Image Styles Starter - viral image style generator
What it does
Give it a style and a subject. It writes a finished image prompt you can paste into ChatGPT, Gemini, Midjourney, or any other image tool.
Try it: type paper-cut shadowbox: a lighthouse at dusk and get one prompt in a code block, plus a one-line fix for the most common failure of that style.
The three styles
Surreal Maximalism
Glossy, candy-coated 3D scenes packed edge to edge with tiny props.
Vintage Travel Poster
1930s flat-color lithograph poster with Art Deco lettering.
Paper-Cut Shadowbox
Layered cut paper in a wooden frame, lit from behind.
Each style was tested side by side on Seedream 5.0 Pro and Nano Banana 2. The images above came from the tested example prompts included in the skill, and the included gallery.pdf shows the image each example prompt produced.
Why the prompts work
- They describe how the image was made, not the genre. "Stone lithograph in six flat spot colors" changes the picture. "Vintage style" barely does.
- The look is locked. Each style ends with a tested Style paragraph that the skill never rewrites.
- You bring the scene. The skill fits your subject into the opening paragraph and adds a few fitting details.
What it does not do
- It writes prompts. It does not generate images.
- One style per prompt, square format.
- Three styles only. Seven more (tilt-shift diorama, bioluminescent ocean, hyperreal macro, retro anime VHS, dark fantasy oil, glass and chrome Y2K, cozy watercolor) are in the full Style Engine skill, also on Capafy.
AI disclosure and limits
This skill instructs an AI assistant to write image prompts. Results depend on the image model and resolution you use; for best results generate at about 2048 x 2048. Output can vary between runs.
