AI / Design Communication & Image Prompting
Reverse-Engineer a Prompt from a Reference Image
Turn an image you like into a reproducible prompt, not just a saved link
This is a reference image I found: [image description, or attach the image]. Reverse-engineer a prompt that could reproduce this style: 1) Break down the composition — subject and materials, light direction and color temperature, lens and depth of field, where the visual weight sits; 2) Extract style cues — brushwork or texture, the closest genre and medium, traces of post-processing; 3) Write one English prompt in the order subject → environment → style → lighting → camera → parameters; 4) List the 3 features most likely to come out wrong, with matching negative prompts; 5) Give an acceptance checklist: which features must be present for the result to count as a successful reproduction. Do not just stack adjectives.
How to use
When you swap the reference, rerun only the first three steps and leave the prompt structure alone — that makes different styles easy to compare side by side.
Good for
- You found an image whose style is exactly right and need to reproduce it on your own material (same style, different subject) — that requires breaking down structure, not copying content: light, material, and lens first, subject second.
- When you are building a reusable style baseline for a team — once the prompt structure is fixed, swapping the reference means rerunning only the first three steps, so different styles stay comparable.
Not for
- The reference is a work you have no rights to and the output is commercial — style can be learned, but the specific composition and characters cannot be copied; settle the legal risk first.
- You just need one usable image and do not need reproducibility — direct generation plus inpainting is faster; reverse-engineering a prompt pays off only when you need a batch.
Example
Input
A reference image: a white porcelain cup on a dark ground, hard light from the left casting a crisp shadow, deep forest-green wall behind, slight film grain
What you get
Breaks down the composition: single subject slightly below center, hard light at 45 degrees from upper left, cool color temperature, shallow depth of field; style cues are still-life photography, low-key lighting, Portra-like grain. Writes it as a six-section English prompt and lists the three features most likely to come out wrong (shadow edge too hard, green wall collapsing to black, blown highlight on the cup rim) with matching negatives. The acceptance checklist is checkable: the shadow edge has a gradient and the background still reads as green.
Common mistakes
- Writing the reference's content (that cup, that character) into the prompt — the style falls apart the moment you swap the subject; keep only lighting, material, and lens language.
- Asking only for a string of adjectives (moody, cinematic, film grain) — those hold for any image and reproduce nothing specific to this one.
Try tweaking it like this
- Replace 'break down the composition' with 'first decide which common style this image is least like' — elimination often locates the style coordinates faster than positive description.
- Ask for 'three parameter variants: conservative, close, and aggressive' — you get three levels of reproduction fidelity in one pass and pick by use case.