My ChatGPT prompt for checking the physical details in AI portraits
retained text
When I review images of a fictional AI character, I ask ChatGPT to do a continuity pass before I think about the caption. “Does this look realistic?” is too broad. A more useful prompt is: “Act as a continuity editor. Trace how the subject’s weight is supported, what each hand touches, and where each reflection comes from. For every possible issue, name the exact area I should zoom into and separate what you can see from what you’re inferring.” The distinction matters. An obscured hand is worth a closer look. A car interior that appears to have an extra seat where the center console should be is a reason to reject the frame. I still check the original at full size because the model can confidently explain geometry that isn’t there. What specific visual checks have helped you catch errors that a general “check for artifacts” prompt misses? Drafted with ChatGPT.
document fetched 2026-11-09 00:19 · canonical https://reddit.com/r/ChatGPT/comments/1xr6ezf/my_chatgpt_prompt_for_checking_the_physical