Skip to content
Edit Text
  • Change a date in an imageMove a schedule, flyer or timestamp forwardEdit numbers and pricesPrices, quantities, times and phone numbersEdit text in a screenshotReplace labels, messages and UI copy in a finished screen
  • Remove text from image
  • Pricing
Checking…
  • Dates
  • Numbers
  • Screenshots
  • Remove text
  • Pricing
  1. Home
  2. /Blog
  3. /
  4. How-to

How-to

How to Match a Font in an Image (and What Cannot Be Recovered)

Matching a font in an image is two jobs, not one. The first is identifying which typeface the picture uses. The second is replacing the words so the new lettering still sits on the same baseline, at the same weight, with the same spacing. Most guides stop at the first job and imply the second one comes free. It does not: a JPG or PNG stores pixels, not a font file, so a typeface can be identified and still not be reusable.

October 8, 2026·9 min read

Quick answer

To replace text while keeping the original look, upload the image to an editor that treats the existing lettering as its reference. It reads the shape, weight, spacing and position of the words already in the picture, then renders the replacement to sit in the same place. You do not need to know the font name, and you cannot install the font - only the visible lettering is matched.

Why matching a font is harder than the name suggests

In a layered design file, matching a font means installing the same typeface and typing into it. That is not available to you the moment the artwork becomes a screenshot, a photo or a flat export: the words are baked into the pixels and there is no text object left to select.

Something narrower and still useful is available. Instead of reproducing the typeface, the replacement can reproduce how the existing lettering looks, so a swapped word does not break the line it sits in. That means matching four visible properties rather than a font name.

  • Shape - the letterforms themselves: how rounded, how condensed, how much contrast between the thick and the thin strokes.
  • Weight - how heavy the strokes read against the surface behind them.
  • Spacing - the gaps between letters and between words, which is what makes a replacement feel too loose or too tight.
  • Position - size, baseline and alignment, so the new word sits where the old one did.

What font identifiers are actually comparing

A font identifier does not read a name out of the picture. It measures the letterforms - stroke thickness, the angle of the terminals, how open or closed the counters are, the width of the stems - and compares those measurements against a catalogue of hundreds of thousands of faces. What comes back is a ranked list of faces whose geometry is closest to yours.

Three things follow from that. The catalogue matters more than the algorithm: a face that is not in it cannot be returned, and smaller foundry releases often are not in it. A shortlist is the honest output, so the top result is the closest match found rather than a verified identity. And a blurred, curved or low-resolution sample degrades every measurement at once, which is why the same identifier can be confident on one crop and useless two inches to the left.

A lookalike is not a copy

Identifiers frequently return a near-identical free or system face instead of the original commercial one. That is often good enough for a mockup and never good enough for a brand guide, so treat the suggested name as a lead to verify rather than as an answer.

How to identify the font in an image

If what you need is the name - for a brand guide, a redesign or a licence purchase - identification is largely a solved problem. The usual route:

  1. 01

    Crop to one clean line of text

    Crop tightly around a single line of large, straight-on, well-lit lettering. Identifiers work on letterform shapes, so perspective, blur and a busy background all lower the quality of the suggestion.

  2. 02

    Run it through a font identifier

    Online identifiers such as WhatTheFont, Fontspring Matcherator or the text tools built into Google Lens compare your letterforms against a catalogue and return the closest matches. Expect a shortlist, not a single verified answer.

  3. 03

    Cross-check the glyphs that give it away

    Look at the lowercase g, the leg of the uppercase R, the shape of the numerals and any ligature. Two typefaces can share a family resemblance and differ in exactly one of these.

  4. 04

    Confirm on the foundry page

    Before the name goes into anything public, confirm it against the foundry listing. An identifier returns its best guess, and a guess that reaches a brand guide is expensive to unwind.

Where identification stops

Knowing the name does not let you reuse the font. A raster image does not contain the file, so a licence that covers the typeface does not transfer to the picture, and an editor cannot install a font it was never given. Identification answers a naming question. It does not answer a rendering one.

How to replace the text so the lettering still matches

  1. 01

    Keep the replacement about as long as the original

    Short swaps land closest. Replacing a short word with a long phrase changes the spacing of the whole line, which is a layout change rather than a font change, and no amount of letterform matching will hide it.

