This link is the ONLY one i can find that specifically says to never use resampling in photoshop and to print at best quality and scan that back in. I have never seen this method used ever and i was wondering if anyone else can corroborate if it is best/better than using photoshop or other resizing software.

method described here

  • 3
    The method described there, print a photo and using a scanner to enlarge it, it is simply nonsense!
    – Rafael
    May 28, 2017 at 18:29
  • Actually, not it's not provided the print quality at the existing image size is good, and the scanner is high quality. It's not something to use as a "go to" method, but it can be better than interpolating a huge change in a digital image. The difference is you utilizes the scanner's interpolations as well as any image editing interpolation. Again.. things have improved nowadays.... but they wren't always as good as they are now with application interpolation.
    – Scott
    May 29, 2017 at 0:29
  • No it makes no sense, you would get errors 3 times whereas resampling would get the error 1 time. Now granted each of those 3 times may be more suitable for your end result in some weird cases
    – joojaa
    May 29, 2017 at 7:39

3 Answers 3


Rescanning? Why not Snake oil and Batwings?

Note that the effect is the sum of printing and scanning. Printers and scanners both (actually their software) try to sharpen the image and it's well possible that rescanning creates sharp edges which in some cases can result good looking appearance. But that still means heavily losing the original data. Do not use it until you have tried also other methods.

Photoshop's Bicubic Enlargening (a little different vesions available) works well as long as we do not try to enlargen so much that original pixels should be visible. Bicubic inserts blur that makes original pixel borders fuzzy. That blur is actual an estimated gradient.

Photoshop's nearest neighbour algorithm estimates nothing, so the image comes blocky as soon as the enlargening makes original pixels visible.

The image enlargenening software does not make the image fuzzy nor blocky, when the enlargening is increased. They lie the image is made of sharp areas and start to generate those when all actual info is runned to end. Those sharp areas get divided to smaller if the enlargening is still increased. That is total fake detail texture, but can sometimes look out plausible. The method definitely is a succes, if the image originally had blocks with sharp edges and those blocks are filled with flat colors or simple gradients. For example low resolution PNGs of logos had been resqued to printable size by using image enlargener programs.

Some examples: At first the original photo

enter image description here

The secod is Bicubic enlargement to 600%

enter image description here

Bicubic seems to have as much detail as the following, but it's not as blocky. The following is made by using "nearest neighbour". The blocks are original image pixels.

enter image description here

Finally the one that is got from an image enlargener (by On1). It looks out sharper, but the sharpness is fake. The program has invented artificial sharp borders. This is as false as the assumed smooth gradients that are generated by Bicubic. But I must admit that the sharpness feels less uncomfortable than the unsharpness:

enter image description here


Printing and then scanning... Totally no!

Printing a photo and scanning it again will not magically make new detail on the photo. You will not see the nonexisting details, new features, etc.

The only thing you will get is secondary details, like the texture of the paper or the droplets of ink that will give you the impression of having more detail.

It is just fooling yourself.

There are some basic principles when enlarging a photo.

1. Averaging new pixels

If you resample a photo exactly 2 times, from let's say 1000px to 2000px the resampling algorithm averages the values of the new pixels and creates a new one.

This works fine for large areas, but in borders, this make the photo a bit blurry.

2. Blur them

If you enlarge a photo at some point you will start to see the little squares that compose the image. Sometimes it is less noticeable seeing a somehow blurry image that the little squares.

3. Re-use them

There are some content-aware algorithms that use the surrounding texture to generate a texture. This is similar of using a clone stamp.

4. Gess it

And there are some other algorithms that especially on the borders try to generate a more contrasting border. Some claim that they are "tracing" the borders to know what the new invented pixels will be, this color or the color of the other side of the fence.

Normally The algorithms used by Photoshop are No 1. And it is a good approach.

a. I only recommend to resample a Photo exactly 2 times or in worst case scenarios exactly 3 times the size, so this averaging works as smooth as possible.

b. You can sharpen a bit the image after this resampling.

c. In some cases where you have a lot of flat zones of color, like a pie graph, drawings, etc, some other programs do a better job, like Benvista PhotoZoom (Which has a free trial) or the one the guy mention in his post https://sourceforge.net/projects/imageenlarger/

Again, I recomend enlargin only 2x or 3x max.


[April 12, 2013] You saw that right? Web sites bury dates any more because they'd rather have the content, no matter how old, and have you visiting their site than actually convey accurate information. It's always important to find the date of the posting if at all possible. Especially if an article involves anything on the technical side. Things change rapidly in the tech world. 4 years is like dog years in tech.

Although, in theory yes, printing on a high quality printer, at the native size, then scanning back in at a larger size/higher PPI will result in decent results at times. Provided both the printer and scanner are good quality devices.

However, today's Photoshop is much better at interpolation with the addition of features such as Content Aware fill.

Ideally you wouldn't use resampling for large changes, but in today's versions some enlargements are possible via resampling. It all really depends upon the image itself.

  • I tried telling the guy im arguing with this but i settled for walking away and just throwing link after link at him about it. I figured it would work too but damn is it old.
    – user56843
    May 28, 2017 at 17:58
  • @user56843 some people do not want to know the truth. They just want magick mumbo jumbo
    – joojaa
    May 29, 2017 at 7:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.