CCalclabhub

Work

How To Resize Product Images

Resizing is one multiplication applied to every coordinate in the frame. The only decisions are which ratio to scale by and what to do about the shape that is left over.

Quick Answer

scale = min(box width / source width, box height / source height) to fit inside, max(...) to fill and crop

s
The scale factor both dimensions are multiplied by
w, h
Source width and height in pixels
W, H
Target box width and height in pixels

A 2000 x 1500 source fitted inside a 2000 x 2000 box keeps its size at 2000 x 1500, is 3.00 megapixels, weighs about 900 KB, and leaves 500 pixels of box height to fill. Scaling 4000 x 3000 to Instagram portrait gives 1080 x 810 and keeps 7.3% of the pixels.

What Is How To Resize Product Images?

Every marketplace publishes an image specification, and a listing that ignores it is either rejected outright or quietly demoted in the results. Resizing is the step that turns a camera file into a listing asset.

Resizing means multiplying both dimensions by the same number. That number is the scale factor, and using one factor for width and height is what stops a mug from turning into an egg.

There are two sensible ways to hit a target box. Fitting inside scales by the smaller of the two ratios so the whole image survives, and filling scales by the larger so no part of the box is left empty.

Fitting inside is the safe default because nothing is lost. The cost is a gap: a four-to-three photograph inside a square box leaves the top and bottom empty, and that gap has to be filled with a background or removed by cropping to the box first.

Filling and cropping is what a platform hero image usually wants. The image is scaled up until it covers the box and the overflow is cut, so the product has to be centred before that happens or the crop takes the part you were selling.

Stretching sets both dimensions to the box, which multiplies them by different factors. That distorts the shape, and no export setting hides it. It is only defensible for textures and backgrounds where there is no recognisable outline.

The aspect ratio is the two dimensions once every common factor has been divided out. A 2000 x 1500 frame reduces to 4:3 and a square box is 1:1, which is why those two shapes never agree without either a crop or a gap.

Downscaling is always safe. Every pixel in the output was measured in the original, so the smaller file is a faithful summary rather than a guess. Upscaling is the opposite: the extra pixels are invented.

Megapixels scale with the square of the scale factor. Halving both dimensions quarters the pixel count, which is why file sizes fall much faster than dimensions and why a fifty per cent resize can lose three quarters of the data.

File size is roughly proportional to pixel count at a fixed quality setting. A web-optimised export runs near 0.15 MB per megapixel, a standard one near 0.30 and a maximum-quality one near 0.60, so the estimate is a multiplication rather than a mystery.

The published targets differ by platform and by position. Amazon asks for 2000 x 2000 on the main image so its zoom works, Shopify for 2048 x 2048, eBay for 1600 x 1600 in the gallery, and Instagram for 1080 x 1080 square or 1080 x 1350 portrait.

The larger target exists so the buyer can zoom. A 300-pixel image on a 300-pixel phone looks identical to a 2000-pixel one until the pinch gesture starts, and then the smaller one turns to mush exactly when the buyer is deciding.

The resizer takes the source dimensions, the platform box, the fit mode and the quality tier, and returns the output dimensions, the scale factor, the crop or the gap, the megapixels and an estimated file size. It is arithmetic rather than a judgement, so every line of it can be checked by hand.

Formula

fitting inside: s = min(W / w, H / h); filling and cropping: s = max(W / w, H / h)

Both dimensions have to be multiplied by one number, so the box can only be met on one axis. Fitting inside chooses the smaller ratio and leaves a gap on the other axis, filling chooses the larger and crops the overflow.

SymbolMeaning
wSource width
hSource height
WBox width
HBox height
sScale factor

fraction kept = s squared, because area scales with the square of the length factor

Halving both dimensions quarters the pixel count, which is why file sizes fall much faster than the numbers on the dimensions box. It is also why an upscale gains almost nothing noticeable.

SymbolMeaning
sScale factor
P_srcSource pixels
P_outOutput pixels
fFraction kept

estimated megabytes = output megapixels x megabytes per megapixel at the chosen quality

At a fixed quality setting, size tracks pixel count. A detailed and noisy frame runs larger than a flat one at the same megapixels, so the figure is a ceiling for planning rather than a promise.

SymbolMeaning
MPOutput megapixels
qQuality tier
BEstimated size

How To Calculate How To Resize Product Images

  1. 1

    Enter the source dimensions

    Both numbers in pixels, taken from the file itself rather than from what the camera was told to shoot. A cropped or rotated file reports the size it actually is, not the size it started as.

  2. 2

    Pick the platform target

    The target is a pair of numbers and platforms quote it as a requirement rather than a suggestion. Choosing the position on the site matters as much as choosing the platform, because the hero image and the gallery image rarely match.

  3. 3

    Choose how to handle the shape

    Fitting inside keeps everything and leaves a gap, filling crops the overflow, and stretching distorts. Make the choice before reading the numbers, because it decides which of the two ratios is even used.

  4. 4

    Read the scale factor

    One number decides everything else. Below 1 is a safe downscale in which every surviving pixel was measured, and above 1 is an upscale that invents detail which was never captured.

