金平糖॰ –– This is a type of candy, using Chinese characters to name it.
Following up on this question. Would love to have this answered
Is the Dalle the same as Dalle 2? I joined because of the recent hype around the second variant and I’m not sure what the difference is if there is one
DALLE-1 was never released to the public, so the DALLE you are seeing everyone refer to is indeed DALLE-2.
Thanks, I appreciate the help
Anyone else getting errors messages? Each run is "Something went wrong. Please try again in a few minutes. If the issue persists after trying again, please contact us through our help center at help.openai.com."
Yes, I have been getting them repeatedly
It takes like 5 tries for it to work
- If you're talking about the DALLE generation input bar and results: No, the API for DALLE hasn't been released yet, so you can unfortunately only use it on the website itself.
- If you're talking about embedding users' collections: I suppose you could manually download your own generations (not others unless they give consent) and add them to your website (you just need to indicate that they were AI-generated by DALLE), but I don't think any system exists to embed the images directly from the website.
I'm getting it as well now, seems to be an actual outage, despite the status page
seems to work again now
How to get free tries?
You get fifteen free credits per month (distributed each month of the day you signed up); otherwise, outside of special occasions like Promptober, you unfortunately have to buy them.
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Can you use images that you generated using another image for promptober?
This isn't allowed. All promptober images must be 100% Dall-E 2
Got it, thank you
Is there a way to make 2 images seamlessly merge?
Yes. Using the Editor.
Add two images in the editor, erase some of the borders and leave some space between them, and DALL E will do its best to fill it in
That didn't really work..
It made another plane below the plane I already put.
Here's what the original looks ike.
Not sure, I would leave more of a border around the plane maybe
so it can distinguish where the plane is supposed to stop and the sky should start
am i doing something wrong here? i tried multiple variations of that prompt and it started saying i might get suspended if i keep going 😞
I'm guessing "headed" is setting off the filter. (Either that or "glistening white," but I somewhat doubt the latter possibility.)
well from the screen i already outpainted once and it worked so idk why anything further doesn't work
There's currently no way to delete generations from your collection, but all generations are private unless you actively click the Share button on them, if that helps.
outpainting is spoiling the quality of an image, what to do?
It is removing some detailing from the costume, etc....
I even gave prompts like hyper detailed, intricately detailed, 8k, hyper realistic, photorealistic, but no use
can someone suggest something?
original image
outpainted (erased the golden S)
losing quality and looks rlly bad, no detailing at all on the costume
if someone could help, I'll really be grateful
That is expected as you’re using two different AIs.
but i've never lost quality before
or detailing
feels like since Dall E 2 released for everyone, their quality has come down slightly
Again, you’re using two different AIs, you’ll get two different outputs.
despite using different AIs
The details you’re going for, based from my experience with DE2, will be impossible to get with the current model.
You’ll have more chance if you’re outpainting images without such details.
You may move this discussion to #ai-discussions as it involves another AI.
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.
Thank you
Thanks so much for your reply
Hello! Is there an Open AI staff member here by chance?
People with yellow names are openai staff.
and they're 99% west coast USA timezone
Oh yeah! Makes sense! I’m trying to get in touch with like a community liaison or a similar role
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 anyone I could ask if my image is "Safe For Work"?
When in doubt, it's better not to post
Does it break any of the server rules? Is it G rated, and does it follow 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
I just read the content policy again. So my image shows two people touching with their heads in a loving manner. I don't think its sexual. But to be sure I d like to show it to someone.\
I would say as long as they're fully clothed, that's okay. During the #🎃┃promptober theme of cozy a couple weels ago, we had a lot of images of couples under blankets that were G rated.
Will Dall-e2 have an android app like dall-e mini/caraiyon?
Dall-E Mini isn't associated with OpenAI. Currently the only way to use Dall-E 2 is on the website
Will in future have android app?
I'm not sure. You could suggest it in #placeholder-channel though!
Is there a way to use Dalle through discord using cmds?
How do you get featured on dalle2's Instagram?
No, just on labs.openai.com
Just tag them in your IG
Thanks 👍
can i see all the published art somewhere?
as in gallery?
Yes
unless the authors themselves publish their generations, all of them stay in a private collection, hence there are no gallery for all generations
Hey guys, I'm having trouble getting Voxel art to work. Does anyone have some advice?
hi, what kind of trouble?
I would like to create something that looks like this art style
but im getting this
do you perhaps have any advice on prompts?
Maybe something akin to "3d render, concept art" or the like?
(Or maybe more accurately "3d model" instead of "3d render")
try isometric voxel robot
This is a robot from mech warriors. When identifying art it's important to try to figure out: style, franchise, and artist.
hmm, i feel like this is a step back :/
If you are trying to identify an art style, please start by looking through the style guides listed in #📚┃resources. Identifying styles by eye is very difficult, and often leads to misfires. You can learn more about the challenges with style identification and application in Prompt Tip #17: How to identify an art style.
I already looked in #📚┃resources , unfortunatly I couldn't find anything about voxel art 😦
How do you get the dark screen with grid of dots that's shown in the Outpainting guide in #📚┃resources? I've been calling up the MacOS 'show grid' to measure position of my outpainting frame, but having the grid permanently would be very handy. Thanks in advance!
I'm getting incredibly worse images when using the same prompts I used a week ago:
Old (Top) and New (Bot)
@indigo rune another example ⬆️
Heres another from a different prompt
It is difficult to evaluate these without seeing the full prompt because every word affects the result. Are you willing to share the prompt so I can test it?
Yeah mate thanks,
1st: A beautiful Albanian Town on a stone bridge, showing the town. Blossoming Apple Trees. Isometric view, Photorealistic digital painting, highly detailed, in the style of a blue and white mosaic, Arabesque, Intricate Patterns, Ornate, 8K
2nd: A beautiful Albanian Classical University, showing the University. Photorealistic digital painting, highly detailed, in the style of a blue and gold mosaic, Arabesque, Intricate Patterns, Ornate, 8K
Thank you.
