Question 1

What does the quality slider actually do?

Accepted Answer

It scales the quantization tables the JPEG or WebP encoder uses, which decide how coarsely each block of frequency detail is rounded off. It is not a percentage of the image, and it is not linear: on a photograph the drop from 95 to 80 often halves the file while being hard to see at all, whereas the drop from 50 to 35 saves comparatively little and is obvious immediately. That is why 80 is the default here and the setting nearly every encoder guide recommends. Below roughly 60 you start to see blocking on skies and skin, and ringing around text and logos.

Question 2

Why did my PNG come back the same size?

Accepted Answer

Because a browser canvas cannot compress a PNG. PNG is lossless, so there is no quality dial, and the real savings in tools like pngquant or oxipng come from reducing the palette and choosing better filters, neither of which a canvas exposes. Re-encoding a PNG through one usually produces a file the same size or larger. Rather than hand you that, this tool keeps your original bytes and marks the card. If you want a PNG to get smaller, convert it to WebP: it keeps the transparency and is routinely a third to a half the size.

Question 3

Can I compress to an exact file size?

Accepted Answer

To a maximum, which is the useful version of the question. Switch to “To a file size”, type a figure in KB, and each image is encoded several times at different quality settings until the best-looking one that still fits is found. No encoder can be told a byte count directly, because the size falls out of the image content rather than the settings, so every tool offering this works by encoding and measuring. Since file size rises with quality, it is a binary search rather than a scan and settles in about seven passes. If even the lowest quality is too big, you are told which images missed, and the fix is almost always to tick “Shrink oversized images” as well.

Question 4

Should I use JPG or WebP?

Accepted Answer

WebP whenever the file is going somewhere that displays it, which today means any browser released since 2020, plus most social and CMS platforms. Google's own comparison puts WebP at 25 to 34 percent smaller than JPEG at equivalent quality, and unlike JPEG it keeps an alpha channel. Stay on JPG when the file has to survive being emailed, opened in old desktop software, or handed to a print shop. Auto splits the difference by leaving JPGs alone, so the filename you get back is the one you had.

Question 5

Does compressing an image lose quality permanently?

Accepted Answer

Yes, for JPG and WebP, which is what makes them small. The discarded detail is not recoverable, and re-compressing an already compressed file compounds it: each save is applied to the artefacts of the last one. Compress from the original whenever you have it rather than from a copy that has already been through a save, and if you are not sure a setting went too far, open the comparison and drag the split across it before you download.

Question 6

How is the comparison sized?

Accepted Answer

The frame is sized to the picture, not the other way round, so there is never a letterboxed strip or a field of empty chequerboard around a small file. Most images are simply fitted whole. A small one is enlarged, but only up to three times, past which it is bigger without being any clearer. A very tall one — a full-page screenshot, say — is the case worth knowing about: fitting it to the height would leave a strip two hundred pixels across in which nothing can be made out, so it gets the full width instead and the frame scrolls. The line underneath gives the pixel dimensions and the scale you are seeing them at, and says when there is more to scroll to. Artefacts show first on smooth gradients such as skin and sky, and as a halo around hard edges like text or a logo.

Question 7

Will my EXIF and GPS data survive?

Accepted Answer

No. Compressing decodes the picture into raw pixels and encodes it again from scratch, and no metadata makes that round trip: camera model, lens, timestamp and GPS coordinates are all absent from the output. That also trims a little from the file. Orientation is the one field that would otherwise cause trouble, and it is applied to the pixels before encoding, so a photo shot in portrait stays portrait instead of falling on its side.

Question 8

How many images can I do at once, and how do I get them back?

Accepted Answer

Up to fifty, totalling 250 MB, with no single file over 30 MB or 80 megapixels, added in as many goes as you like. Those limits are about time and about what a browser canvas will allocate rather than about memory, since each image is decoded, encoded and released before the next one starts. With one image loaded the button downloads it directly; with several, each card has its own save row and the button below the settings packs the lot into one ZIP, with the original filenames preserved and duplicates numbered so nothing is overwritten.

tooldura

Image Compressor

About Image Compressor

How to use Image Compressor

Frequently Asked Questions