JPG to PNG Converter

You have been staring at it for twenty minutes: the client's logo, perfect except that it arrived as a JPG with a hard white rectangle where transparency should be. Dropped onto the site's dark header, it sits there like a sticker on a night sky. The fix everyone recommends is the same sentence — convert it to PNG — and this page is the version of that sentence with a tool attached and the fine print attached to the tool.

The JPG to PNG Converter above this article re-encodes images entirely in your browser: drop a file, get a PNG, download it. No upload, no watermark, no signup wall after the third image. For photos the output will be larger, for graphics it will be cleaner, and the reasons for both are genuinely interesting — they are the story of two compression philosophies that have quietly coexisted since the 1990s.

One honest thing before anything else: converting JPG to PNG does not create transparency, because the JPG never had any to give. Every pixel in a JPEG is opaque, edge to edge. The transparent-logo workflow is a two-step job — background removal, then PNG export — and this page covers how to chain it properly rather than pretend one tap fixes everything.

JPEG and PNG both answer the question how do we make this image smaller, and they answer it in opposite directions. JPEG, finalized in 1992 from work by the Joint Photographic Experts Group, is a lossy format built on a psychological wager: human vision tolerates losing certain high-frequency detail if the overall structure survives. PNG, developed in the mid-1990s as an open replacement for patent-encumbered GIF, is lossless and refuses the wager entirely — every pixel that goes in comes out.

That philosophical difference produces everything you will notice in practice. JPEG crushes photographs to a fraction of their raw size and shows artifacts around sharp edges when pushed too far. PNG preserves sharp edges exactly, supports 8-bit transparency that JPEG simply lacks, wins on file size for flat graphics — and balloons embarrassingly on photographs. Neither format is better. They are specialists, and most image problems are a routing question: which specialist does this file actually need?

Three decades of newer formats have tried to retire both and failed to finish the job — WebP, HEIC and AVIF each win somewhere and lose somewhere else, while JPEG and PNG remain the only two formats every browser, editor, printer and marketplace accepts without a compatibility check. When the destination is unknown, the safe pair is still the old pair.

The converter answers the routing question in one direction — toward the lossless specialist with alpha support — and the mechanics of both formats are worth understanding before you trust any converter with a whole folder.

The tool is above; the ritual is below. It takes about as long as dragging a file does.

1. 1

   Drop your JPG files

   Drag one or many onto the drop zone, or click to browse. Multiple selection and folder drops both work — each file gets its own row with size, dimensions and progress.

2. 2

   Review the preview

   The first file renders immediately with its dimensions shown. Check the orientation is upright — the converter applies the EXIF orientation before encoding, so rotated phone photos come out correctly.

3. 3

   Convert

   One click decodes each JPG, draws it to an off-screen canvas, and encodes the canvas as PNG. Processing is local; the progress rows fill in as files complete.

4. 4

   Download

   Grab files individually or take them all as a set. Output filenames keep the original name with a .png extension — no random suffixes, no watermarks.

5. 5

   Verify the size trade

   The result row shows the new file size next to the old. Photos will be bigger; screenshots and graphics may be smaller. If a photo ballooned further than expected, the section on file sizes below explains exactly why.

A JPEG encoder chops the image into 8×8 pixel blocks and runs each through a Discrete Cosine Transform — the DCT — which converts the 64 pixel values into 64 frequency coefficients describing the block as a sum of cosine wave patterns. The first coefficient is the block's average brightness; the rest describe progressively finer variation within it. It is the same mathematics that powers MP3 audio and much of signal processing: move the data into the frequency domain, where human perception has known blind spots.

Then comes the lossy step, and it is brutally simple arithmetic. Each coefficient is divided by a number from a quantization table and rounded to an integer. Small coefficients — subtle detail the eye was not going to miss — round to zero and vanish forever. Large coefficients survive, roughly. The quality setting on any JPEG encoder is just a scaling of that table: quality 95 divides gently, quality 60 divides savagely, and the same image becomes two different artifacts. Everything else in JPEG is lossless bookkeeping — the surviving coefficients are zig-zag ordered, run-length coded and Huffman compressed.

Two consequences follow. Chroma subsampling — typically 4:2:0, storing color at one quarter resolution while keeping brightness full — does its own quiet damage to colored edges. And because every block is transformed independently, heavy compression leaves visible 8×8 seams: blockiness, plus the mosquito-noise shimmer around sharp transitions known as ringing. A 4032×3024 photo contains 504 by 378 — exactly 190,512 of these blocks, which is a lot of independent little canvases on which to notice a seam.

PNG never discards anything, so its compression has to be cleverer about what pixels have in common. Before compressing, each scanline passes through a prediction filter — None, Sub, Up, Average or Paeth — which replaces each pixel with the difference between it and a prediction from its neighbors. Smooth gradients, which are nearly impossible to compress as raw bytes, become long runs of small numbers after prediction. The Paeth filter, named for its inventor Alan Paeth, picks the nearest neighboring pixel by a three-way distance test and encodes against that guess.

The filtered bytes then go through DEFLATE — the same LZ77-plus-Huffman algorithm inside zip files and gzip, which finds repeated byte sequences and replaces them with back-references. This is why PNG loves flat graphics: solid areas compress into almost nothing, and the prediction filters make gradients nearly free. It is also why PNG struggles with photographs: sensor noise means no two pixels are ever quite the same, back-references rarely fire, and the file ends up holding most of the raw data.

The alpha channel is PNG's structural superpower. Each pixel carries red, green, blue and an alpha byte — 256 levels of transparency from fully clear to fully opaque, with per-pixel control. Compositing software multiplies against that alpha to blend the image over any background. JPEG has no equivalent; its designers judged, reasonably for 1992, that photographs did not need transparency, and every logo owner since has been paying that particular toll.