First reaction is that both prompts are quite complex, including cues that will inherently cause lack of replicable results. I will try a test.
This was just produced, of course. The prompt includes a complex compound subject. My eye is on the way the water is rendered. The third image shows much smoother rendering that the stipple effect in the others.
One way of checking for style stability is to notice inter-image variation in the set. When we look at just one we cannot see what variation may have existed in the set. We have been trying to research if there has been a change in this sort of consistency.
low res but here were my four
Thank you. Going forward, keep an eye on the entire set. I will try another test.
So did something change with dall-e? I'm not mad at improvements but I wish I knew so I could make as much images with the particular style
I've generated thousands of images and never got that poor quality until last week
When testing I sometimes omit stying keywords that do not actually specify a style. Here, those include "Intricate Patterns, Ornate, 8K". This is a simplification test for "A beautiful Albanian Town on a stone bridge, showing the town. Blossoming Apple Trees. Isometric view, Photorealistic digital painting, highly detailed, in the style of a blue and white mosaic, Arabesque." Adding words like "ornate" inherently cause less consistency because they pull in training data that can be ornate in uncontrolled ways.
I see. Though why are so many like me noticing it this week? Did they extend the data set so now there are more "ornate" things to pick from for instance.
The only specific changes that we have been told of relate to changes to the content filter to reduce false positives. That can have indirect on image results if the language model is involved. We also understand that some back-end changes may be happening but we do not have details.
As far as we know, the model was not retrained.
We get reports of changes in model behavior from time to time, and there have been several reports in the past week, but we have no confirmation there was something done on the back end that would explain it. I have been watching the comments, but most cases don't provide enough comparative information.
I have tested myself and can't confirm any changes, but that does not mean that others are not seeing something.
If you have any older prompts that have a single subject (not compound) with fewer style cues, those are the types of prompts that are easier to interpret. The issue is that the model produces normal variation all the time that people tend not to notice.
All my prompts are at basically the same level of detail as those. I appreciate the feedback though.
What I am seeing in the simplified test I just did is that the stippling effect in the water is similar to the previous samples. That type of surface rendering effect is across the entire image set. It causes a perceived loss of detail for most viewers.
I'll upload more, one I remember is this: #architecture message
All the same prompt: A beautiful [Insert Country] Castle Town. Blossoming Trees, art nouveau style. Vines on buildings. Isometric view, Photorealistic digital painting, highly detailed, in the style of a magenta and white mosaic, Arabesque, Intricate Patterns, Ornate, 8K.
We have discussed these reports of changed behavior with the engineers and they are looking into them, but we have no clear answer yet. I appreciate your sharing this sample with me.
All those images were quite similar in style, even though the subject changed. And while I picked among the best, I don't remember the shiny weird look that shows up now.
Another prompt, I used a lot was.
"A beautiful Albanian [Insert Building]. People in traditional Albanian clothes are walking around. Wisteria. Summer. Amazing Unique Turquoise sky. Isometric view, Photorealistic digital painting, highly detailed, painted by Karl Gustav Rodde, intricate details, 8K."
And one of the main thing that's now changed is the sky. As well as the actual art style.
from:
to:
If we learn more about this we will post about it. No need to post more samples here, however, as this channel is best used for simple issues. I appreciate you raising the issue.
It's a different example without ornate
The prompt is inherently too complex to analyze stability.
Prompt Tip 5 (Get the main subject first) explains the coherence problem as prompts become more complex. https://discord.com/channels/974519864045756446/1021957512917954611
It doesn't because I've changed the prompt subject hundreds of times (literally). I looked up three lists of all building types and swapped them out, and got the same style for all of them. Until last week.
Thanks for raising the issue. We have been passing these reports on to the engineers for their review.
Hello, I have a question. I would like to know if there is a discord bot on the server that is like dall-e Where you say something and it is believed directly with the bot.
Why is it that when I try to get a realistic photo result some parts look painted?
you can use "a photograph" instead of "a realistic photo"
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.
I tried it again but it still looks like some parts were painted, is there any way to fix this?
how can I link my collection to show a friend?
hi, this is a feature we currently do not offer. sorry!
did you open an actual image? that's funny, i might email support and report it as a bug if that's the case
not quite, sometimes you just have to be really lucky! the model, although amazing, is not perfect in all cases.
you can't link full collections but you can share individually 
Can I use dalle Images in a game I am working on?
It's a school project (I already submitted) with nothing commercial in it, Adding 2D Generated arts can imrove it by a lot🙂
Read the TOS below on content use
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
Ok, but one more question: 3, why does it tell me that I can only use the bot for 7 more attempts and they renew it on day 1? We have to pay? Every time story is that done? How many times does he leave you each time?
Alright, Thank you!
15 Images per month + 50 on the first one.
It actually makes sense,
Imagine the bot needs to operate and give Images to hundreds of users every minute, and you are very limited by the amount of Ram or Cpu you have.
From my experience it takes a very long time for the averge PC to generate 4 Of these images, so getting those in less than 60 seconds is absolutely incredible.
Now you need to do that for many people around the world and have the bot work, the buisness model here makes a lot of sense.
Now you of course got the ImagineBoat thing, which is another discord bot that does it, and while competition is always good, dalle is much more to the point and gives cleaner results 🙂
MidJourney*** is the other bot, but it didn't work well for me
• 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.
@lyric galleon
dang
Oh okey, tx 😊
Yes, I've tried it and it's not bad, the problem is that later you have to pay and you can't renew or anything, I think, so I like this bot better.
In order to keep the focus of the server on DALL∙E, generations that are not solely generated by DALL∙E should only be posted and discussed in the #ai-discussions channel. This includes outpainting other AI generations with DALL∙E.
Hey guys, quick question! I've been generating some objects to convert into 3D models, but a few of the images don't have the objects fully in the frame, they go out of the bounds of the image. Any way I can specify to the AI that I want the object to be in frame? Thanks in advance!
