#Feet is always out cut from the image
1 messages · Page 1 of 1 (latest)
Have you explored describing the shoes, and/or where the person stands?
Well, no. I promoted it to generate a random individual.
How you word the prompt will surely matter.
To make sure you see the feet, help guide the model's attention to the feet in a way specific enough to get details.
'how they stand' kinda worked for one, not the other, I expect it works some percentage of times if I tried that a lot.
Asking the model to 'note their shoes' still left it random, but put attention on that detail needs to be in the picture, in case that helps your exploring.
The only issue I have is that arms are missing
And now apparently, I asked DALL-E/ChatGPT to note arms... it couldn't process it
Interesting. Is it fit to share screenshots of prompt and output?
Nvm
I think a shorter prompt fixed it
I labeled overalls under negative but that did not work. How do I make DALL-E/ChatGPT to not generate certain things?
Prompt
Generate an adult female character
Full body, character illustration,
Enhanced image, High quality, professional quality, high visual details, high saturation, high quality shading
Horizontal image,
Casual outfits, front and back, 2D anime art style
Negative:
- Chibi
- Overalls
"How do I make DALL-E/ChatGPT to not generate certain things?"
I usually handle this by telling it to make something else that would conflict with what I don't want.
If I don't want 'a red ball', I might describe the ball being green, or a green box, or whatever, so the idea is not even in the model's attention.
I usually do not use any negatives, especially with image gen
Ok
After the first image generation, you can ask it to preserve the image and zoom out. It usually works for me.
Ok