Photo too large to upload? How file size limits work
What KB and MB really mean, the two things that decide a photo's file size, and measured results for shrinking real 12 MP phone photos to 20 KB–1 MB.
A photo straight from a phone can be several megabytes: our three test photos were 1.62, 4.28 and 6.25 MB. Application forms, government portals and school systems often accept far less: 100 KB, 200 KB, 1 MB. This guide explains what those limits actually measure, what you give up to get under them, and how to pick a size that passes on the first try. All the numbers below were measured with Hanji’s own tools on real 12-megapixel iPhone photos.
What “KB” and “MB” mean, and why two counts exist
A file’s size is a count of bytes. The trouble is that “kilobyte” has two meanings in everyday use:
- 1 KB = 1,000 bytes (the decimal unit). Current macOS and iPhones count this way, and so does Android since version 8.
- 1 KB = 1,024 bytes (strictly a “kibibyte”, KiB). Windows File Explorer counts this way but still writes “KB”.
The same file can therefore show two different sizes. A file of exactly 100,000 bytes is “100 KB” on a Mac, but only about 97.7 KB when 1 KB is counted as 1,024 bytes, as Windows does. For megabytes the gap grows: 1,000,000 bytes versus 1,048,576 bytes.
You rarely know which rule an upload form uses, so Hanji always aims at the smaller one. When you ask for 100 KB, the result is at most 100,000 bytes, which is also under 102,400 bytes. It passes whichever way the form counts.
The two levers: pixels and compression
A photo’s file size is decided by two things.
- How many pixels it has. A 12 MP phone photo is 4032×3024, about 12 million pixels. Halve the width and height and you keep only a quarter of them.
- How strongly those pixels are compressed. JPG and WebP throw away detail the eye barely notices. The “quality” setting controls how much.
Quality is the cheaper lever at first. Measured on the same three photos at full size:
| JPG quality | Portrait (simple background) | Street scene | Beach and rocks |
|---|---|---|---|
| 95 | 2.07 MB | 4.01 MB | 4.98 MB |
| 90 | 1.62 MB | 2.81 MB | 4.10 MB |
| 75 | 0.84 MB | 1.55 MB | 2.44 MB |
| 60 | 0.59 MB | 1.13 MB | 1.88 MB |
Going from 95 to 90 saved 18–30% with no visible change. Going from 90 to 75 cut another 40–48%. Below that the savings shrink and blocky patches start to show in skies and skin, so at some point removing pixels is the better trade.
Hanji’s compressor works this way: it tries quality between 92% and 45%, and only when even 45% is too big does it reduce the resolution, then searches for the highest quality that fits again. It never goes below 45%, because a smaller image at decent quality looks better than a full-size one full of artifacts.
Notice also how much the subject matters. At every setting the beach photo, full of fine sand and rock texture, is about two and a half to three times the size of the portrait. Detail is what costs bytes, which is why two photos of the same pixel size can need very different treatment.
What each target does to a real photo
Here is what happened when the three phone photos (1.62 MB, 4.28 MB and 6.25 MB originals, all 4032×3024) were run through the compressor with JPG output:
| Target | Portrait | Street scene | Beach and rocks |
|---|---|---|---|
| 20 KB | 496×372 | 399×300 | 357×267 |
| 50 KB | 861×646 | 651×488 | 627×470 |
| 100 KB | 1340×1005 | 1030×773 | 887×665 |
| 200 KB | 2282×1712 | 1548×1161 | 1246×934 |
| 500 KB | full size | 2560×1920 | 1969×1477 |
| 1 MB | full size | full size | 3094×2321 |
Every result landed just under its target (for example 98.6 KB, 99.7 KB and 99.4 KB for the 100 KB row). A 1280×900 screenshot behaved differently: it reached 100 KB without losing a single pixel, and for 200 KB and above it stopped at 128 KB, because at top quality it was already small enough. The tool does not pad files to reach the target.
Saving as WebP instead of JPG keeps more pixels for the same budget. At 100 KB the portrait came out 2212 pixels wide as WebP against 1340 as JPG. The catch is that many upload forms accept only JPG, so use WebP only when you know it is allowed. The image formats guide compares the formats in more detail.
How to pick the right size
Start from what the photo is for, not from the smallest number you can reach.
- The form states a limit. Use that number, or the nearest preset below it. For a 300 KB limit, choose Custom and type 300; you do not need to aim lower “to be safe”, because the 1,000-byte counting already gives you the margin.