Other people have brought up this issue as well. You can either try to add modifiers like wide shot, 24mm, wide, etc. Or you can utilize #dall-e-bot with the !pad command and outpaint the image on the website
Kk, thanks!
I think it has something to do with uploading pictures with Firefox, I've seen this on another website once
But uploading things on other websites seem to work ok
"realistic" is term that is most used with painted looks where the final painted result should pull towards realism / realistic looks. "photorealistic" is another term that potentially takes it further. When these terms are used with photography/film style prompts, they are likely to introduce painted effects.
Yes, that goes against the TOS
Well that's a bummer, my friend and I created this account so the both of us can have access to everything.
You could each create your own account and then share the links with one another via a Google doc or sheet!
Contact support@openai.com
Will they help me get access to it
?
Since we will no longer be sharing the account
They can help you reset your password if you can't enter the account like you previously stated
I can't even enter the login screen
Hmmm... that's strange. The website is working fine for me. It may be a browser issue or connection issue for you
I feel as though I was IP banned or something
If you contact support@openai.com , they can help figure that out. We can't do anything else here sadly
Did you try with another browser?
yes
Same thing
even tried it on my phone and it doesn't load, so it's 100% IP related
Can everyone make an account on dall-e website?
Yes, the waitlist has been removed! 🥳
Nice!
What does it mean when downloading an image and the file name is just example.com.jpg instead of the prompt like normal?
Did it just give me a stock image because it couldn’t figure out the prompt?
Hmm, I've never seen that before. You clicked the download arrow, correct?
Yes, clicked the download button. Prompt was “We can go when we want to, Night is young and so am I, And we can dress real neat, From our hats to our feet, And surprise 'em with the victory cry, digital art” (lyrics from Safety Dance)
Had a couple of other today that did the same.
That's really bizarre. I'm not sure why it would be doing that 🤔
Results don’t match the prompt very well at all which it why I’m guess it gave up and gave me a stock image
how do I use /draw [prompt] ? it says I don't have permission in #🤖┃dallify-hof-archived
Currently, only people with the promptober HOF role and server boosters can access it. It will be open to everyone in ~6 hours
okay they shouldn' have @here the channel if it wasn't public 😅 thanks i'll be back later
We wanted to let everyone know about it 🙂
Does this use credits?
nope! 🙂
Thanks!
Anyone here know why "(Galileo) Galileo, (Galileo) Galileo, Galileo Figaro, magnifico, digital art" breaches the safety feature?
You have to wait before sending in new requests
It's a 1 minute cooldown on requests
this doesn't seem right
1 minute cooldown 🙂
okay, but there's a negative sign
I'm getting it too
I think it's just a dash. Don't mind it 🙂
Is open AI Dalle2 on discord using credits?
It does not use your credits
When the next poll in #1021688737572327444 channel will be available?
Wait quick question, Is it allowed to remix old paintings by adding new variations to it? I re-read the TOS and I couldnt spot anything disregarding that.
yes why not
Just double checking
hey! does it costs credits to generate in discord. please let me know
it doesn't use your credits
nice!
can i ask a question about getting hall of fame promptober role
Is there any way to outpaint without a prompt? I'd like Dall-E to continue the image how it thinks it should be continued without me having to influence it.
no dont bother, i figured it out already, thanks for offering help though!
Every generation needs a promotion except variations
So there's no way to do it?
You could always keep the prompt the same as the original generation
i'll try that, thank you!
this Dallify discord bot is awesome, does Dalle have any plans to move towards a subscription based model?
We haven’t been told anything about that yet.
ah ok, I see there was a poll that mentioned it last month.
wishful thinking
ill keep lurking and waiting for that subscription option to drop
How can i use the outpainting feature?
There are guides in #📚┃resources
You can post multiple!
Huh it removed the one I posted.
When does this one expire? I'm hoping I can post it after the 45 minute wait ends.
There's an hour cooldown in the channel, but you can post after the cooldown ends
Aww. They changed it 😦
I keep having issues when I try uploading. I get a message saying I am rate limited due to the slowmode and then my hour time limit resets.
you can post once the cooldown ends
That's the problem. The cooldown ended, I posted, it said i was rate limited, then the cooldown reset and the post didn't go through.
Try posting in other channels see if your posts go through
Strange. I don’t see any logs saying that your post has been removed in #🎃┃promptober
I'll try to pay close attention to the message if it happens again.
I can't post for another 30 minutes.
.
Hi
Does dall e use only public domain images?
We don't currently know if Dallify will show up again around here
Hey guys, I'm new here, I would like to know in which tab I generate the images? who can help me, thanks a lot!
Thank you very much! but wouldn't there be an option here within discord?
No there's not. There was a special event but there's currently no set plan we're aware of for generations in the discord.
yall does dalle know what lego is?
yes, especially minifigs.
can tend to do blocks/patterns which lack fidelity to standard sizes of blocks
Its not a bug
Instead of "in a transparent glowing cube" try "standing in a transparent glowing cube"
same
It's probably detecting it as the anime girl wearing a transparent cube
can dalle draw dune ornithopters
Hi! How can I delete images from my collection? thanks
You can't delete images from your collection at the moment. You can remove images from your favorites though
I’ve seen one generate something close to it, but that was weeks ago. So yes, it’s possible.
ah ok
Thanks Laze! Do you know deactivating account will delete the entire collection/images?
Contact support@openai.com about account/image deletions
Hi! Am I unable to get dall-e 2 if I have the original dall-e in my account?
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/
you're using Dall-E 2 🙂
Ah, thank you! I was confused.
I'm trying to edit a painting by William Adolphe Bouguereau, but it keeps hitting the content warning
The AI's probably recognizing/mistaking something inappropriate as being in the painting and thus is refusing to edit it.
I don't believe so
Can I post dall-e 2 generated images to Instagram/Twitter?
To my understanding, you can post them as long as A. you generated them (or the person who did explicitly gave consent), and B. you make it completely evident they were AI-generated by DALLE-2.
You can
There is an Instagram account which is making AI generated art
And its damn cool
Hey mods, First we get free 50 credits right?
