| name | soft-objects |
| description | Generate coherent Soft Objects editorial illustration sets using soft 3d with optimistic visual direction. Use when illustrating articles, blogs, newsletters, essays, or social posts in this named style, or when the user asks for reusable, consistent visuals matching Soft Objects. |
Soft Objects
Create a coherent set of article illustrations using the Soft Objects visual system. Preserve the style grammar across the set while inventing a fresh, article-specific composition for every image.
Read references/style-guide.md before planning or generating. Use the images in assets/examples/ only for low-frequency visual calibration; never copy their composition, people, objects, or wording.
Workflow
- Read the full article or supplied passage.
- Identify the core argument, contrasts, turning points, system relationships, and conclusion.
- Select 1-9 cognitive anchors. Do not illustrate paragraphs at even intervals.
- Write a shot plan before generating. Give each image one core idea, one structure, and a different primary composition.
- Compile every shot with the visual DNA and negative constraints from the style guide.
- Use the host agent's available image-generation tool. Generate one standalone 16:9 image per request.
- Inspect every output against the QA checklist. Regenerate or make one targeted edit when a failure signal appears.
- Deliver ordered files and preserve original generated files.
Shot plan
For every proposed image, specify:
- Placement after the relevant passage
- Theme and core idea
- Purpose in the article
- Structure type
- Composition and reading direction
- Suggested elements
- Essential short labels, or none
Prefer:
