Because of censorship, removed

Stop giving them ideas!! lol

We got 10 I believe in the first Labs iteration…

Double might come back as the network stabilizes, I bet.

Next up is -2 images! You must supply the AI with two of YOUR drawings that it requests before it does anything! Muahahaha. :wink:

1 Like

Sometimes I have two, when I explicitly state N:2.

But since the new version of ChatGPT, it doesn’t listen to my custom instructions very good.

I asked to show the exact JSON formatted payload it send to the image_creator and the exact parameters it got back.

That worked perfectly, but now it just spits out images, without showing these (useful) parameters.

I made thousands of drawings by hand in my life, so that’s no problem.

With the new Hot Mod model I am turning them into new creations, which is really fun to see.

Cartoons from the 80’s and 90’s (I drew back than) are now redrawn by GPT in the Hot Mod model.

1 Like

I found now I can only receive one (1) images by default :cry:.

1 Like

Correct, the system can’t handle the load. They are even blocking new customers.

Even when you explicit set “N : 2” it wil create one.

But you can say “Another one!” and it will do.

I used to, when I reached the DALLE limit, GPTs let me know. Reaching the DALLE limit wasn’t as difficult as I originally thought, I already had a DALLE burner type implementation.

There isn’t a separate cap for image generation. You have a limit of 40 messages every 3 hours, but this cap can fluctuate depending on the number of people using the service simultaneously. If you send a message at 5 o’clock in the evening, you might only be able to send around 20 before you get shut down. However, if you’re sending messages at 4 o’clock in the morning, you’re more likely to be able to send all 40 messages.

1 Like

Sorry for the late reply, I hope it was clear enough.
Since I have a working model created with a single prompt for more than 30 images, even though in actual operation it cannot be that continuous, it is faster to create than any method.

I hope that clarifying this matter won’t cause another few days of blocking work or creating limits from the system.

I’m tired of giving useful information and experiencing betrayal from the staff. Moreover, I don’t adhere to these things, while Sora is admirable. I found DALLE’s amazing response to "i think you show image in wrong side. can you flip it " so perfect.

If people tried to find out what DALLE could do, Sora would definitely come out more complete.

ในวันที่ ศ. 23 ก.พ. 2024 เวลา 03:14 Elijah via OpenAI Developer Forum <notifications@openai1.discoursemail.com> เขียนว่า:

I was originally confused by this post, because DallE has not been having issues creating multiple images for some time now.

However, I went back to a previous thread of mine of some icon generations where DallE could not tell its right from its left and posted your “i think you show image in wrong side. can you flip it” prompt just to see what happened.

The result of randomly selecting the 4th generation, using code interpreter to flip the image and provide me a download link for the new version was unexpected. I assumed it would just say “sure” and then generate another incorrect facing Arrow. The icons were bound for Adobe anyway, but the result was still interesting. Thanks for the share.

2 Likes

Please create a new topic if you wish to continue this conversation.