Style Roll example output generated by Doodle AI — A real run of this skill. The input is a synthetic portrait, not a customer photo.

The photo used as input for Style Roll — A real run of this skill. The input is a synthetic portrait, not a customer photo.

Ready to use

Style Roll

4 distinct art styles from one photo — pick your favourite

Use this skill

Your photo: Required
Output: 4 images, square 1:1
Category: packs
Cost: 4 credits

style-sampler
multi-style
pick-favourite

What it does

What you get back

Addresses the #1 pain point of single-roll AI art — you never know if you will like the result. Style Roll generates four materially different illustrations of the same person from one photo: bold felt-tip marker, waxy crayon, cross-hatched ink pen, and soft watercolour wash. Same person, same framing, four different media. Pick your favourite and run with it.

When the agent picks it

Use when the user wants to see their photo in multiple art styles to choose from, or asks for different drawing styles, a style sampler, or "give me options".

The full skill file

Every skill is one SKILL.md — this is the exact file the agent reads.

src/mastra/skills/style-roll/SKILL.md

Style Roll

Upload one photo, get four genuinely different hand-drawn styles of the same person — then pick the one you like best.

When this is the right skill

Pick this skill when the user wants:

Multiple art styles to choose from (“give me options”, “show me different styles”)
A style sampler or comparison of illustration media
To try different looks before committing to one
“I don’t know what style I want” or “surprise me with variety”

Do NOT use this for a single specific style (use normal, crayon, or mini-me), for multiple expressions of one style (use expression-sheet), or for seasonal variations (use seasonal-pack).

What to draw

Four separate 1:1 images from the same uploaded photo, each rendered in a materially different hand-drawn medium:

Marker — bold felt-tip on sketch paper, thick outlines, flat colour fills, directional hatching
Crayon — wax crayon on textured paper, soft waxy strokes, paper grain showing through
Ink Sketch — black pen cross-hatching on cream paper, no colour, pure linework depth
Watercolour — translucent washes on cold-pressed paper, bleeding edges, paper-white highlights

Character consistency is paramount: hairstyle, hair colour, face shape, skin tone, accessories, and outfit are identical across all four. Only the drawing medium changes — that IS the product.

How to run it

Confirm a photo is attached. If none is available, ask the user to attach one.
Call generateDoodle with skill: "style-roll" and the uploaded photo as imageUrl.
Pass refImageUrl when the message includes a separate style reference.
The tool returns FOUR separate images (one per medium). Cost: 4 credits (one generation per image).
Present all four and ask which style the user prefers.

Following up

Once the user picks a favourite style, offer to generate more images in that specific style (suggest the matching single-style skill if one exists), or offer to re-roll one of the four they didn’t like with a tweaked prompt.