Reduce image resolution means cutting the pixel dimensions, the actual width and height in pixels, not changing a DPI number. Downscale a 4000 by 3000 photo to 1600 by 1200 and you have genuinely fewer pixels, a smaller file, and a display size that fits where it is going. Change only the DPI field with resampling off and none of that happens, the pixel count and the file size stay exactly the same.
Resolution gets confused with two different things: the actual pixel grid a photo is made of, and a print density number a printer reads off it. Only the pixel grid matters on a screen, and only the pixel grid changes your file size. This guide covers the difference, the real arithmetic behind how much a resize actually saves, and the exact steps to downscale a photo on Mac, on Windows, and from the command line. It is part of our complete guide to compressing an image.
| Your situation | What to do | Why |
|---|---|---|
| Photo displays oversized on a screen or page | Resize the pixel dimensions (this guide) | Fewer pixels means a smaller file and a display that actually fits |
| File is too many bytes for an upload or email limit | Compress first at quality 80 | Often clears the limit with the pixels untouched |
| Printing at a specific size | Match pixels to print inches times DPI (below) | Print, not screen, is where DPI is real |
| Both a display problem and a byte limit | Resize, then compress | Removes pixels the page never shows, then re-encodes what is left |
What does image resolution actually mean?
Image resolution almost always means the pixel dimensions, not a DPI number. DPI stands for dots per inch, the ink droplets a printer lays down per inch of paper. PPI stands for pixels per inch, the number most editors actually show you, and it describes how densely a fixed set of pixels gets packed into a printed size (Adobe). Neither one is a pixel count by itself.
For anything shown on a screen, a monitor renders pixels one to one, there is no ink density to account for. What actually changes how large an image looks, and how many bytes it takes to store, is the pixel dimensions themselves, the real grid of 4000 by 3000 or whatever the camera wrote. If you reduce image dimensions, meaning fewer total pixels, the file gets smaller as a direct result, not as a side effect.
If you are prepping a photo for the web, ignore the DPI field entirely and look at the pixel count instead, that is the only number I check before uploading anything. That is also why any real guide to how to lower image resolution has to start with what resolution means before it gets to steps.
DPI vs pixel resolution: what actually changes the file?
Two numbers live in the same size dialog, and only one of them is real for a screen. Knowing which is which is most of the fix.
Why DPI is a print concept, not a screen one
With Resample switched off, typing a new DPI value does nothing to the picture itself, it only relabels how the existing pixels would print at a different physical size. Open an image's size dialog and you will typically see two numbers: pixel dimensions and a resolution value in pixels per inch. Change the second one alone and it does not add or remove a single pixel, so the file size does not move either. This is the single most common mistake people make trying to reduce resolution for the web: they change the DPI field, save the file, and find it is exactly the same number of bytes it was before.
300 DPI and the print pixel math
Print is where DPI is real, and the math is simple multiplication: print size in inches times target DPI equals the pixels you need. 300 DPI is Adobe's standard for a sharp photo print, so a 4 by 6 inch print needs 1200 by 1800 pixels, and a 5 by 7 inch print needs 1500 by 2100 pixels (Adobe). A large poster viewed from a few feet away can get away with closer to 100 DPI, since the eye does not resolve the same detail at a distance, while fine magazine or gallery work sometimes calls for 600 DPI. None of this applies if the image only ever goes on a screen, a screen does not read the DPI value stored in the file, only the pixel count.
| Use case | DPI target | Pixels needed (example size) |
|---|---|---|
| Screen or web display | Not applicable | Match the display width, typically 800 to 2000 px wide |
| Standard photo print | 300 DPI | 4x6 in = 1200x1800 px, 5x7 in = 1500x2100 px |
| Large poster, viewed from a distance | 100 DPI | 24x36 in = 2400x3600 px |
| Fine magazine or gallery print | 600 DPI | 8x10 in = 4800x6000 px |
How to reduce image resolution on Mac, Windows, and the command line
Cutting pixel dimensions takes a different tool than compression, our compressor does not resize yet, resizing and cropping are on the roadmap, not shipped, so here is how to do it with what is already on your computer, then bring the result back through the compressor.
How to resize image resolution in Photoshop
Go to Image, then Image Size. Make sure Resample is checked, that is what actually removes pixels rather than just relabeling them. Enter the new width and height in pixels, or switch the unit to percent.
A 4000 by 3000 photo cut to 40 percent lands at 1600 by 1200, sixteen percent of its original 12 million pixels, since both dimensions shrink together. The Resample dropdown also picks the interpolation method, Bicubic Automatic is a safe default for downsizing a photo and is what most people should leave selected.
Resize on a Mac with Preview
Open the photo, choose Tools, then Adjust Size. Enter a percentage or exact pixel dimensions, keep Scale Proportionally checked so the photo does not distort, and the new size is confirmed at the bottom of the dialog before you save (Apple Support).
