Prompts / art_direction_brief
Art direction brief for AI image generation Free
Writes image prompts and a style guide that produce a coherent set of AI-generated images with a point of view, avoiding the stock-illustration and glowing-blob look.
Run it
- Claude Code:
/mcp__designers__art_direction_brief - Claude Desktop and claude.ai: plus menu, Add from Designers, Art direction brief for AI image generation
- Any app: Use the art_direction_brief prompt from the Designers server on what the images are for.
Arguments
| Name | Meaning |
|---|---|
project * | What the images are for |
mood | Feel, references, what to avoid |
count | How many images you need |
* required
What your assistant is told to do
The full instructions, so you can see exactly what you are asking for:
Write an art direction brief for generated images that look like one considered set.
Project: <project>
Mood and references: <mood>
Count: <count>
1. Read resource designers://reference/slop-tells for the image tells to avoid (isometric blobs, glowing gradients, floating dashboards, stock people at laptops, perfect symmetry, the same lighting on everything).
2. Choose one mode and commit: documentary-style photography (real subjects, available light, specific lens and framing), or collage (cut-outs, mixed scale, torn edges, type interacting with image), or a defined illustration system. Do not mix.
3. Write the style guide: subject matter rules, camera and framing (focal length, angle, crop, aspect ratio per placement), lighting, palette (hex values tied to the brand and checked for contrast with overlaid text via check_contrast), texture and grain, what is always in frame, what is never in frame, and how people are depicted (diversity of age, body and setting, no stock poses).
4. Write the image prompts: each names subject, setting, action, framing and aspect ratio, lighting, palette and mood in concrete terms, plus negative constraints. Write them as a photo editor would brief a photographer. Vary the framing across the set (wide, detail, mid, close) so it does not read as copies.
5. Add review criteria for choosing results: does it match the guide, are hands and text correct, does it read at small size, does it work with the text overlay, is it different from the others?
6. Add alt text guidance: describe what the image conveys in context, not "image of".
Provide a shot list table: id, placement, aspect ratio, prompt, alt text, overlay text and contrast check.