  5. 5

    Check the gap, the crop and the weight

    The leftover height or width is the part the box still wants, the crop is the part that will be cut, and the megapixels and byte estimate tell you whether the upload will be accepted at all.

Examples

Example 1: A 4:3 photograph into a square box

Source width
2000
Source height
1500
Target
Amazon main image, 2000 x 2000
Fit
Fit inside the box
Quality
Standard
StepCalculationResult
Ratio from the width2000 / 20001
Ratio from the height2000 / 15001.3333
Scale used when fitting insidethe smaller of the two ratios1
Output dimensionsround(2000 x 1) by round(1500 x 1)2000 x 1500
Box height still empty2000 - 1500500

Result: The width ratio of 1 is already the smaller of the two, so the source is left at 2000 x 1500 and no scaling happens at all. The result is 3.00 megapixels and about 900 KB at standard quality, with 500 pixels of box height still to fill and nothing cropped, because every pixel of the original is kept.

Example 2: A camera file down to Instagram portrait

Source width
4000
Source height
3000
Target
Instagram portrait, 1080 x 1350
Fit
Fit inside the box
Quality
Standard
StepCalculationResult
Ratio from the width1080 / 40000.27
Ratio from the height1350 / 30000.45
Scale used when fitting insidethe smaller of the two ratios0.2700
Output dimensionsround(4000 x 0.27) by round(3000 x 0.27)1080 x 810
Share of the source pixels kept0.27 squared, as a percentage7.3

Result: Scaling a 4000 x 3000 photograph by 0.2700 to fit Instagram's portrait box gives 1080 x 810, which keeps only 7.3% of the original pixels. The frame is 0.87 megapixels and roughly 262 KB at standard quality, and 540 pixels of box height are still empty because the width ran out first.

Calculator

Prefer a full-width tool? Open the How To Resize Product Images calculator page.

Common Mistakes

  • Upscaling to reach a target

    Scaling a 500-pixel image up to 2000 pixels produces a 2000-pixel file of a 500-pixel picture. The upload requirement is met and the listing looks worse, because the buyer's zoom reveals the blur within a second of opening it.

  • Stretching to fill the box

    Setting both dimensions to the target multiplies them by different factors and changes the shape. A round bowl becomes an ellipse, and no amount of export settings puts it back.

  • Filling a box with an off-centre product

    Filling and cropping cuts from the edges of the scaled image. If the product is not centred, the crop removes the part that was selling it, so centre the subject before the resize rather than after.

  • Reading the megabyte estimate as a guarantee

    The estimate is megapixels times a quality constant. A noisy or highly detailed frame runs larger than a flat one at the same pixel count, so treat the figure as a ceiling rather than a promise.

  • Ignoring the aspect ratio until the upload fails

    A 4:3 source and a 1:1 target cannot agree without either a crop or a gap. Choosing one of those two is the decision, and leaving it to a default is how products end up with a white band down both sides.

  • Rounding the dimensions by hand

    A scale of 0.27 applied to 3000 gives 810, but 0.3333 applied to 3000 gives 999.9 which is 1000. Rounding at the wrong step drifts by a pixel or two and breaks the exact-match uploads that some marketplaces still enforce.

  • Shooting small and planning to fix it later

    The resolution you capture is the resolution you keep. No resize step adds information, so the fix belongs in the camera setting rather than in the export dialogue.

FAQ

What size should product images be?

Around 2000 pixels on the long side for most marketplaces, because that is what makes the buyer's zoom work. Amazon, Shopify and eBay all sit in the 1600 to 2048 range, and Instagram wants 1080.

Does resizing reduce quality?

Downscaling reduces file size while keeping every pixel faithful to the original. Upscaling reduces quality while increasing file size, so the direction matters far more than the amount.

Should I crop or pad to reach a square?

Crop when the product fills the frame, because padding leaves bands that buyers read as an unfinished listing. Pad when the product needs breathing room or when the shape itself is part of the message.

How much does a downscale reduce the file size?

By roughly the square of the scale factor. Halving both dimensions leaves about a quarter of the pixels and about a quarter of the bytes at the same quality setting.

What is the difference between 2000 x 2000 and 2048 x 2048?

Nothing visible. Powers of two are a legacy of older storage and tile sizes, and a 48-pixel difference on a listing thumbnail is beyond any buyer's notice.

Why does my upload still get rejected?

Specifications usually cover dimensions, file format, colour space and a maximum size in megabytes. The resizer handles the dimensions and estimates the weight, but the format and colour space are export settings you choose yourself.

Can I resize without changing the shape?

Only by cropping or padding, because the source ratio and the target ratio are different numbers. Keeping the shape and filling the box at the same time is the one combination that is not available.

References

  1. [1]Amazon Seller Central, Product image requirements for listings — https://sellercentral.amazon.com/help/hub/reference/G1
  2. [2]Shopify Help Center, Image sizes and theme specifications — https://help.shopify.com/en/manual/online-store/images
  3. [3]eBay Seller Center, Picture sizing tips for sellers — https://www.ebay.com/help/selling/pictures/picture-sizing-tips