Refine a character image using GPT Image 2. Upload one source photo, write a prompt describing what to keep and change. The output is a new AI generation based on your instructions.
Refine an existing character image using GPT Image 2. Upload the source photo, describe what should remain and what should change in your prompt. Output is a new generated image, not a pixel edit.
Select one clear reference image of your character. This will be the starting point for the new generation.
In your prompt, state which elements to keep (e.g., armor, pose) and which to change (e.g., expression, lighting). Framing or crop requests must be written as words.
Inspect the new image for consistency with your source and prompt. Adjust the prompt if needed and regenerate.
Keep the female warrior in silver armor, sword raised, and the misty forest background. Change her expression from angry to a calm, focused stare. Maintain the same lighting and camera angle.
Confirm the armor, sword, and forest are unchanged, and the new expression is clearly visible.
Describe the exact expression, e.g., 'calm, focused stare with relaxed eyebrows'.
Yes, write 'keep the warrior, change forest to a dark dungeon' in your prompt.
Describe the sword in detail in your prompt, as the model may reinterpret it.
