JPG vs PNG vs WebP: Which Format Wins

Formats • Compression • Transparency

Almost every "why does this image look bad" or "why is this page so slow" problem traces back to one decision made without thinking: which format to save in. The three that matter are JPG, PNG, and WebP, and they are good at genuinely different things. Once you know what each one throws away the choice becomes mechanical. For the click-by-click version, see how to convert image formats.

JPG ➝ WEBP

The one distinction that explains everything: lossy or lossless

Lossy compression permanently discards information the format's model considers visually unimportant, and gets small files in exchange. Lossless compression finds redundancy and encodes it more efficiently, so decompressing returns the original bytes exactly, and the files are bigger. JPG is lossy. PNG is lossless. WebP can be either, which is most of why it exists.

The corollary matters: quality lost to a lossy save never comes back. Converting a JPG to PNG does not restore detail, it just stores the damaged image losslessly and at a larger size.

JPG: photographs, and nothing else

JPG compresses by transforming blocks of pixels into frequency components and discarding the high-frequency ones people are least sensitive to. On continuous-tone imagery - skin, foliage, sky, fabric - that is close to invisible and the size reduction is enormous. On content with hard edges it fails visibly: text, logos, screenshots, and line art pick up a halo of ringing artefacts around every edge, because a sharp edge is exactly the high-frequency information JPG is designed to throw away.

JPG also has no alpha channel, so it cannot store transparency at all. And because every save re-encodes, repeatedly opening and saving a JPG degrades it a little each time. Edit from an original, export to JPG once.

Use it for: photographs, photographic backgrounds, email attachments, anywhere size matters and there is no transparency or text.

PNG: transparency, text, and pixel-exact graphics

PNG is lossless and carries a full 8-bit alpha channel, meaning each pixel can be fully opaque, fully transparent, or anything in between - which is what allows a soft shadow or an antialiased logo edge to sit cleanly on any background. Its compression works by finding runs and patterns, so it is brilliant on flat colour and terrible on photographic noise.

That is the whole tradeoff. A screenshot, an icon, a chart, or a logo saved as PNG is small and perfect. The same photograph saved as PNG can be five or ten times the size of the JPG and look identical.

Use it for: logos, icons, screenshots, diagrams, images with text, and anything needing transparency - including exports from the SVG to PNG converter and the rounded corners tool, where the transparency is the whole point.

WebP: the one that does both

WebP has a lossy mode derived from video keyframe coding and a lossless mode, and both support transparency and animation. That combination is unusual: it is the only one of the three that can give you a photograph with a soft transparent edge in a small file. At comparable perceived quality WebP typically produces meaningfully smaller files than JPG, and its lossless mode usually beats PNG on the same graphic.

The catch is reach rather than quality. All current browsers support WebP, but plenty of desktop software, older devices, print workflows, and third-party upload forms still do not. So WebP is the right answer for images you serve on the web and the wrong answer for images you email to a colleague or hand to a printer.

Use it for: production web images of every kind, where you control the delivery.

BMP: when uncompressed is the point

BMP stores raw pixel data with essentially no compression, so files are large and nothing is ever approximated. It is not a delivery format. It survives as an input for legacy Windows software, some embedded and industrial tooling, and occasional test fixtures where you want byte-predictable pixels. Our format converter supports it for exactly those cases.

A decision rule

If the image is...	and it needs transparency	and it does not
A photograph, for the web	WebP (lossy)	WebP, with a JPG fallback if you need one
A photograph, to send or print	PNG (accepting the size)	JPG
A logo, icon, or diagram	PNG, or WebP lossless on the web	PNG, or keep it as SVG if you can
A screenshot with text	PNG	PNG - never JPG

Conversion is a translation, not a repair

Three habits save a lot of grief. First, keep a lossless master - the camera original, the layered file, the SVG - and treat every JPG and WebP as a disposable export. Second, convert from the master rather than from a previous export, so you never stack lossy generations. Third, do not convert to "fix" quality: a blurry JPG becomes a blurry PNG that is four times the size.

Also note that converting to a format without transparency destroys it. A rounded PNG saved as JPG comes back with solid corners, which is the single most common surprise in format conversion. If you need both small files and alpha, WebP is the answer.

Batches, and where your files go

The converter takes a queue of images, applies the same target format to all of them, and returns a ZIP, which is what you want when a client sends forty PNGs that should have been JPGs. It also accepts HEIC input from iPhones - see HEIC compatibility and quality for what that conversion costs. Everything is decoded and re-encoded on a canvas inside your browser, so an entire batch of personal photos or unreleased product shots is processed without a single byte being uploaded, and there is no file-count limit because there is no server capacity to ration.