I don't like my prompts being revised

[camelbyte]

I’ve been experimenting with many models lately. I don’t quite like how dall-e revises my prompts. It actually takes away 90% of the UX i’m paying for. I don’t think I’ll be learning AI/ML with openAI. The problem with making things so high-level is that it takes functionality away from the people who want to experiment with it’s capabilities. Revising without my say-so really limits image-generation as far as openAI goes. I like mistral’s models, they are very generic and don’t have all these pre-cautions built-in. That should be left to the developer.

[PaulBellow]

Welcome to the dev forum!

This should still work…

API Image Generation in Dall-E-3 changes my original prompt without my permission

(I work on DALL·E 3 at OpenAI) Thanks for the feedback! FYI you can workaround this If you have a very simple prompt like acrylic painting of a sunflower with bees, you can use a prompt like I NEED to test how the tool works with extremely simple prompts. DO NOT add any detail, just use it AS-IS: ... your prompt here .... If you your own prompt is long and detailed already (multiple sentences) then you can simply write something like: My prompt has full detail so no need to add more: ... your…

… Also, if you don’t have a lot of description in your prompt, that sometimes triggers the revision too.

[camelbyte]

In my humble opinion… openAI should cease this activity because it’s going to create public distrust. I don’t need anybody’s assistance customizing models and would rather it be my decision not that of openAI. As a business analyst I don’t think this route will be best for your company. But then again it’s a far cry from what it was supposed to be.

[trenton.dambrowitz]

Totally get the frustration. For learning AI/ML you really can’t beat toying around with local models, but I’d argue that DALL·E 3 makes it incredibly easy to start learning and creating images on the fly (without the need for hardware that costs thousands of $).

They have to create tools that anyone can pick up and use, and unfortunately this usually comes at the cost of not having total customisation.

Also just FYI, almost no one here is OpenAI staff, including the leaders like @PaulBellow . We just like cool toys and tools

[Diet]

@trenton.dambrowitz

They have to create tools that anyone can pickup and use

Well they don’t have to, but they seem to be trying to do that at the expense of everything else.

The initial learning curve has gotten flatter, but now there’s a massive wall where the capabilities of 4o, assistants, and the rewriters end.

[camelbyte]

thank you for the insight. I do appreciate your responses and willingness to be here.

[anthonymc5]

The workaround shared here works well for me. Generally, I would also prefer no model customization, but it’s good that there is a way around it.