I really like the new shared image template feature in ChatGPT Images.
It is a very useful and creative way to let other users try the same idea with their own images, and I think this is an excellent addition.
I noticed one behavior that may be worth improving for clarity.
Here is an example shared template:
I intentionally used a very simple prompt here as a test case, so the behavior would be easy to verify.
The preview image shown in this template was originally generated as a morning café scene.
When creating the shared template, I changed the prompt text from “morning” to “evening.”
The preview image remained the original morning image, but when I actually used the shared template, the newly generated image followed the edited prompt and became an evening scene.
I will attach the resulting evening image as an example.
So the current behavior appears to allow:
- the preview image to remain from the original generation
- the shared prompt to be edited afterward
- the edited prompt to be used for new generations
This flexibility is useful, and I do not consider this a bug.
However, from the viewer’s perspective, it becomes difficult to know whether the preview image was actually generated from the prompt currently shown in the shared template.
I think the feature would become even better if the relationship between the preview image and the current shared prompt were made clearer.
For example, one of the following could help:
- keep the original generation prompt associated with the preview image
- show a small note when the shared prompt has been edited after the preview image was generated
- distinguish between “original prompt” and “shared template prompt”
- optionally regenerate the preview image after editing the shared prompt
The shared image template feature itself is excellent, and I would love to see it continue to evolve.
A small improvement in prompt-to-preview transparency would make shared templates feel more trustworthy and easier to understand.
Thank you for considering it.