Does lossless WebP preserve the RGB of transparent pixels?

The original control

We authored a 240 × 120 RGBA PNG. The left half stored pink RGB (198, 52, 122) with alpha 0; the right half stored teal (21, 127, 124) with alpha 255. We encoded that same input twice with Pillow: lossless=True alone, then lossless=True with exact=True. No photograph, palette reduction or image resizing was involved.

14,400 hidden RGBA pixels changed by default — Results from our authored fixture and the exact comparison described below.

What the saved files measured

The default WebP decoded its left sample as (0, 0, 0, 0). The exact WebP retained (198, 52, 122, 0). A full-grid comparison counted 14,400 changed RGBA pixels in the default file and zero in the exact file. Both retained the opaque teal sample. The saved default and exact files measured 50 and 52 bytes in this run.

Original test file	Detected format / mode	Dimensions	Bytes	Distinct decoded RGBA values
hidden-rgb-source.png	PNG / RGBA	240 × 120	448	2
lossless-default.webp	WEBP / RGBA	240 × 120	50	2
lossless-exact.webp	WEBP / RGBA	240 × 120	52	2

These are source-file measurements made with Python 3.12.14, Pillow 12.3.0 and libwebp 1.6.0. The manifest records complete file hashes, decoded RGBA hashes and the relevant coordinates. File sizes are measured bytes for these inputs, not forecasts for other images.

How to interpret this comparison

We then composited all three decoded inputs over solid white and over solid black. The complete composite grids matched within each background. That control separates hidden-channel preservation from visible equality. It explains how a file can satisfy this visible comparison while failing an RGBA byte comparison. If hidden RGB drives texture processing, record that requirement explicitly rather than relying on the word lossless.

Original experiment file previews on white, with detected formats and dimensions — These are previews of authored test files, composited on white for the diagram where necessary. They are not screenshots of a browser test.

Repeat this exact test

Run the supplied experiment and compare the RGBA samples and full-grid hashes in measurements.json.
Inspect the two composites_equal entries. They test each complete composite, not a screenshot.
Keep exact=True in a WebP export when this specific hidden-data requirement applies; retain the original PNG as the reference.

Download and unpack the ZIP. With Python and Pillow installed, run the command below. Compare the generated measurements.json with the included report. The script regenerates fixtures and computes the measurements; it does not obtain values from this article.

python reproduce.py --case lossless-webp-transparent-rgb --out rerun

Scope and limits

This is a local encoder and decoder experiment, not a claim that a browser color picker can recover invisible RGB. Alpha-zero pixels have no visible color in these composites. Other encoder versions can produce different file sizes; the manifest records the actual runtime.

Our own sample is a diagnostic control. Applying its conclusion to another file requires checking that file and the operations performed on it. We have not measured a global error rate or established that an automatic palette is an exhaustive list of image colors.

Did the visible teal change?

No. The opaque control decoded to the same RGB in all three files. The changed samples were in the fully transparent half.

Watch the source-file comparison

Original diagrams and synthetic English narration. Caption sentence timings are approximate.