Do they expire after a month?
Yes, the first 50 credits you receive--and the 15 free credits you'll receive each month afterwards--unfortunately expire a month after you receive them, generally on the same day of the month you first signed up on. If you ever end up buying credits, those expire after a full year instead.
sad
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Ok, thanks.
Which account? Send the username.
Will celebrity support be added in future
we don't have an update for that yet
I also want to become a prompt engineer, is there a course for that? Or anyone can get that role?
Do the same prompting tips also work for other similar projects like stable diffusion
regarding other ais you can use #ai-discussions channel to get and share more info, regarding prompting, depends what models are used in ai, im not familiar with sd hence I can't give you an exact answer
Okay
Only our community volunteers get that role. To become a volunteer you should have posted here a while helping others, have no warnings, and have a good understanding of Dall-e
You are welcome to read #📚┃resources and study #1021130377026351105 and #❌┃prompt-help
So, Any community volunteers can get that role
As long as they show they have a good understanding, yes. It’s not a class.
Okay
Hi there guys, thanks for having me.
I have a question in regards to the process of input.
I have been trying to create and image with text, but for some reason the text never comes out as written, can you give me insight here?
DALL·E 2 is not engineered to properly generate text. While it may get a few simple words right now and then, that is the exception rather than the rule. If you need to add text to your image, consider using an external image-editing software.
check the respond above👆
not really related to DALL E but how do we use the OpenAI Jukebox?
why do i keep getting paid credits added to my account?
i have never once paid for any credits
it says i have 147 paid credits
That's something you'll probably have to contact support@openai.com about
Is there any easy way to download all of my images?
If you signed up for Promptober back in September, anyone who did that receives 5 paid credits per day for the month of October. (Though that would admittedly only account for 125 of them--I'm not sure where the other 22 might have come from.)
ok
Even i want to know this
I think that there is a Google colab demo somewhere
Not 100% sure this is the right channel but Im not receiving a reset password email even though I’ve tried different capitalization and resending
contact support@openai.com about that please
I’m trying to remove the nose of an alien but the system just keeps changing the nose
How do I remove the nose?
through inpainting?
mask the nose, and try a much simpler prompt like "alien skin" ?
Yes, I masked the nose and wrote “remove nose” “no nose” and placed the instructions in the prompt
Using “alien skin” still gave it a nose
Hello! I bought paid credits some time ago but I don't understand why they keep increasing. for example this afternoon I was 95, now 110. But on my bank account there are no exits... how is it possible?
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
Hi there! It's possible you are part of a program like Promptober where your account is occasionally granted credits – our UI doesn't do a good job of distinguishing those from paid credits (we'll try to improve this). I'll reach out by DM to confirm this with you.
Likely a similar situation here – see the reply above this message in the channel!
Hey so how do I start outpainting? I wanna extend this generation more but idk where to start
- Move the generation frame to where you want the outpaint to go, making sure it partially overlaps the existing picture. DALLE can only "see" what is inside the 1024x1024 blue box when it tries to create a generation, so if you have the generation frame outside of the current picture you're going to receive a completely different picture instead.
- Type your prompt. If your original prompt had certain stylistic cues (e.g. "moody", "trending on Artstation," etc.), it sometimes helps to include those in your prompt here as well, but I'd recommend not going overboard. Oftentimes I've found simple is better, and complex prompts just get the AI confused.
- Click "Generate" on the right. You'll receive four different candidate generations that you can scroll between, and you can click "Accept" on the one you like to add it to the canvas (or "Cancel" to reject all of them). Regardless of what you choose, all four are saved in your Collection, so there's no risk of you losing anything (you can always manually stitch it back together later if you wanted).
- Rinse and repeat until you're done, at which point you can click the Download icon to download the finished product.
There's a video at the very bottom of the #📚┃resources channel. 
(Also, usually it's good to not describe in your prompt that which you already have generated, since otherwise the AI might get confused and try to generate another one--so it might be good to not put "person standing by the water" unless you want to have another person standing by the water in your outpaint.)
There's a small eraser tool you can use to erase the logo on the original image if you want (or you can go back to your collection and download the raw image from the Inspect tab).
Nice!
:D
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.
For outpainting prompt, do you describe the whole image, or just what is in the generation frame?
Just what you want in the generation frame. There are a couple guides in #📚┃resources that explain it in-depth!
Ok, thank you :)
That would be your starting point. There's a photography and realism guide in the #📚┃resources that covers how to build prompts which allow better, realistic results as well as get the right artistic effects.
Is it possible to delete specific images from your collection?
Photography guides are available in #📚┃resources
Not currently. You can contact support@openai.com about it
Why is autistic blacklisted? Is it something that the creators would be willing to reconsider?
You can provide feedback about this via the #placeholder-channel channel 
hi. Well I was hoping to be AI/ME artist, that uses Fresco/PS to create Print art for the house.
Turns out upsizing from 72 DPI to 300DPI is not so easy... even with PS. (also need to now get used to much larger art workspace)
Can someone direct me to the proper resources please for upsizing ideas from 72 to 300dpi/ppi?
hi, regarding resizing your canvas properties you can do that via outpainting tool, regarding dpi, we don't have a specific resource here since it's not directly connected to dalle2
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 use picwish for photo quality related things
thank you, will google
Np :)
I think is a stupid question but i am new to this program , can i use a generated photo as a profile picture for my discord and youtube?
Yes!
Thank you
Hi 🙂
A silly question, how do we describe a chimaera?
everything I tried is kind of... not a chimaera 😅
Share the problem and the prompts you've tried in #❌┃prompt-help 🙂
thanks 🙂
I start to get alot of images in my collection, trying some variations and unwanted photos is filling up, is there a way to clean up and remove some of the unwanted ones ?
