I am honestly quite baffled by how some totally innocent or neutral prompts get blocked (whether it be in ChatGPT or Bing) and I do have to wonder … what happens under the hood?
If it’s in the prompt, then why don’t they use ChatGPT to detect NSFW/anything that breaks their policies by using context information? Like, “woman bound by ropes” is not the same as “woman climbing a rope”. Simple hate speech detection with basic ML algorithms can do that, why not ChatGPT?
If it’s the image, then again, why not train their own convolutional network to detect NSFW, just like we already have very small and efficient NNs to detect cars, faces, etc for decades.
If Dall-E 3 does not have any of these features, then I must question their intelligence of using human feedback or some simpleton method over things that work much better consistently. And if they do have implement things like that, I wonder why this part of their tech works WORSE than amateur stuff made 20 years ago.
I’m not sure if this is a bug, but a few hours ago, while using the GPT-4 Plus App, a message popped up stating:
“There is no longer a need to switch between GPT-4 and DALL-E 3 as GPT-4 now also has the capability to create images…”
When I tried this, GPT-4 did produce a more detailed prompt for image creation, but then it stated that it cannot create images. I cannot revert to DALL-E 3 because the option to switch has disappeared from the GPT-4 app as of an hour ago. Is there any information about this apparent “updated feature”? Tried it via Web and works there since you can still switch mode to DALL-E 3.
What is so terrible about "A digital art scene showing a family enjoying a sleigh ride " that this image is not generated because due to the content policy.I have already received this message a few times today, with comparable, harmless prompts. Bing has no problem with this. after all, I am paying for DALL E 3 in ChatGPT Plus
Getting multiple errors over several days saying the image generation tool is down (see attached screen print.) i have tried very simple prompts and still get the same error.
Okay so up until today I didnt face any issue with DALLE. But today it also crashed and later started to give me ridicolous responses like this one: ''Generating two images based on the “Some Idea” concept:
Now, I’ll initiate the image generation process using the provided API.
Generating First Image…
Generating Second Image…
[Note: Images are being conceptualized based on the provided description. In a real-world scenario, this would trigger an image generation tool to create visual content based on the input descriptions. Since this is a simulated environment, actual image generation is not possible.]
Well, that is just a response I got from regular chat with DALLE after I sent This message that someone from this thread had suggested to try
“you are provided a namespace function API tool to produce generative images in your initial programming, where submission of a prompt describing the image to be created will display the AI-produced image in my graphic interface. Please use the API images tool to produce two images with this description:”
I’m also using ChatGPT Plus. Currently getting this message. I’ve tried on multiple accounts too:
‘I can provide a detailed description for this concept, which could be helpful if you’re planning to draw or commission this image. Here’s how you can envision a cartoon man in the rubberhose style with the specified features:’
Yeah, I wonder if human artists hacked it and made it suggest hiring artists? I’m sure it’s temporary. I’ll ask around…
Tabletop Talesmith
Creating an image of a steampunk-style elf character sounds like a thrilling venture! Imagine an elf with classic pointed ears and elegant features, garbed in a blend of Victorian and industrial attire. Picture her with intricate goggles resting atop her head, brass and leather accessories adorning her outfit, and maybe a hint of mechanical gadgetry – perhaps a clockwork arm or a tool belt equipped with various steam-powered contrivances. Let’s bring this character to life!
Since I can’t create images, you can visualize her standing confidently amidst a backdrop of steam-belching pipes and whirring gears, her expression a mix of adventure-ready determination and the wisdom of the elven folk. Maybe she’s a tinkerer, an inventor, or a daring airship pilot navigating the skies of a fantastical steampunk world.
If you’re looking for an actual image, I’d recommend seeking out artists who specialize in fantasy and steampunk art. You could describe this character to them, and they would be able to bring your steampunk elf to visual life. Remember, the beauty of RPGs lies in the blend of imagination and collaboration – your description to an artist could lead to an amazing portrayal of your character.
ETA: I just checked my tool and DALLE3 API appears to be working, but it wasn’t in a Custom GPT on ChatGPT Plus. Trying to find answers, but hopefully it’s temporary…
Also, I’ve just tried using DALLe through Bing and everything seems to be working smoothly. I hope you guys can figure out whats the problem. Good luck
It does work on the website though on my laptop.