  2. 02

    Upload the image and mark the words to change

    Marking matters more than it sounds: the editor locates the text before it generates, so it knows which pixels are in scope and which are not.

  3. 03

    Type the replacement exactly as it should read

    Include the capitalisation and the punctuation. A trailing period or a changed case is the most common reason a result looks wrong at a glance.

  4. 04

    Review the changed area at full zoom

    Check the stroke weight against the words next to it, the gap after the new word, and whether the baseline of the line is still straight. This takes seconds and catches almost everything.

  5. 05

    Re-run it if the new word stands out

    A replacement that looks too sharp or too heavy next to the original lettering is usually a rendering difference, not the wrong typeface. Running the same edit again costs nothing else, because nothing outside the words moves.

When the match holds and when it does not

Honest expectations, from the cases that come out closest to the ones that need another attempt:

How closely the replacement lettering tends to track the original

CaseHow close it lands
Short words on flat surfaces - a label, a button, a plain posterClosest. There is little texture behind the text to compete with.
Ordinary sans-serif and serif faces at a normal sizeClose. Shape and spacing are easy to read from the image.
Curved, slanted or decorative display letteringLess close. The replacement is rebuilt from a description of the shapes rather than from the shapes themselves.
Text over a busy texture or a photographThe lettering can match while the surface around it shifts slightly.
Very small type, or words sitting over a gradientThe hardest case. Fine strokes lose their edges first.

Have an image where one word has to change?

Replace text in an imageSee the same-font workflow

A short checklist before you download

  • Is the new word spelled exactly as it should be, including case and punctuation?
  • Does the replacement read at the same weight as the words around it?
  • Is the space before and after the new word the same as the original spacing?
  • Is the line still straight, or has the replacement pushed the words beside it?
  • Has anything next to the text changed - an icon, a logo, a border or a background edge?

One thing that is not reversible

Once a word has been replaced in the image, the original lettering is gone from the result. Keep your source file, and check the changed area before you publish rather than after.

If you still have the design file, use it

Everything above describes the case where a flat image is all that is left. When the layered source still exists - a PSD, a Figma file, an Illustrator document - the original font is already inside it, and the correct move is to open the file, keep the typeface, and edit the text object directly. Replacing pixels is a repair for artwork that has already been flattened, not a replacement for a file you still have.

Frequently asked questions

Can I match a font in an image with a free tool?

You can identify the typeface with a free online identifier, but identification and matching are two different jobs. Replacing the words so the lettering still looks right can be done in the browser without installing anything, because the editor matches the visible lettering rather than the font file.

What font is used in my image?

Crop to one clear line and run it through an identifier such as WhatTheFont or Fontspring Matcherator. These return a best guess from a catalogue, so check the lowercase g, the numerals and any single distinctive glyph before you rely on the name.

Does replacing text keep the original font size and spacing?

The workflow measures the existing lettering and aims to leave size, weight and spacing where they were. Short replacements land closest. A replacement much longer than the original changes the spacing of the line, and that is a layout change rather than a font change.

Can the exact font file be recovered from a JPG or PNG?

No. A raster image stores pixels, not outlines or font metadata, so a typeface can be identified but not restored as a file. A tool that offers you the font from a screenshot is returning a lookalike suggestion, not the original the artwork was set in.

Why does my replaced word look slightly different from the rest?

Usually a sharpness or weight mismatch rather than the wrong typeface. Generators redraw the text at a higher resolution and scale it back down, which can leave the new word slightly crisper or heavier than the words around it. If it stands out, run the edit again and compare the area at full zoom.

Related

Keep reading

  • Edit text in an image with the same font
  • Remove text from an image
  • Canva vs Photoshop vs an AI image text editor
Edit Text

Replace words already inside a finished image, compare the result with the original and download the edited file.

Contact ussupport@rewordimage.com

Tools

  • Edit text in image
  • Edit text in a screenshot
  • Edit text with the same font
  • Change a date in an image
  • Edit numbers and prices
  • Remove text from image

Site

  • Pricing
  • Blog
Reword Image
Launched on FazierFeatured on Twelve ToolsFeatured on Findly.toolsListed on ToolPilot.aiListed on Dang.aiMossAI ToolsSubmit AI Tools badgeFeatured on Wired BusinessFeatured on TinyShelf

© 2026. All rights reserved.

Acceptable usePrivacy policyTerms of serviceRefund policyAbout