Dimensions are a pixel count, not a size on paper
An image's dimensions are simply how many pixels wide and how many pixels tall it is. A 1200 × 800 image contains 960,000 pixels; a 600 × 400 image contains 240,000. Halving each side removes three quarters of the pixels, not half — this square relationship is the single most useful thing to remember about file size.
Dimensions say nothing about how large the image will look in print or on screen. The same 1200 × 800 image can fill a phone screen or print a small postcard; that depends on the display and on the DPI used when printing. Dimensions describe detail capacity, not physical size.
Which change moves file size the most
- **Format** — moving a photograph from PNG to JPEG or WebP is often a 5–15× reduction before anything else is touched, because PNG stores every pixel losslessly while JPEG and WebP discard detail the eye is less likely to notice.
- **Quality** — within a lossy format, quality is a continuous dial. On a detailed photo, the difference between quality 95 and quality 75 is frequently a halving of file size with little visible change at normal viewing distance.
- **Dimensions** — quadratic. Cutting each side in quarters cuts the data to roughly one sixteenth. Very effective, but it also removes real detail, so it is the change you make last and only when the requirement allows it.
- **Content** — a flat blue sky compresses far better than a gravel path. Two images with identical dimensions and settings can differ in size by an order of magnitude.
Choosing what to change
- Read the requirement carefully: is it a pixel dimension, a byte limit, or both? Image Metadata tells you the current dimensions and byte size before you start.
- If only a byte limit is stated, start with format and quality in Compress Image and leave the dimensions alone.
- If only dimensions are stated, use Resize Image and accept whatever size results — a dimension requirement is not a size requirement.
- If both are stated, fix dimensions first with Resize Image (so the requirement that cannot be negotiated is met), then bring the bytes down with Compress Image or Exact KB Photo Compressor.
- Re-check the result: dimensions, then byte size, then a visual look at 100% zoom.
A note on upscaling
Enlarging an image adds pixels but not information. A 400 × 400 photo stretched to 1600 × 1600 has four times the pixels and none of the extra detail — it is simply interpolated, so it looks soft, and the file grows while the quality does not. If a form demands a minimum dimension your photo cannot reach honestly, it is usually better to re-photograph or re-scan than to upscale. Where a small upscale is unavoidable, do it once, in Resize Image, rather than repeatedly.
Downscaling is different: it genuinely discards detail you had, so it is safe in the sense that nothing is invented — but detail removed cannot come back later.
Common questions
Why did halving the width and height not halve the file size?
It removes three quarters of the pixels, so the file should drop by far more than half if settings are unchanged. If it did not, something else changed too — a different format, a quality floor the encoder would not go below, or metadata being added back in.
Is a bigger image always a better image?
Only up to the point where it exceeds what you will actually use. Extra pixels beyond your display or print size cost space and upload time without adding visible quality, while a file that is too small cannot be sharpened back.
Do megapixels tell me the file size?
No. Megapixels count pixels — a 12 MP image is 12 million pixels. File size depends on how those pixels are encoded, which is why two 12 MP photos can differ enormously in size.