File-size claims should come with arithmetic, so here are three conversions run through the actual pipeline — the same code path your files take.

The pattern in one sentence each. Photographs: JPEG wins on size by three to six times, PNG wins on preservation and editability. Flat graphics: PNG wins on both size and fidelity simultaneously, which is why screenshots default to it. Anything with transparency: PNG wins by walkover, since JPEG cannot enter the ring.

JPEGs from cameras and phones carry an EXIF block — often thousands of bytes of metadata about the camera body, lens, exposure, timestamp, and, on phones, a GPS IFD containing the latitude and longitude where the photo was taken. EXIF is genuinely useful: the orientation tag, formally tag 0x0112, is how a portrait photo knows it is a portrait. It is also a privacy leak with a memory: share the original file — not the social-platform copy, which is usually stripped — and you share the location.

Canvas re-encoding drops all of it. The decode step applies the orientation tag to the actual pixels — so the image stays upright — and the encode step writes a fresh PNG with no EXIF block at all, because the PNG the browser produces contains image data and nothing else. Your GPS coordinates, device model and precise timestamp simply cease to exist in the output. For anyone publishing original photos, converting through this tool is the cheapest metadata scrub available.

Back to the logo with the white box. The correct pipeline for a JPG that needs transparency runs through two tools, in order: remove the background, then export as PNG. The Background Remover does the pixel work — separating the subject from the white or near-white backdrop and writing an alpha channel where the backdrop used to be. A PNG export preserves that alpha perfectly, which is why the format ends up on the end of the pipe.

This converter is the honest second half of that workflow, and occasionally the first: converting first gives the background remover a lossless canvas to work on, so its edge detection is not fighting JPEG artifacts around the logo's boundary. Chain them as converter-then-remover for a battered JPG, or remover-then-export if the source is already clean. Either order beats trying to fake transparency by matching a background color to the destination — a trick that dies the first time the destination background changes.

For the related routing problem — PNGs that need to become smaller JPGs, with background flattening and a quality slider — the PNG to JPG Converter is the mirror-image tool, and the two pages are written to be read together.

The converter is exact about what it does. These are the boundaries of that exactness.

• Artifacts are preserved, not repaired. Converting freezes the JPEG's existing blockiness and ringing losslessly into the PNG — useful for preventing further decay during editing, useless for undoing the past.
• Progressive JPEGs and CMYK JPEGs convert fine — the decoder handles both — but extremely old or exotic files (think 1990s camera firmware oddities) occasionally fail to decode and are reported per-file rather than stalling the batch.
• 16-bit-per-channel PNG output is not offered; the browser pipeline is 8-bit, which matches the source JPEG's own 8-bit depth, so nothing is lost in translation.
• Animated files do not survive: a Motion JPEG or a multi-frame file converts as its first frame. Use a dedicated video or GIF tool for those.
• Color profiles are a soft edge: mainstream sRGB files round-trip exactly; exotic wide-gamut profiles may be normalized to sRGB by the browser's decoder, which slightly shifts saturation in saturated extremes.
• Very large images are limited by device memory rather than the tool — a 100-megapixel file needs roughly 400 MB of canvas memory and a browser tab willing to allocate it.

Designers and web developers dominate: logos, diagrams and UI assets arrive as JPGs from clients who do not know better, and the working format for all of them is PNG — alpha-capable, artifact-free, re-editable without generational loss. The dark-header logo problem is the canonical case, and it arrives in support email roughly weekly.

Photographers use the tool as a pipeline stage rather than a destination: converting a JPG to PNG before heavy editing — compositing, retouching, repeated saves — stops quality decay during the work, then the final export returns to JPG or WebP for delivery. Archivists take the same logic further and keep the PNG permanently for anything precious enough that further loss is unacceptable.

The third group is privacy-motivated and growing: people who publish original photos — marketplace listings, dating profiles, forum posts — and want the GPS coordinates gone before the file leaves the house. The metadata section above explains why this tool is a one-step scrubber for exactly that.

A quick heuristic serves all three groups when the source is ambiguous. Zoom to 400 percent and look at any sharp edge — a letterform, a power line, a jawline against the sky. If you see stair-step blockiness or a faint shimmering halo, the file is a JPEG showing its quantization scars, and converting to PNG before any further work stops the scarring from compounding. If the edge is clean at 400 percent, the file is already lossless somewhere in its history, and the PNG conversion is merely a container change rather than a rescue. Either way the check takes ten seconds and settles the question more honestly than any file-size table can.

Ways to get a PNG out of a JPG, ranked by friction and trust.

The routing between formats goes both ways — when the destination is JPG, the PNG to JPG Converter handles the mirror direction with flattening and a quality dial, and the WebP to PNG Converter completes the triangle for WebP sources.

Eight questions cover nearly everything the feedback inbox raises about this direction of conversion. The longer treatments live above.

Format conversion is a small ecosystem, and these three share the most workflows with this page.

• PNG to JPG Converter — the mirror direction, with white-background flattening and a quality slider for the size-versus-fidelity call.
• HEIC to JPG Converter — iPhone photos in the wild, translated before anything else will accept them.
• Image Compressor — when the PNG is right and the size is not, for graphics that still need to travel light.

The compact version of everything above: JPG to PNG is a lossless re-encode that trades file size for preservation, transparency support and a scrubbed metadata slate — done here in your browser, in batch, with no upload and no watermark. Photographs will grow and graphics will shrink, the DCT's past damage travels with the pixels, and the transparency you want is one background-removal step away. Route accordingly.