- No limit, just “too large” errors. Try 1 MB first. Two of the three test photos kept their full resolution at that size.
- ID and profile photos. These are shown small, so 100 KB or 200 KB is plenty; 1000 pixels across is far more than a screen thumbnail needs. If the form also asks for specific pixel dimensions, see the next section.
- Very small limits (20–50 KB). Expect roughly 350–860 pixels across for a 12 MP photo. Faces stay recognisable, but small text in the picture will not. Crop away everything you do not need before compressing, so the pixels go to what matters. The 20 KB and 50 KB targets are made for this.
- Printing. Do not compress at all. Printing needs pixels, and a 4032×3024 photo at full quality is the right input.
When to resize first
The compressor picks pixel dimensions for you. That is what you want most of the time, but not always.
Resize first with Resize Image when the form specifies dimensions as well as a size, such as “600×800 pixels, under 100 KB”. Set the exact width and height there, then run the result through the compressor. At those dimensions the photo will usually fit at high quality, so no further pixels are lost.
Signatures and scanned documents
Signature uploads can be capped very low, at 20 or 50 KB. Signatures are easy to make small because they are mostly blank paper:
- Sign on plain white paper with a dark pen, and photograph it in daylight without shadows.
- Crop tightly around the signature. A full A4 photo with a small signature in the corner wastes almost all of its pixels on white.
- If the signature image has a transparent background, saving as JPG fills the transparent area with white, which is what forms usually expect.
For scanned documents, a single page as an image can go through the same compressor, but text needs more pixels than a face to stay readable. If the form allows PDF, a PDF is usually the better container for multi-page documents. See Compress PDF and the guide on scanning documents to PDF.
Why a photo can’t get “smaller than X” without losing pixels
People are sometimes surprised that a 6 MB photo cannot become 100 KB at full size. The arithmetic explains it. At 100 KB, a 4032×3024 photo would have about 0.07 bits per pixel. No photo format can describe a detailed scene with that little information. Even at quality 60, the beach photo was still 1.88 MB at full size.
So for any given photo there is a floor: below a certain size, the only way down is fewer pixels. Where that floor sits depends on the content. The plain portrait kept full resolution all the way down to 500 KB; the detailed beach photo had to shrink even for 1 MB. A tool that claims to put any photo under 20 KB “without quality loss” is reducing the resolution and not telling you.
Common mistakes
Renaming .heic to .jpg. Changing the extension does not change what is inside the file. A form may accept the name and then fail to show or process the image, or reject it later. Convert it properly with HEIC to JPG. Hanji’s tools recognise a HEIC file even when it has been renamed, so a mislabelled file can still be fixed. Why iPhones produce these files is covered in the HEIC guide.
Sending a screenshot of the photo. A screenshot captures the photo at screen resolution, often with bars and status icons around it, and is often saved as PNG, which is a poor format for photos. You lose detail and usually end up with a bigger file than a properly compressed original. Compress the original photo instead.
Compressing a compressed copy. Each time a JPG is re-saved it loses a little more. If you made a 500 KB version yesterday and now need 100 KB, start again from the original photo, not from the 500 KB copy. (Hanji leaves a file untouched if it is already under the target and in the chosen format, so running the same file twice through the same target does no harm, but other apps may re-compress it anyway.)
Converting to PNG to “keep quality”. PNG is lossless, but it cannot restore what JPG compression already removed. Converted to PNG, the three test photos grew to 14.3, 21.1 and 25.2 MB, four to nine times the size of the originals, with no visible improvement.
Guides
- HEIC photos from iPhone: why forms reject them and fixes
- JPG, PNG, WebP, AVIF and HEIC compared
- Pixels, cm and DPI: how to convert, with tables
- How to Make a PDF Smaller (and What You Give Up)
- Scan Documents to PDF with Your Phone
- ID and passport photo sizes in mm and pixels
- How to Take a Passport Photo at Home That Gets Accepted
- Photo file rules for online applications: pixels, KB and age
- Do Online File Converters Keep Your Files?
- Print Passport Photos on 4×6: CVS, Walgreens, Walmart
- Character and Word Counts: Why Counters Disagree
- QR Codes That Keep Working: Static, Dynamic and Print
- Email Attachment Size Limits: Gmail, Outlook, Yahoo, iCloud
- Why Text Extraction From Photos Goes Wrong, and How to Fix It
- Shrinking Images to a File Size: What We Measured