Compress product photos to a file-size limit
A file-size limit is not a quality setting, it is a budget. You decide how to spend it across three things: pixel dimensions, fine detail, and file bytes. Spend on dimensions until the image reaches the platform size, then spend the rest on quality and stop when you can no longer see the difference at 100%.
Most bad compressions fail in the same way: the image is compressed first and resized afterwards. That order throws away detail twice. Resize to the exact target pixels first, then compress the final image.
Opens the tool with no preset. Pick a platform and rule first, then drop your images in.
Which platforms actually cap file size
Not every marketplace publishes a byte limit, and a limit that does not exist cannot be failed. Check before you spend an afternoon optimizing files against a rule nobody enforces.
| Marketplace | Main image size | File-size cap |
|---|---|---|
| Temu | 800×800 | 2048KB |
| Shopee | 800×800 | 2048KB |
| Shopify | 2048×2048 | 20480KB |
| Amazon | 1000×1000 | not specified by the platform |
A cap shown as "Not specified by the platform" means this site has not verified a byte limit for that rule. Do not invent one; keep the file reasonable and move on.
What actually eats bytes
Compression cost is mostly detail. A smooth white background behind a hard-edged product compresses extremely well, because large areas of identical pixels cost almost nothing. Noise, grain, fabric texture, foliage, and gradients cost a lot, because every pixel differs slightly from its neighbor and must be encoded separately.
This is why two product photos of the same dimensions can differ in file size by a factor of five. A high-ISO shot of a knit sweater has more visual information than a matte bottle on pure white, and information is what bytes pay for.
Sharpening makes it worse. An unsharp mask creates bright and dark halos along every edge, which is exactly the kind of high-frequency detail a JPEG encoder has to spend bits on. Sharpen lightly, after resizing, and re-check the file size afterwards.
The order of operations
- Crop and resize first: Bring the image to the exact target pixels, for example 800×800. Resizing after compression destroys the quality you paid for.
- Fix the background before compressing: A flat, single-color background is nearly free in bytes. Cleaning it up first gives the encoder more of the budget for the product.
- Export at high quality once: Start from a quality level where artifacts are invisible, then step down. Starting from the top and stopping early is faster than rescuing a crushed file.
- Read the actual byte count: Compare against the cap, such as 2048KB, and record the number. "Looks small" is not a size.
- Step quality down in small increments: Change one setting at a time. A large jump can hide a visible artifact that a small step would have exposed.
- Compare at 100% on a detail area: Look at edge detail: stitching, printed text, metal highlights. If the artifact is invisible on that crop, a shopper will not see it either.
What "no visible loss" really means
Lossless compression exists but does almost nothing for photographs. Real product photos are compressed lossily, and the honest goal is not zero loss but invisible loss at the size the image is displayed.
That gives you a way to measure instead of guess. Compare the original and the compressed file at the exact display size, then at 100%, on the region with the most fine detail. If the two are indistinguishable in the detail crop, the compression is done. If you can see blocky patches around text or a halo along the product edge, step back up.
Do not compare at 400% on a thumbnail-sized display and call it a loss. Detail that no one can see at display size is not worth bytes.
When not to compress
- When the platform publishes no cap. Turning a 900 KB file into 300 KB helps page speed, not approval, and it should be a deliberate choice.
- When the image is a flat graphic, logo, or line art. PNG keeps edges crisp where JPEG introduces ringing.
- When the source is already compressed several times. Re-encoding a re-encoded JPEG adds artifacts without saving much.
- When the product detail is the selling point — fabric weave, engraving, label text. Compress less and reduce dimensions instead.
Compress a whole folder
Open the batch image tool, pick the platform and rule, and drop the folder in. Each file is resized to the exact spec, padded if the background requires it, and compressed in one pass — the byte budget is checked against the platform rule rather than against a slider you have to eyeball.
Processing runs inside the browser: images are decoded and re-encoded locally, so a folder of unreleased product photos never leaves your machine and nothing waits in an upload queue.
Frequently asked questions
How do I compress a product photo to a specific file size?
Resize to the exact target pixels first, remove noise where you can, then step the quality down until the file lands under the cap, checking a detail crop at 100% after each step.
Does compressing an image reduce its dimensions?
Not by itself. Compression changes how the file is encoded, not how many pixels it has. If you need exact pixels, set them explicitly: 800×800 for Temu, for example.
What is Amazon file-size limit for main images?
This site records it as not specified by the platform. In practice that means no verified fixed cap, so optimize for page weight rather than for approval.
Is PNG or JPG smaller for product photos?
JPG is usually far smaller for photographs, because it discards detail that the eye is less sensitive to. PNG wins for flat colors, logos, and transparent areas.
Can compression remove the white background?
No. Compression does not change background pixels into anything else, and it does not add transparency. Prepare the background first, then compress the finished image.