Windows, no extra software
Open the photo in the Photos app, click the three dot menu, and choose Resize image. Pick a smaller preset or enter a custom pixel width, then save to create a resized copy alongside the original. If you resize photos often, Microsoft's free PowerToys utility adds a Resize pictures option straight to the right click menu in File Explorer, which is faster for a whole folder than opening each photo one at a time.
The command line, with ImageMagick
convert photo.jpg -resize 1600x1200 photo-small.jpg fits the image inside that pixel box while keeping its aspect ratio. convert photo.jpg -resize 40% photo-small.jpg does the same cut as a straight percentage, no need to calculate the exact pixel numbers yourself (web.dev).
Then compress what is left
Fewer pixels compress to fewer bytes at the same quality setting, so running the resized file through the free image compressor afterward is usually smaller than either step alone, and it strips the EXIF metadata at the same time.
How much do fewer pixels actually save in file size?
Reduce image resolution by half in each dimension and pixel count does not drop by half, it drops by three quarters, because both dimensions shrink at once, and the byte savings follow close behind.
Here is the arithmetic, worked from the same 4000 by 3000, 12 megapixel photo used above. Resize to half the width and half the height and you do not keep half the pixels, you keep a quarter, since area scales with the square of the linear reduction.
| Resize to (percent of original width and height) | New pixel dimensions | Pixel count | Pixels remaining |
|---|---|---|---|
| 100% | 4000x3000 | 12,000,000 | 100% |
| 75% | 3000x2250 | 6,750,000 | 56.3% |
| 50% | 2000x1500 | 3,000,000 | 25% |
| 40% | 1600x1200 | 1,920,000 | 16% |
| 25% | 1000x750 | 750,000 | 6.25% |
Pixel count and byte count do not move in perfect lockstep, a compressor still finds more or less redundancy depending on what is actually in the photo, but they move in close proportion at a fixed quality setting. Serving oversized dimensions can waste two to four times the data a page actually needs, the same finding covered in the next section, and that relationship holds in reverse too: cut the pixels to 16 percent and expect the byte count to land in that same neighborhood, not just a little smaller.
The 4000 by 3000, 4.5 MB phone photo used across these guides already compresses to about 900 KB at quality 80 with pixels untouched. Resize it to 1600 by 1200 first, 16 percent of the original pixels, and that same proportional relationship puts the compressed result somewhere in the 140 to 160 KB range. That is a proportional estimate to plan from, not a guaranteed result, the picture's own content, not just its pixel count, affects how well it compresses.
Should you serve multiple image sizes for the web instead of one giant file?
If you control the website, yes, serving one size to every device wastes bandwidth on small screens, and the standard fix is well established.
Serving a desktop sized image to a phone can waste two to four times the data the small screen actually needs to display it (web.dev). The common fix is keeping three to five differently sized copies of the same image and letting the browser pick the right one with the srcset attribute, something like photo-480w.jpg 480w, photo-1080w.jpg 1080w. That is a developer workflow for a page template, not a one off resize, and it sits outside what a single file compressor does, our tool exports one size per upload, the size you gave it.
A photo shown 400 pixels wide on a phone screen does not need to arrive at 4000 pixels wide no matter how good the camera was, and every extra pixel is real load time on a slow connection. If you are just prepping one photo for one upload, the single resize covered above is the whole job.
Why upscaling does not fix low resolution
Adding pixels a photo never had does not add real detail. Resampling up interpolates new pixel values from the ones around them, essentially an educated guess, and the result usually looks softer under close inspection rather than sharper. If a placement genuinely needs more pixels than your original has, the fix is finding a higher resolution source, not stretching the one you have got. Treat any AI upscaler claim the same way, it can produce a plausible looking result, but it is not recovering detail that was actually captured.
When should you not reduce resolution at all?
Keep an untouched master copy of anything you might crop tighter, print larger, or reuse bigger later, downscaling only goes one way, there is no getting the removed pixels back afterward.
Do not resize an image that is already small, a 300 pixel wide icon or screenshot has little left to give up, and cutting it further mostly costs you usability, not meaningful bytes. And if the actual problem is a byte limit rather than an oversized display, an email attachment cap or a form's upload limit, compressing at quality 80 usually clears it without touching the dimensions at all, which is the faster fix when the pixels themselves are not the issue.
If a resize does not seem to be shrinking the file much, check that Resample was actually turned on. The most common troubleshooting case is exactly the DPI trap covered above: someone changes the resolution field, sees no pixel-dimension change in the file, and assumes resizing does not work. The second most common case is resizing a screenshot or a graphic with sharp text, downscaling those can blur fine lines in a way a downscaled photograph rarely shows, so check the result at full zoom before you use it.