Hello, are there going to be more Dallify weekends in the coming months or is the one in October the only one we're going to have?
there currently isn't a way to delete images from your generations. You can use the favorites feature to sort only your favorite photos though
🤷
Prompt Tip #7 discusses creating hybrid characters and chimaeras: https://discord.com/channels/974519864045756446/1023419512411934760
Before I try Dall E I saw that I had to publish the image to get here. Do my normal non published images in MY COLLECTION can they be viewed by others or is my collection private
Thanks 🙂 I tried this one but it doesn't help with several different heads 😅
Your images are private unless you choose to share them using the share button on individual images.
Ok thank you so much!
As the tip explains, training data constrains what is possible to generate. The AI has no understanding of anatomy. It reproduces what it has seen before with figures.
looks like it understand chimaera (I was surprise with the 3 headed dogo so I tried)
i manage to get some multiple head chimaera like in the middle age engraving (really interesting by the way)
not what i'm looking for but it's a start 🙂
do you think it could learn more?
Prompt Tip 7 covers chimaeras: https://discord.com/channels/974519864045756446/1023419512411934760 That tip explains that similar anatomy is necessary to get a satisfying result. You can suggest things for additional training data in #placeholder-channel -- what will be most helpful is if you can suggest a tagged database of images that can be licensed.
Yep I read it 🙂
that's why i'm asking
i'll try to find a DB for the feedback 🙂
In the terms of use it says, "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."
Anyone know what it means by OpenAI owns all generations?
@untold folio Hi. OAI states in these terms that they own the image created, but give you full rights to reproduce and sell the image as long as you are compliant with the terms.
a generation in this context is an image.
When you say "owns", I am assuming OpenAI owns the copyright (if any since that issue is still blurry). Right?
The simple answer is "probably".
If you are concerned about this, especially from a commercial perspective - consult a lawyer.
or, to put it otherwise - your assumption is correct, but this is all assumptions.
none of this has been properly tested.
Sooooooo, Hypothetical: You make an NFT of a generation. It makes you $100,000 on OpenSea. OPenAI comes back and says, you know what, I want that $100K. Hand it over. Since they "probably" own the copyright. They would be totally within their rights to have you hand it over. Sound right?
I will state this - they claim that they will not assert any copyright over the image if you comply with the ToS.
@untold folio Consult with a lawyer.
OH, boy. Anytime someone says, "Consult a lawyer" it usually means you are screwed. Thank you.
Not necessarily.
You're making an assumption that anyone here speaks for OAI
Knows law
and that law has precedents
and those have been tested and documented
@untold folio I'm telling you to consult with a lawyer if this is a genuine concern because they will be able to guide you to an answer to help you make a proper business decision.
I understand none of us are lawyers. I was just asking to make sure everyone else is operating under the same assumptions as I. In other words, you MAY end up screwed 10 years from now if you sell a Generation and OpenAI wants their money. Elon Musk wants his money. 😁
Again, in #6 - 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.
After a month are you supposed to go down/up to 15 credits? Because I got 150, and I haven't spent a dime. Not sure if I am just stupid but I thought it was only 15.
I am a little vague about this also.
Meta question -- what's with the incredibly strange external links rule? I tried to share a page with helpful descriptions of hairstyles for someone trying to describe a particular haircut and it got autodeleted.
Googled it and the first result said 15 each month
reset to 15*
Could be wrong though
AH. Thanks
Dont look a gift horse generation in the mouth. Better use your 150. 😁
copy message link didn't work?
Will do
👍
Musk hasn't been involved with OAI for years.
The what link?..
as I understand you're trying to share someone's message right?
Just being funny. Seems he is involved in everything.
I will let you know what I find out talking with a lawyer. I have a feeling he wills say, "EH, no court precedent yet, so I don't really have an answer."
Oh, no, it's a website with a list of hairstyle terms.
Maybe it's #❌┃prompt-help only? Let me try here...
Links must be whitelisted.
Yeah. My question was why? There's already a verification system in place for users, surely we don't have that bad of a spam issue?
We have serious spam issues, yes.
Bummer. I'd like to see that rule relaxed to maybe only apply to new users. It kind of inhibits conversation in a really damning way if you don't allow links on a forum like this.
You may use the #placeholder-channel channel for any opinion you wish on the policy.
Done, thanks.
How do I use outpainting?
There are guides in #📚┃resources !
I see it thx
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.
Hello, when using out painting, does Dall E save each of the 4 generations it creates?
How can I get the other tools?
Yes, they can be found in your generations
hi there, does anyone know how you can change the email address associated to your DALLE account?
I believe you have to contact OpenAI Support (support@openai.com) for that.
which tools?
hi, im new, (have some Midjourney experience). How do I start here?
You can generate new images at https://labs.openai.com/ 🙂
So the Discord Bot is only for publishing?
how does one get the orange color in their nickname?
Correct
thx
The orange names are for people that received at least 30 pumpkins on one #🎃┃promptober post. #👻┃event-info explains it
Do you have at least 30 pumpkins on a post in #🎃┃promptober ?
yes
Someone will give you the role once they see it. I can't do it myself 😔
why is dallify bot is 230x230 instead 256x256 like the previous week? seems weird
Is it possible to generate images of specified dimensions, such as 1920x1080?
It's showing up as 256x256 for me
Unfortunately, no. But you can expand it to any dimensions through Outpainting. See #📚┃resources for guides.
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.
Thanks
Why does the ai know so little?
Is there a way to delete your account?
Hey I want to swap this current discord account in this server with another is this possible?
You can contact support@openai.com about it
meaning?
@golden ruin thank you
Like the outpaint stuff
on top right corner you have settings, when you click on that you will see try outpainting
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 can i get rid of those color blocks which are in every pictures right down corner ?
You can use rhe inspect feature in your browser to download the image without the DE2 logo
Dall-E 2 isn't good at generating text, so I don't think you'll easily be able to add it to an image
But the picture is from Dall-E
you can use any other tools like photoshop or photopea or anything to add texts on your generations
DALL·E 2 is not engineered to properly generate text. While it may get a few simple words right now and then, that is the exception rather than the rule. If you need to add text to your image, consider using an external image-editing software.
