Image To Text🔒 100% local
Extract image metadata as text — dimensions, aspect ratio, transparency, dominant colors, ready-to-use embed code. Runs in your browser, no uploads.
Image → text report
Analyzes the image in-browser: dimensions, palette, brightness, transparency, and generates Markdown/HTML embed text.
Reads an image you choose from disk, draws it to an offscreen canvas and reports its metadata as text: pixel dimensions, megapixels, aspect ratio and orientation, the transparency share, average brightness, a count of high-contrast edges and the five most common colours. It also writes Markdown and HTML embed tags for a file of that name. The browser decodes the file locally; there is no OCR and no upload.
How to use it
- Choose an image file with the picker.
- Press Analyze so the browser decodes it into a canvas.
- Read the report top to bottom: dimensions, palette, brightness and the embed snippets.
- Copy Output to take the whole report as text.
Worked example
a 100x50 solid red PNG named 'solid red.png' (143 bytes)becomesFile: solid red.png (0.1 KB)
Format: image/png
Dimensions: 100 × 50 px (0.01 MP)
Aspect ratio: 2.000 (landscape)
Transparency: none
Avg brightness: 54/255 (dark image)
High-contrast edges (400×400 sample): 0
Dominant colors:
1. #FF0000
Markdown embed:
HTML embed:
<img src="/images/solid-red.png" width="100" height="50" alt="" loading="lazy" />Verified against a generated 100x50 pure-red PNG: brightness in this tool is 0.2126R + 0.7152G + 0.0722B, which for pure red is 54, and a flat-colour image has no edges. The palette and edge count depend on the file you supply.
When to use it
- You need the exact pixel size and aspect ratio of a screenshot before dropping it into a layout.
- You want a rough palette check of a logo or photo without opening an image editor.
- You need ready-made Markdown or HTML embed tags carrying the correct width and height attributes.
What it does not do
- It does not read text out of an image, despite the tool name - there is no OCR, no handwriting recognition and no scene-text detection.
- Dominant colours are quantised to four bits per channel, so near shades merge into a single bucket and the averaged value can be a couple of steps off the exact colour.
- The edge counter samples at most a 400x400 block from the top-left corner, so it is a rough signal rather than an edge map of the whole picture.
- The transparency figure counts any pixel with alpha below 0.5 across the entire image, so a soft drop shadow or a faint overlay inflates it.