Quotes are extracted from the original book text (no inventions)

Meet your favorite characters with VisuLit — 5 free generations daily

VisuLit extracts character lists and appearance quotes from public-domain books, then builds photorealistic portrait prompts you can reuse, tweak, and regenerate in seconds.

Free to try · 5 portrait generations per day — no credit card · No signup required

Find characters

Start with one book, let VisuLit prepare the cast

Search by title, pick from our curated list, then prepare the book. If a character is missing, type the name and we'll add them by searching the book text.

40+ curated picks · Extended catalog matches when your title isn't on the shortlist · Quote-grounded portraits

Book search

No books returned. Check your connection and refresh the page.

Cast

Select a book above to load the cast.

Missing a character?

Didn't find someone in the cast?

Type the name as in the book. We search the text and add them to the cast.

Your own character

Create a branded VisuLit prompt in 30 seconds

Use a guided form (hair, eyes, era, clothing) to generate a clean prompt in the VisuLit style. Your prompt stays on this page for easy reuse and fast iteration.

Prompt

Name

Edit / extend

Your generated prompt and portrait are ready for reuse.

Portrait gallery

Photoreal examples from VisuLit — the same quote-first flow: prepare a book, pick a character, generate.

Elizabeth Bennet·1 of 7

History

No generations yet.

FAQ

Why these quotes?

We prioritize lines with stable visual details: face, hair, eyes, clothing, figure, and color terms.

Why does the portrait differ from my imagination?

The prompt is quote-first. When source text is sparse, the model fills gaps with neutral, era-consistent details.

Can I regenerate?

Yes, use the Regenerate button under each portrait card for a new variation.

Can I change scene but keep the same face?

Use New scene (same character) under a portrait: describe emotion, pose, and place. We send your current image as a reference and stress identity in the prompt. Results vary by model; perfect likeness is not guaranteed without a provider that accepts reference images (see server env IMAGE_REFERENCE_BASE64_FIELD).