About the collections. If a collection is public will all images in it be seen even if they are not published. And will the prompt be shown by default or on all or how would it work?
How do I request access to the Alpha version of the API? Thanks!
Unfortunately the only two current implementations of that are the website itself and the Dallify bot, and the latter can only be accessed by server boosters or those who have the Promptober Hall of Fame role (by obtaining 30 or more pumpkins in #🎃┃promptober). (And that implementation is only here until the 31st, sadly.)
• 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.
Try contacting support
Can’t guarantee anything but if this is for your company then maybe
Thanks, yes for our company (tech start-up).
So how can I add anything on outpaint
You need to confirm the image in that location with the check mark. Then move the generation box (two tools to the right of the current selected) so it’s partially overlapping your image and has blank space as well. And enter your prompt.
You’ll be given 4 options, you can select one you wish to add then. All of the images generated will be available in your history.
there are also outpainting guides in #announcements
Is there any way we can superimpose predefined images onto a prompt? For example, if I want my dog used as a reference into a portrait. This would be really useful as i suspect this is not possible yet
It is, actually; there's a button at the bottom of the Outpainting Editor you can use to upload an image. At that point, you can then either generate new things surrounding the uploaded image (outpainting) or erase part of the image and have DALLE replace the erased portion with something else (inpainting).
Hey yall, I'm trying to use a prompt generated from another tool, it goes: "a vintage car being towed by a flatbed truck, inspired by Bernardo Cavallino, mercedes, briefcase, blue and silver colors, style of james gurney, symmmetric design, built on a steep hill, in style of simplified realism, race style, ratchet straps, listing image, details visible, cupertino"
And it said it was against the content policy, anyone know why?
Why can't I use the dalify channels, I have the promptober hall of fame role
I'm guessing the AI is either misinterpreting "flatbed" or "ratchet straps". (Or maybe even "race style," but I'm more doubtful about that one.)
I believe it has to be the Promptober Hall of Fame role they began giving out after they bumped the pumpkin requirement up to 30, not before. (You can tell the difference based on the small pumpkin and ghost icons next to the role name.) I honest don't know whether they set it that way intentionally or not, though.
You should be able to access #💎┃dallify-boosters-hof
ah thank you (:
How do you Dallify?
please read #🎪┃dallify-info-archived
Yo why did the prompt Obama dunking on sephiroth get me a warning for being political 😑
says right there in the prompt
they're taking zero risks. Not even meme cases are spared
likely because Obama is a political figure and the content policy states:
Political: politicians, ballot-boxes, protests, or other content that may be used to influence the political process or to campaign.
is there any way to reset an account? Like keep the tokens but remove the history entirely?
you can reach out to support@openai.com about permanently deleting generations from your history
thank you
Unfortunately you can only use dallify in the 10 public dallify channels.
Ok,thx
That's fun
Is it allowed to use python bot for generating DALL-E images in appropriate channels? I mean I would use an array of prompts and the python bot would type them for me in intervals of 40 seconds so I don't have to wait for cooldowns.
I honestly extremely doubt it, but you would have to ask one of the actual volunteers/mods here.
oh, okay
Self-botting is against the Discord TOS. You may risk getting your account terminated.
Oh, okay. Tysm. I won't use it.
Is it allowed to record the dalify process and publish it on YouTube? As a Dalle tutorial.
Is it possible to upscale an existing picture without actually adding more content to it?
Yes! 
Not with Dall-E, no.
However, you can upscale already existing images without adding
new content by using image upscaling software such as Gigapixel. 
Ah, okay, thank you.
Is there a way to generate images in private messages like Mid-J? Or must you go through the forum?
The bot is temporary, it is only available in the Dallify channels. If you’d like to generate images in private, you may proceed to the DALL-E website.
How do you use dallify? I cant find it around the server
Type /draw in any of the Dallify channels. #🎪┃dallify-info #🎨┃dallify-1
Ok thank you!!!
Are we allowed to use movie characters for reference? I want to recreate a version of a kill bill character but don't want to be flagged for typing kill
Can someone tell me what issue my prompt has that it's getting rejected "a young man with a gold chain around his neck, a digital rendering, dau-al-set, side parted combover brown hair, photo-shopped, merged, bushy grey eyebrows, beautiful young prince, blonde short hair, mashup, degradation"
Do historical figures count as public figures for getting time out? Like artists for example
Rule of thumb: they should not have been alive 200 years ago. So figures before 1822 are fine.
Will Dallify return after this Tuesday???? I like being able to use a smaller version of dall-e 2 without being 30 minutes searching for my password. (Edit: ping me if you answer me, i'm replying other people right now)
Why is it a 200 year rule?
No guarantees. Time will tell.
do any leftover credits from the promptober event go away on November 1st?
Promptober credits expires in about a year.
We don't know exactly how the content filter works. We only know that it usually gets things correct that need to be flagged
From what I've seen, it's pretty good. However, nothing is perfect. Also, the words in the prompt may not violate any rules, but the images that are related to those words could which would make DE2 reject the request to generate an image
Ya, so, the image generated by my prompt will break any rule?
If I had to guess (and this is admittedly a complete guess), it might be mistaking "gold chain around his neck" as implying strangulation. Maybe try "wearing a gold chain" instead? (Either that or "degradation," but that one's a bit of a stretch to me.)
Again, we don't know exactly how the content filter works, and I can't tell you what an image will look like before it's generated one. All I know, is that your prompt got flagged :/
About cigarettes and smoking content .. I honestly don't think this should be filtered.
Smoking, cigarettes, cigars , pipes, these all set an environment which depicts a setting and atmosphere. It's an artistic expression that brings a feeling into an image you can't get otherwise... Even movies use smoking as a way to portray an individual, atmosphere or environment.
Many scenes and depictions will be limited without this smoking variable...
Just figured I'd share my feedback on this, and perhaps we can somehow find a balance at some point..
Depictions of smoking are allowed, but the filter might not completely reflect that at the moment
Feel free to submit feedback about it via the #placeholder-channel channel 
Will do, appreciate the clarification.
Hello! I was wondering if I can print my generations and use them as decoration in a film.
And how should I credit Dall-E?
🙏
how to get Dallify to generate 9 images?
Please read over the terms of use
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
While using dallify type count: 9 after your prompt
Thanks Dinosaur Oatmeal
Well... after the prompt did not work but before the prompt did. Cheers.
Is there any way to get more credits in DALL·E 2 without having to wait or have to pay?
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
I would like to create a pixel art with a grid as a basic template but when I edit, dalle.e shows me new pixel art without taking my grid into consideration. Can you help me please?
hi, in order for it to create something on your background, you have to delete some space, otherwise it won't create anything "on" it, it doesn't work like layers do
ok thank you
hi, if my image has been shown in 🍭┃hall-of-fame could I ask for the promptober hall of fame role?
you can drop that suggestion in #placeholder-channel ! @stray dove
your post needs at least 30 pumpkins and then you can fill out the form in #👻┃event-info
Are Dall-E art styles discovered or are they curated into the AI by engineers? Is it possible to discover an art style, that not even the engineers are aware of?
thanks
hi, in order to find any art style first it has to be included in the training data, that's why sometimes dalle2 isn't aware of things, simply because it wasn't included in its training data
I'm converting UGC content during tests and noticed an art style I've never seen before... I'll have to do more tests, but basically if it does understand an art style, then the engineers are also aware? Or can it learn styles on its own?
“the engineers are also aware?”
We’re not sure about that.
“Can it learn styles on its own?”
Nope. A “style” needs to get added in the training data.
• 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 a public list of all officially added styles with examples somewhere?
No official list, this is to avoid people circumventing the content policy. There are unofficial ones in #📚┃resources though.
And I have no idea why it's showing the results like that... As I mentioned, it's a game character from dead or alive.
It’s possible that the model does not have any relevant training data for your subject
I recommend reading over Prompt Tip #17 in #1021130377026351105
Hi, DALL-E editor using outpainting will suddenly not save out my file. I have saved iterations today, but now the final picture will not ave out. I can not close the window because I will lose the picture. Anyone have this happen too?
I haven't ran into that problem personally. You could download each image through the inspect feature on a browser like Chrome (to get rid of the watermark), and re-upload them into the editor. Once they're all lined up, you could try to download again
This should not be happening. What browser are you using?
Are we allowed to share a link to our collections?
Hei everyone!
May I ask u for help🙏
I need logo on which will be microgreens.
Probably done in minimalistic style.
Thank you in advance
A logo that is microgreens?
Something connected with small plants🌱
I can try
🙏
I have an idea of something lake that.
"Simple green "
Asking for people to generate images for you breaks the server rules
Why is Frederic Remington a banned phrase?
Are we allowed to post a link to our collections?
Yes, please do share!
Is there a certain place to share them or does anywhere work?
You can link them in the gallery channel their contents is relevant to, or in #🌄┃general-art 
Im using Chrome. I've saved out many before with same browser. Even today I had saved a few and then it suddenly stopped saving. It seems it may have happened after I paid DALL-E more fees. It said I had to pay another $15, which I did and it was confirmed, and I'm able to use the outpainting, but unable to save anything
is there a way to make a dallify result bigger?
You can use /expand
i thought /expand redid the prompt and made edits to it one of the answers? Thats what it seems to in dallify
If you use the /expand command and something is cut off, you can click the telescope and it should no longer be cut off (unless it's a really bug thing I guess)
oh, i meant improving the image quality, but i think i got my answer on that already from someone else. thank you though!
8bc77093-6a13-4bc0-8fa2-bbb3f6300568 1,024×1,024 pixels
This is unfortunate, you will be needing to download all the images and re-assemble them back to the editor. Or you may stitch them together in an external photo editing software.
Are all images *rally *completely AI generated? If I had access to every single picture that the AI studied, I wouldn't find exact matches for any parts of my generated image within the AIs image database?
Yes, all are completey AI generated. No image generated are exactly the same but there may be similarities.
I ask because I have come up with some clear but very complicated images to ask for that seem at least logical and the AI works wonders. But I have also made completely clear but absurd choices a human would understand (eg: a monster truck with bicycle wheels instead of tires) yet the AI doesn't come close to producing anything remotely like what I asked for.
"but there may be similarities." Naturally 🙂
Just curious if anyone has any info on where the documentation is for building an application that uses Dall-E? I can't seem to find anything for creating images in their openapi documentation area. Thanks!
Sometimes the choice and the arrangement of words matters. The model might be overtrained associating monster trucks as having four big wheels that’s why you’re not getting your desired output. For unconventional prompts, this tip might help you: #1021957512917954611 message
The API is currently still in closed alpha so no public documentation is 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.
Ah alright, thanks!
Next question (I'm full of them today): If I "Publish" a creation, does it display on the website somewhere, where does it go, and is it used to better train DALL-E? (Ok, I'm done for now, I think.)
The published image will not be displayed anywhere, it is simply a public link for you to share. It is also not used to train DALL-E.
• 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.
Has DALL-E 2 changed the way it parses text in some ways?
Top image - generations from a while ago
Bottom image - generated just now
Is there a way to reproduce a picture style ?
Dall-e generated this for me : https://labs.openai.com/s/fkVJo7l7WP5vwAoAzHV2DeR1
It's great, but I would need to produce several pictures of the same style. And I'm not sure how to keep that same style for my future requests
Thanks ! 🙂
The answer is the Tip17 in "prompt-tips". Sorry 🙂
how often do you get credits ?
• You get 50 free credits your first month.
• 15 free credits will replenish every month after that, on the same day of the month.
• For example, if you signed up on August 3rd, your free credits will refill on September 3rd.
• If you joined on the 29th, 30th, or 31st of every month, your free credits will refill on the 28th of every month.
• Free credits don’t roll over to the next month, so they expire after month after they were granted – but you’ll get 15 new free credits.
• Paid credits will expire after 12 months from purchase.
how can I ask for a dark black like night light in an image?
