# Does resizing a photo larger or smaller destroy the image quality more?

There are various interpolation settings that image editors use to rescale photos larger or smaller. For some reason, resizing a small photo larger sounds more artificial and destructive than doing the reverse: shrinking it. Is this the correct intuition and why?

• What is the point of this question? What problem are you trying to solve? Nobody chooses to upscale or downscale based on which one is least "destructive". People upscale for some reasons and downscale for different reasons. Commented Jan 28, 2022 at 19:39
• well if fitting the dimension to some size is the aim, only the aspect ratio should really matter. But for whatever reason, it would be good to know whether resizing the aspect ratio based on pixel dimensions would deteriorate image quality or not. Commented Jan 28, 2022 at 20:04
• So you're saying you want to change the aspect ratio of an image? I would do that only through cropping, not by resizing, which would distort the image. Commented Jan 28, 2022 at 23:09
• technically it depends on the image, a black square can be losslessly reduced down to a very few number of pixels while a circle cannot be reduced at all without losing information... Commented Jan 29, 2022 at 9:54

I would argue the opposite:

Enlarging a picture does not add any new information, but neither is anything discarded. If you view an enlarged picture from an equally increased distance, it should appear pretty much the same as the original. If you shrink it to its original size, you'll get the original picture (ideally, supposing compatible algorithms).

In contrast, shinking a picture does throw information away, and no matter how close you get, you will have lost detail. Enlarging this image will never result in the original again.

• last sentence there seems contradictory that shrinking and enlarging are both destructive ("lost detail" by shrinking) Commented Jan 28, 2022 at 0:48
• If you start with an image 1000px square, then enlarge it to 2000px, you still have in effect 1000sq px of information, just spread out into 2000 sq px. If you instead halve it to 500px, you've thrown away 3/4 of the information [halve both dimensions, you quarter the number of px], which you cannot ever get back. Commented Jan 28, 2022 at 8:14
• So shrinking an image is destructive and enlarging it is not? Commented Jan 28, 2022 at 20:05
• yes. that's the gist of my answer.
– ths
Commented Jan 28, 2022 at 20:13

It is true that shrinking an image throws away information that you will never get back. But the story is more complicated than that.

When you shrink an image, you make any anomalies in it smaller and harder to notice. You also reduce noise. Many times I've taken an image that I thought was hopelessly blurred and made it acceptable by shrinking. Here's one example:

On the other hand, enlarging will blow up those anomalies and make visible what might have been missed.

• What are technical examples of anomalies Commented Jan 31, 2022 at 12:55
• @user610620 I gave you one example, blur. Another common one is CA. It's basically a catch-all term for anything that's different from the scene you were taking the picture of. Commented Jan 31, 2022 at 13:14

This is somewhat true. The information in a digital picture is more or less carried by the pixels, the more pixels, the more information.

When you scale up an image, you create more pixels but don't add any information, so the same information is spread over more pixels and the image is blurry.

When you scale down, you are in the opposite situation, you have to discard information, which is done by operations akin to averaging pixels. The perceived quality of the result is however variable, and depends on the scaling factor: we don't look at a tiny image exactly the same way as a big one, and you can have artifacts caused by aliasing.

• So in general is creating more pixels without adding any information less destructive than scaling down by discarding information? Think the eyes perceive bluriness/pixelation to be a hallmark of poor image quality Commented Jan 28, 2022 at 3:36
• One could consider it as not destructive at all. However, if you then try to scale back to the original size, you will get a different picture. Commented Jan 28, 2022 at 7:39
• @user610620 you see the blurriness only because you look at it from too close. every picture has its optimal viewing distance. think of a billboard up close.
– ths
Commented Jan 28, 2022 at 17:26