01
Source image
A storefront photo reading 'OLD TOWN BOOKS' in condensed cream lettering, painted onto a weathered green sign at a slight upward camera angle.
Typography-preserving replacement workflow
Change the wording while keeping the lettering visually consistent with the source. Describe the exact replacement and protect the font style, weight, texture, spacing, perspective, and shadows.
Worked example
Use this runnable brief for a photographed bookstore sign. It makes the typography constraints measurable and does not claim that the general FireRed showcase below is this exact output.
01
A storefront photo reading 'OLD TOWN BOOKS' in condensed cream lettering, painted onto a weathered green sign at a slight upward camera angle.
02
Replace the headline with 'RIVER CITY BOOKS'. Retain the condensed uppercase style, weight, color, worn paint texture, tracking, baseline, and perspective.
03
Confirm the new words are exact, lettering feels native to the sign, shadows follow the surface, and brickwork, windows, reflections, and people are unchanged.
“Replace 'OLD TOWN BOOKS' with 'RIVER CITY BOOKS' on the storefront sign. Change only those words. Preserve the original condensed uppercase font style, weight, cream color, worn paint texture, letter spacing, baseline, size, perspective, edge softness, lighting, and shadows. Keep every other part of the photo unchanged.”
A convincing text replacement depends on more than finding a similar typeface. The lettering must also inherit the source treatment and the physical conditions of the scene.
Describe serif or sans serif structure, width, capitalization, stroke weight, and distinctive display characteristics visible in the source.
Preserve fill color, outline, bevel, paint wear, foil, glow, grain, or printed texture so the replacement matches the existing production method.
Keep size, spacing, baseline, perspective, curvature, occlusion, shadows, and reflections aligned with the original surface and camera view.
FireRedTeam's official showcase includes stylized text reference across posters, photographed signs, and illustrations. It demonstrates the broader capability behind typography-aware replacement, not the exact bookstore example above.

Decorative and dimensional examples show why a replacement prompt should name more than the words and nominal font category.
Photographed text demonstrates the importance of surface, angle, lighting, and environmental context.
The showcase is published with FireRedTeam's open-source project materials rather than presented as an unverified user result.
Use an exact replacement instruction, name the visible typography traits, and validate the generated lettering at full resolution.
Note case, width, weight, color, outline, texture, spacing, perspective, and shadows. Start with a clear image or crop when the words are small.
State exactly what to replace and what it should say. Add 'change only these words' followed by the typography and scene invariants.
Check every character, then compare width, baseline, tracking, texture, perspective, and lighting. Narrow the next prompt to the remaining mismatch.
Use this workflow when a flat text overlay would break the material, typography, or perspective already embedded in an image.
Update names, hours, or event wording on photographed surfaces while preserving paint, wear, viewing angle, and natural light.
Swap a headline or date while keeping decorative lettering, visual hierarchy, effects, and surrounding artwork consistent.
Test alternative product names while retaining print texture, material highlights, label geometry, and studio shadows for internal concepts.
Use precise visual constraints instead of asking only for the same font. Replace the quoted examples with the wording and traits in your source.
“Replace 'OPEN DAILY' with 'OPEN LATE'. Preserve the original bold condensed lettering, painted texture, cream color, spacing, perspective, weathering, and cast shadow.”
“Replace the headline 'SUMMER NIGHTS' with 'CITY LIGHTS'. Keep the same decorative serif style, letter weight, gold foil texture, size, alignment, glow, and background artwork.”
“Replace only 'COFEE' with 'COFFEE'. Match every neighboring letter's font style, height, stroke width, color, grain, baseline, spacing, and lighting.”
Font-preserving text can look right at thumbnail size while failing under inspection. Review both the word and the way it occupies the image.
Verify spelling, punctuation, repeated letters, counters, serifs, and unusual glyphs. Do not approve a replacement based only on its overall silhouette.
Compare cap height, width, weight, tracking, baseline, color, outline, texture, and visual hierarchy with the source lettering.
Inspect edges, perspective, curvature, occlusion, reflections, material texture, and shadows for signs that the new text was pasted onto the scene.
This page focuses on same-language or approved-wording replacement where typography continuity is the primary constraint. A precise scope keeps the prompt and review criteria useful.
Choose this workflow when the new words must appear to use the original lettering style and physical treatment.
The edit references visible letterforms and styling from the image, but it does not identify or deliver a licensed font file.
Translation, unrestricted rewrites, typo repair, and other text tasks are covered by the main AI image text-editing workflow.
Answers about matching typography, photographed surfaces, exact fonts, prompt structure, and production review.
It can reference visible lettering and reproduce a similar style while changing the words. Specify the traits to preserve and compare the result closely with the source.
No. Describe visible traits such as condensed sans serif, bold weight, uppercase letters, cream paint, worn texture, and narrow spacing. A font name can help but is not required.
No exact font-file retrieval or licensing is implied. The model uses the image as visual context, so treat the output as style matching and inspect distinctive letters carefully.
You can request the same perspective, baseline, curvature, and surface alignment. Clear source detail and an explicit prompt make those constraints easier to follow.
Include each effect in the prompt. Ask to preserve outline thickness, bevel direction, glow, cast shadow, highlights, and their relationship to the scene lighting.
Longer wording may force smaller letters or tighter spacing. Provide concise copy and state which boundaries, hierarchy, and minimum size matter most.
Yes. Quote the incorrect and corrected word, say to change only that word, and ask the model to match neighboring letters in every typography and lighting detail.
Use a sharp, high-resolution source with readable letters. Crop closer when text is small, blurred, heavily occluded, or occupies only a tiny part of the image.
It can attempt handwritten and decorative styles. Review repeated letter shapes, stroke rhythm, joins, spacing, and whether the result still feels naturally hand-made.
The instruction should say to change only the named words and preserve all non-text details. Always compare objects, textures, faces, and reflections against the original.
Technically similar edits may be possible, but you are responsible for trademark, copyright, licensing, and brand-approval requirements for the source and result.
Enter the exact old and new wording above, list the visible font and surface traits to preserve, open the editor, upload the image, and inspect the result at full size.
Carry a constraint-rich prompt into FireRed and review the result for exact text, typography continuity, and scene integrity.
Name the old text and approved replacement.
Protect weight, spacing, texture, geometry, and light.
Check every glyph and the surrounding pixels.