Bài viết này hiện chỉ có bằng tiếng Anh.

WebPMar 3, 20266 min read

WebP Compression Artifacts: How to Spot the Bad Ones

Compression artifacts are not all equally important. A small change in a background texture may be invisible in the final page. A halo around product packaging, fuzzy UI text, or banding in a hero gradient can damage trust quickly. WebP quality review should focus on artifacts that users can see and that change the meaning or polish of the image.

A useful QA process teaches reviewers what to look for before they approve a batch.

Screenshots of random artifacts are not useful without a threshold, source file, and production decision. A useful artifact review names the source set, the conversion settings, the page context, and the action taken.

Know the Common Artifact Types#

WebP can be used in lossy or lossless modes, and Google's WebP documentation explains the format background. In practical review, the common concerns are visual:

  • blocky or smeared detail
  • ringing around sharp edges
  • banding in gradients
  • color shifts in product images
  • muddy texture in fabric, hair, or foliage
  • fuzzy small text in screenshots
  • noisy transparent edges
  • flicker in animated images

Not every artifact is a blocker. The decision depends on image role and visibility.

Inspect the Hard Areas First#

Do not judge quality from the center of an easy photo. Compression problems often appear in specific areas:

  • skies and smooth gradients
  • dark shadows
  • skin tones
  • product labels
  • thin lines
  • transparent edges
  • fine repeating patterns
  • UI text and icons

Zoom in enough to understand the problem, then zoom back to the actual display size. A defect that is only visible at extreme zoom may not matter. A defect visible at normal page size should be addressed.

Compare Against the Right Source#

Compare the WebP output against the best available source, not against a thumbnail or already-compressed derivative. If the source is a low-quality JPEG, the WebP output may inherit old artifacts. That does not mean the new conversion created them.

Keep a side-by-side review:

source export
current published image
new WebP output
rendered page view

This helps the team decide whether to re-export from a better source, adjust settings, or keep the existing file.

Record the Conversion Settings#

When quality is being challenged, the team needs the exact settings, not just a comment that the image "looks bad." GetWebP's CLI command reference documents the --quality, --format, --output, --recursive, and --json flags, and the original files are not modified or deleted.

For a repeatable review, write outputs into a separate folder:

npx -y getwebp ./artifact-sample --format webp --quality 85 --output ./artifact-review/q85 --recursive --json

Then read the NDJSON output and keep the per-file outputPath, originalSize, newSize, saved, quality, qualityMode, and status fields with the QA note. That makes the artifact report auditable instead of subjective.

Last reviewed: 2026-03-03
Source: product-label-hero.jpg
Output: ./artifact-review/q85/product-label-hero.webp
CLI record: quality=85, qualityMode=auto, saved=31.4%, status=success
Page context: PDP hero, desktop 1440px and mobile 390px
Finding: label text has visible edge ringing at normal mobile size
Action: raise quality for product-label images and rerun page-level QA
Owner: ecommerce content lead

If the output is larger than the original or the visual issue remains after raising quality, the correct decision may be to keep the original for that asset rather than forcing WebP everywhere.

Check Screenshots Differently#

Screenshots fail differently from photos. The biggest issues are fuzzy text, color bleeding around UI lines, and loss of clarity in small controls. A setting that works for photos can be poor for interface captures.

For screenshots:

  • inspect text at the final rendered size
  • check charts and grid lines
  • review light and dark mode variants
  • avoid unnecessary resizing after conversion
  • consider lossless or lighter compression

If a screenshot teaches a workflow, readability matters more than maximum savings.

For browser-extension users, the quality settings guide gives a simple starting range, but it should not replace review. A screenshot with 10 px interface text may need a different setting from a full-width photo even when both are exported as WebP.

Check Product Photos by Purchase Detail#

For product images, ask what the shopper needs to inspect. A shoe buyer may care about stitching. A furniture buyer may care about wood grain. A cosmetics buyer may care about color. A hardware buyer may care about ports and labels.

Artifacts that hide purchase details are important even if they affect only a small part of the image. Review product images in the gallery, zoom view, and variant view.

If the image changes how the product appears, the setting is too aggressive for that role.

Watch Transparent Edges#

Transparent graphics can show artifacts around edges, especially when lossy compression is used. Test logos, badges, and cutout products on multiple backgrounds: light, dark, colored, and patterned.

MDN's image file type guide is useful when deciding whether a file should remain PNG, move to lossless WebP, or be handled as SVG.

If a brand mark gains a visible outline, keep a cleaner format or use lossless settings.

The output formats guide is the right place to anchor this decision: lossy WebP is useful for many photos, while lossless output is the safer review path when graphics, icons, or brand marks need maximum fidelity.

Do a Page-Level Review#

File-level review does not cover how visitors see the asset. The page may crop the image, place text over it, scale it, or show it next to related assets. Compression can become more visible in those contexts.

Check:

  • mobile crop
  • desktop crop
  • card thumbnail
  • hero placement
  • dark-mode background
  • hover or zoom state
  • adjacent images in a gallery

Images should be approved where users see them.

Use a small blocker matrix so reviewers apply the same standard:

Image roleBlockerUsually acceptable
Product detailcolor shift, hidden label text, damaged textureminor background smoothing
UI screenshotfuzzy instructional text, blurred controls, unreadable chart labelstiny compression noise outside the instructional area
Brand graphichalo around logo, noisy transparent edge, wrong colormodest file-size increase from lossless output
Editorial photobanding in visible sky or skin-tone damagetexture loss that is invisible at final crop
Thumbnailartifact visible in listing griddefects visible only at extreme zoom

This matrix gives editors a basis for rejecting an output without turning every tiny pixel difference into a blocker.

Decide With a Clear Action#

When an artifact is found, choose an action:

accept: not visible in final context
raise quality: visible but fixable
use lossless: sharp graphic or text needs protection
resize first: dimensions are the real problem
replace source: input is already damaged
keep original: WebP tradeoff is not worth it

This turns QA notes into production decisions.

For WordPress batches, pair this visual review with a delivery check. GetWebP's batch conversion documentation describes presets, re-optimization, and deletion of generated variants, while frontend delivery explains how the converted file is served on the page. A clean file in the media library still needs to look right after the theme crops it, lazy-loads it, or places it on a dark background.

WebP artifacts matter when they change clarity, trust, or brand quality. Review the hard areas, compare against the right source, and judge the image in its real page context before approving the output.

Jack avatar

Jack

GetWebP Editor

Jack writes GetWebP guides about local-first image conversion, WebP workflows, browser compatibility, and practical performance checks for teams that publish images on the web.