FIELD GUIDE / LEONARDO AI

Leonardo AI Image Guidance: Style, Content and Character References

A reference image can influence different things: appearance, layout or a recurring subject. Leonardo’s official Image Guidance API documentation separates these tasks, which helps designers avoid asking a style reference to solve a.

A reference image can influence different things: appearance, layout or a recurring subject. Leonardo’s official Image Guidance API documentation separates these tasks, which helps designers avoid asking a style reference to solve a character-identity problem.

Three different constraints

Leonardo documents Style Reference, Character Reference and Content Reference. Style reference seeks a shared aesthetic; character reference emphasizes resemblance to a subject; content reference aims to influence structure. The company explicitly says character reference is not intended for face swapping and does not guarantee a perfect human replica. That limitation should be quoted accurately when planning multi-scene storytelling or branded mascots.

Version-level compatibility

The API reference uses different guidance/preprocessor IDs and model-specific capability tables. A preprocessor that works with one base model need not work with another. Before automating a workflow, register the base model identifier, guidance type, supported strength values, width and height and source image type. An example snippet that happens to contain an old model ID is documentation, not proof that an arbitrary current account can use it.

How to conduct an evaluation

Hold one aspect constant per batch. For illustration, use the same prompt and character source across five scenes to inspect hair shape, color, clothing geometry and accessory position. For product design, inspect the product silhouette and logos more strictly. Rate failure types separately: incorrect identity, style drift, object addition, misspelled text and frame cropping. Do not declare “perfect consistency” from one aesthetically attractive sample.

What remains outside the model

Consent to use a person’s likeness, approval to upload a client asset, enterprise retention and regional restrictions are procurement and legal questions. This source does not verify global availability or current price. For brand-critical subjects, prefer direct compositing or editable vectors for fixed marks. Record image reference permissions and hand a versioned score sheet to the next designer.

Worked scenario and failure test

A publisher has an original children’s-book character and must illustrate eight pages. First try a Character Reference for recurring face and costume, then use a separate Style Reference for watercolor appearance; do not assume the same slider provides both effects. Compare the character’s collar, ear shape and eye color and accept page layouts only after a continuity review. The official API example includes model-dependent preprocessor identifiers, so production code must pin the exact supported base-model pairing. The source’s own warning against assuming a perfect face replica belongs in the editor checklist.

Decision guide and practical handoff

Choose a specific reference type and base-model combination; do not promise exact likeness. A responsible review packet names the exact source or model version, the permitted input assets, the requested operation and the image characteristics a reviewer must check. It should also contain an unaltered reference, the accepted derivative, and a short record of what changed. A supported feature is not necessarily a reliable one for every brief, and the team must approve both technical appearance and real-world truthfulness before distribution.

Availability, price and geographic limitations

Research cutoff: 25 October 2026. Release dates and historical tier details above are explicitly dated. Unless independently confirmed for a named country and account tier, current availability, billing currency, credit consumption, API quota and commercial entitlement are unknown—not universal. Consult official regional terms and AI 10’s dated price snapshots before procurement; do not treat a historical announcement as a live entitlement. No application, model, API response or image output was hands-on tested for this article.

Sources & Further Reading

CONTINUE EXPLORINGExplore related files ↗