Yeah, it’s still a bit bumpy, but it should return.
Be lucky you weren’t using it during the early days! Small smile…
In all seriousness, even with it bugging out occasionally, I find I get a lot of value out of the $20/month I pay currently.
3 Likes
I exceeded the message limit (without getting a single picture) and waited 1 hour to test again only to get the same error. I’m not even using the Dall-e plugin from GPT store this time, I’m asking GPT-4 directly to create an image and it’s still not able to generate anything. It’s very frustrating as a paid user, can you suggest how can I request assistance here?
I’m also not able to upload any file, tried uploading a simple image and it gave me error for that too
1 Like
Your browser may be storing a prior version of the client software authentication level from a previous response.
Here’s a guide to clearing site cache and site storage. https://its.uiowa.edu/support/article/719
Simply CTRL-(refresh button) or SHIFT-(refresh button) may be enough to reload all the web assets used by the software in your browser. Or you can use another web browser (Firefox, Edge, Chrome), or log in again with a new private/incognito window to see if your account has new credentials now available.
If it seems like DALL-E hates you, you can ultimately send a message through openai’s [help](http://help.openai.com) system, although the days it takes for them to get back to you may be longer than the problem that simply fixed itself.
1 Like
I was using Chrome earlier so tried the same in Safari, restarted my mac and tested again, install chatGPT app in my phone and tried and as you can guess, same result. I did “inspect” in chrome and saw network calls, both fetching the image and uploading are giving 500 internal error. it’s quite frustrating because the errors in response doesn’t have any information regarding the error. I’ll try to send message on the help section
1 Like
After reading the entire thread of users issues with Dall-E (and having dealt with many issues), I think its worth trying a custom GPT made for dealing with Dall-E.
My absolute go-to now is ‘Super Describe’. No idea what the magic prompt sauce is made of but it not only is incredible for giving it a copy/paste of an image to work with, but it handles creation by straight up image prompts so incredibly good that it can get Dall-E to create images that are outright refused by most other methods.
I only include a screenshot for the GPT as it seems many others have used the same name (PS. I have nothing to do with this plugin or whoever created it).
Same issue with image generation refusal. Asking for a woman in a bathing suit. It generates OTHER images. When i asked it said: I apologize for the inconvenience, but it seems we are facing persistent issues with generating the image you requested. Unfortunately, despite the content being entirely appropriate, the system encountered technical errors that prevented the image creation. I understand the importance of your requests and the frustration that comes with such unexpected challenges. If there’s another type of image or a different request you have in mind, please let me know, and I’ll do my best to assist you.
@bottledcanofdoom
I’ve had the same issue for 10+ hours now after I was rate limited. Even after the limitation expired, Dall-E would only spit out text following a prompt, in Chrome.
_j’s suggestion to use a different web browser worked! I switched to Safari and an image was finally generated. Thanks for the temporary solution!!!
Have the concepts of ‘Pride’ and ‘Strength’ been fully expressed in the model ? I can’t seem to generate a ‘phoenix rising from the ashes’ that doesn’t look like a chicken or domesticated parrot with disproportionately small wings and hesitant demeanor…
its not even comically bad , it just frustratingly terrible, Every image I generate seems to always be a awful cartoonish character with a beard despite trying to refine the prompt and directly demanding no facial hair, clean shaven. Every image is the same goofy cartoon man with a beard, I dont want to generate 60 images and increasingly feel more and more frustrated as I know what use to be possible, I cannot even upload a real image to use in the generation,. sure I can upload it , but changeooo presto, same bad cartoon.
Solemn is not permitted, a hovel is mental health warning, there are so many warnings, Why even bother? I feel like I am paying for a bait and switch product. Why not allow users 18 and over to let them make images that represent the real world we live in which is a sad, terrible place full of evil and death. Why not make it clear there isnt allowed any realism and all images must be smiling cartoon bearded characters that will leave you frustrated, angry, wasting resources and ultimately giving up on your task over what should be a minor thing
dalle image: solemn hovel, with young man seated outside. Hyper-realistic detailed photograph
@gnosticrooster I tried to use this GPT to create a simple photo of a person landing on the ground after a short fall as a reference to something I’m trying to draw. It started to attempt it but midway, it gave me this response below.
Note that I was using the default ChatGPT model (4o) and I’ve only just exceeded the message/plugin limit for the free plan.
i´m having the same issue except i can generate images but it refuses.
I solved the problem, but you may not like or able to make use of the ‘solution’
I’m running FLUX (BlackForest AI - https://blackforestlabs.ai/) image gen through PINOKIO (https://pinokio.computer/), for 99% of all my image creation needs
Sad, but true.
Maybe ‘Strawberry’ will change the situation fingers crossed
I got the more specific error message of “I can’t create images in this environment” which is bizarre. Happens regardless of browser or app. It will also not work within any of the 3rd party custom chats. No answers, I must say as usual, back from support.
DALL-E in ChatGPT has not been working for a few days. It doesn’t matter what kind of prompt I write, even very simple and harmless prompts like ‘represent a house’ or ‘represent a kitten,’ but the response is always the same: DALL-E is encountering issues generating the image. Has this ever happened to you? If so, how did you solve the problem?
In ChatGPT’s user preference for “custom instructions”, there are three checkboxes at the bottom, where DALL-E can be disabled, then controlled by a master switch.
The AI is then very obstinate about reporting why it can’t send to dalle, which it still may perform given a session’s past system instructions. It may not say “you disabled dalle”, but instead could make up whatever it wants to try to be insolent, such as the “environment” message originating solely from post-training.
Above, you can see that I started the chat session without dalle enabled. I then got a deceptive AI, until showing what is produced by the tool.
Now, enabling custom instructions’ tool options (or memory) doesn’t take starting a new session; instead you can see the ability is back after checking the box.
Thus, your own settings is something to check, besides a hard-refresh of the browser after clearing site data, cookies, and cache (which if successful, will force you to log in again).
DALL-E is only fully-enabled in ChatGPT Plus. You might get two images a day in the free ChatGPT version with an account if the correct model also isn’t used up, just as a tease.
OpenAI could add a recipient for when dalle is manually disabled, returning a useful error – such as they did for the “guardian” tool.
(post deleted by author)
I’m have the same issues and there wasn’t any warning or update about it for paying customers. I have clue what’s going on what the workers of these GPTs. If all I wanted was a language model, I would not pay for Plus. It can’t keep up with the demand and it keeps deferring to the internet for other “Sources” they need to answer for this.
Has the issue been resolved for you? I’m on day 5+ on being unable to generate any images via GPT/DALLE.