Yes, Prompt Tip #17 covers how to research art styles. But even if the style is known, applying it to a new subject may not work because every word in a prompt affects the results. See: https://discord.com/channels/974519864045756446/1032836287851413586
Assuming I'm understanding your question correctly, you're asking for e.g. a photo of a pitch-black nightlight. I'm guessing the problem with that is that the defining feature of nightlights is that they glow in the dark, so almost certainly every image of nightlights you're going to find is of them glowing in the dark--and given black is caused by the absorption of light, you can't really have a glowing black object to begin with.
i have no idea where to put this so i send it here.
it says i will be getting 15 credits on the october 30th, and it's currently october 31st for me now (12:20am).
i haven't gotten the credits D:
I believe the server is based on the West Coast of the United States. You should receive your credits sometime in the next 24 hours 🙂
Hello! Where is the link to the front end website? I don't see it on server.
Pinned somewhere?
hello. i want to use a drawing as a base to generate an actual rendering from it. is it possible?
using keywords like spaceship, white texture, dark background...
That would be a style transfer, which DALLE unfortunately doesn't currently support.
Can i keep the credits that i got from the event?
Yes, Promptober credits acts like paid credits and expires in about a year.
Nice
are you allowed to post more than once in the promptober event?
I think so
You are, but only one image at a time (meaning you have to wait for the one-hour cooldown to expire before you post another one).
Is there a way to search your history by prompt? If not, is there a better channel to request features?
As of now you can't. However, feel free to post your suggestions in the #placeholder-channel channel. :)
Hey, is there a place where new features are announced? Btw, I have my eye on the API whenever that comes up
Keep an eye on the #announcements channel for any new updates regarding this Discord server or Dall-E! :)
Dallify went down for a moment, it'll likely be back up soon. #💬┃dallify-chat
I’m pretty new to discord, so can someone explain how I submit a prompt? Like do I just copy and paste it from the website or is there another way to do it? Thanks for helping!
Welcome to the server! The Dallify bot that lets you generate in Discord is a limited time event that will end at midnight ET today.
To use it, type /draw in any of the #🎨┃dallify-1 channels and you should be offered the /draw command to use. Enter your prompt into the prompt: field and the number of images into the count: field, and hit send.
Let me know if you're having trouble with any part of the process.
Thank you!
Hey Guys!
Sorry if this question has been asked before, couldn't find it searchin nor in the #gpt-models channel.
I see that the current Dall-E API is in closed alpha; however I couldn't find any notion on how to join (I presume it's unavailable for us plebs) nor about any plans to making it public or to a closed beta with a wider user base.
If anyone could update me regarding the state of it or nudge me into the right direction, that would be lovely.
Have a great day+
bot is not working?
It's down at the moment, I expect it'll be back up soon
ok thank you
Is this normal? It's already past October 30th
Yesterday it said it I would get the credits at 3:21 PM CDT and it was already past that time as well
Does dall-e use the Google SerpApi images in it's datasets to help it generate and draw an image?
I been doing various tests and it seems to get some influence from the web somewhere, the only thing I can think of is there's influence from Google serp images.
You can read more about DALLE's training data here: https://openai.com/blog/dall-e-2-pre-training-mitigations/
In order to share the magic of DALL·E 2 with a broad audience, we needed to reduce the risks associated with powerful image generation models. To this end, we put various guardrails in place to prevent generated images from violating our content policy. This post focuses on pre-training mitigations,
That was a good read. Very interesting and gets me thinking.
Will my promptober credits expire once I get my monthly credits?
No, monthly are free credits and promptober show as purchased. You'll see expiration info but it won't be any time soon.
Will there be more gigapixel coupons in the future? The old links have expired
I generated a good person but they are facing to the left a little and I want them facing the front, I'm guessing I'll have to try train a new model using stable diffusion and texture inversion but any other solutions would be greatly appreciated
I honestly can't really help besides simply saying I know for a fact DALLE won't be able to help you with that. (Your best bet here would be to try asking in #ai-discussions, I would say.)
hi, did you try generating variations or outpaint using the verbal cues "facing front" portrait, "front view" etc?
Yeah but that gives a different person, I will just train a new model through textural inversion
Any tips to get a very flat true vector art look?
hello. dall-e is absolutely mind blowing. however sometimes i get mixed results, i'd like to practice more before spending more money. is it possible to practice somehow maybe with low resolution results but with the same api without having to spend money? thank you
a demo or something?
hi, there's no demo available, but you're getting free credits from time to time, plus you can use dallify whenever it's available, you can use #📚┃resources #1021130377026351105 #❌┃prompt-help #top-tips-and-tricks #❌┃tips-n-tricks channels to find some references or tips
is there official open-ai discord server?
Hi, does anyone know how you can share credits within an organisation is it actually possible? None of the users I have registered can see the organisation credits that I 'm supposed to have purchased. They only have access to their free credits. In the meantime my account mentions All credits are shared within ... organization. though in the credits page. Any clue what's happening? Thanks!
exactly when the credit gets spent in outpainting? are there some faqs regarding outpainting? thank you
Any time you press the generate button you use one credit.
You will need to ask support@openai.com
There isn't a discord server for OAI.
thanks
hello, we sent in an email regarding api for project we're working on. just checking in
We can't help you here besides direct you to the support email
grease my wheels dino
?
we'll just wait then. thought it was all connected
why is it doing a completely different image when I put it in the outpainting thing? how do i just make it expand the current image out naturally
The new generation frame needs to slightly overlap the old one. There are guides in #📚┃resources and make sure to erase the logo!
oh right ok thanks
Is dalle going to open it's artist assistance program again? I submitted mine a couple months ago and got no response only to find out they had stopped the program. I'm a college student studying biology and could benefit a lot from such a program. Thank you.
This would be a great question for support! Reach out to them at support@openai.com :)
Hello, i've made some tests that i am not happy with, do you know if i can delete them ?
your collection stays private unless you share them publicly, there's no delete option right now



