Hi,
I am using In DALL-E to create images. In general, I create three-panel images, and there is a scene from a sequence in each panel. I need the result to follow my instructions.
It happens to me often that the basic image is quite good, but when I try to edit it a little bit, DALL-E does not do what I ask, and I do not know how to force him. I spend a lot of time on useless back-and-forth. My strategy is to use the edit tool and give instructions, add an example of what I want to change, specify instructions in the same response, and give feedback on the results…
I attached an example.
My prompt:
Creates a triple panel with three pictures of the same woman on a white background. In the first, the woman takes off her earrings. In the second, she washes her face. In the third, and dries them with a towel. 3D cartoon style with different expressions in each panel
After reasonable back-and-forth iteration, I got the following image:
My problem with this image is that in the first image, it is not clear she takes off the earing, so I uploaded an example of the required position,
used the edit tool, and added the prompt: "Use this example to change the first image in the panel. "
This was not working, so I tried to give a prompt that illustrated the hand position, and still nothing happened.
I would appreciate any advice:
How do you make it follow the prompt? When do you stop trying? Do you use other AI tools that can do this task better?
Thanks