How to enlarge images with no pixellization and minimal quality degradation (using free software)

Have you experienced a situation where an image you are interested in using is smaller than you would like it to be, whether it might be for printing purposes or for use in illustration projects or even for usage on the web, etc? Typically, straight-enlarging of an image comes with too much of a degradation in quality, with visible pixellization and undesirable artifacts, and a general loss of detail.

This post will describe a process that can lead a different outcome.

To be clear, the steps described here will not magically produce a perfect image out of a source which (by definition) is far from optimal.

Rather, what is on offer here is the prospect of processing the enlarged image algorithmically to get a much better/cleaner looking resized image than otherwise would be the case.

Clean enlarge illustration2

Nor is this post about manual image processing or Photoshopping techniques. The tool used here is a free software called SmillaEnlarger, and it can perform image enlargement quickly and easily (and automatically) in one or two clicks.

To demonstrate the kind of results that can be obtained, I will attempt to enlarge a test image using the different profiles that are pre-built into the application.

First, step by step instructions, then an overview of a test resize.


Step by Step:

Step 1: go to the SmillaEnlarger web site and download the latest version. Unzip then run SmillaEnlarger.exe (no installation needed).

Step 2: Once the program runs, simply drag and drop the image you want onto the interface.

Smilla 0 Cr

Step 3: Enter the desired output height in the top left. Note: the output width will be calculated automatically and will assume the same aspect ratio as the original.

Smilla1 Cr

Step 4: use the dropdown under the “Enlarger Parameter” section to select one of the available presets.

(Optionlly) press the “preview” button on the right to get a preview of what the enlarged image will look like under each setting. You (move the image slightly with the mouse to get back to the default, no-processing enlargement).

Smilla2 Cr

Step 5: enter a desired filename for resized image (or use the default provided) then click “Enlarge and Save”.

Smilla3 Cr

Step 6: (optionally) go back and redo steps 4 and 5 for each settings profile in the dropdown. Later, you can compare the different outputs to see which one you like best.

That’s it, you’re done!


Overview of test results:

Original image: 250×147, see below (actual size). The desired output size was 470×267.

RH Orig 250x147

I choose this output size (roughly double the original) because I wanted to publish the resized images on this blog in their actual size, and I am restricted by the width of this column.

But I’ve performed enlargements that are x5 or x6 the size of the original, with the same general results as the ones presented below.


1. Control: The enlarged image below was not done in SmillaEnlarger, rather it is a normal resize performed in Photoscape (Lanzcos resizing filter). I include it here for reference.

RH Orig 470 -lanzcos

Notice the pixellizations around the text and mountain areas, and the chunkiness of the sky in the midde.


2. Default: this is the “default” settings profile in SmillaEnlarger.

RH default 470

Both the text and mountain objects are cleaner, and the chunky blocks in the open sky is replaced by a more textured/grainy look (which may be much more preferable in print or other contexts than the pixellized chunks).


3. Sharp: this is the “sharp” settings profile in SmillaEnlarger.

RH sharp 470

This profile produces a less grainy image, which might be better suited for some projects, but the artifacts around the objects (text/mountains) are a little more visible than the ’default’ profile.


4. Sharp & noisy: this is the “sharp & noisy” settings profile in SmillaEnlarger.

RH sharp&noisy 470

It removes the blocky chunkiness, opting for a more “cloudy” and less grainy texture. Although in my opinion much better than the control, it may not be the best treatment of this particular image, but I’ve seen this profile do an excellent job on a different kind of image that is busier and features more objects.


5. Painted: this is the “painted” settings profile in SmillaEnlarger.

RH painted 470

This profile does an excellent job with both texture and detail, but will give your image a certain “drawn” quality. It works really well for illustration type projects that don’t necessarily have to look like photographs, but is not very well suited for, say, pictures of people (see a tiny webcam picture that I snapped of myself and enlarged below, where the beard and ear and other elements look like an oil painting).

WebcamImage-267 painted


6. New: this is not a built in settings profile but is rather one that I made, intended to be a cross between the “default” and “painted” profiles, which are my two favorites.

RH new 470

I kind of like both the grainy texture and the sculpted edges around the visible elements simultaneosuly. I find that I use this profile quite a bit

If you want to create this profile click the “new” button under the preview section, then check “allow changes” and use the following settings. You can edit the name is well; it doesn’t have to be called “new” ;)

new profile


The point is that with a tool like this one, you can really get a much better result than a simple image resize. All of the versions above (2-6) are vastly superior to the simple resize in my opinion.

If you know of other tools or methods to do this sort of thing, please let us know in the comments.


 
 
 
Samer Kurdi

Samer Kurdi

Has been reviewing software since 2006 when he started Freewaregenius.com
Samer Kurdi
April 24, 2011
Samer Kurdi
11
flattr this!
  • http://carbonize.co.uk Carbonize

    From what I can see this program does nothing special other than automatically applying a filter after resizing. I always apply sharpen after enlarging an image as most resizing algorithms make it slightly blurry.

    But as I’ve already said this program does nothing your existing editor cannot do. Resize the image in your favourite editor (Paint.NET for me) and then apply whatever effect filter you want be it sharpen, noise reduction, paint effect and so on.

  • Chris

    Whoa! Radiohead, Iron & Wine. My kinda tunes!

  • jasray

    Hate to reveal the reality–it’s absolutely, totally noticeable–the enlarging. So easy to see for anyone who works with pics. Maybe it’s like counterfeit money: Anyone who works with bills all day long (I mean really observes and is aware) can pick out a counterfeit in a wink. Other may accept the final counterfeit result without question.

  • boony

    Tried it on a few images. Not bad. This might be a solution for someone who doesn’t want to play around with filters after enlargement. Sort of a passable one click solution.

    jasray: Read the second paragraph again. The counterfeit money comparison is not correct in this case. It’s not meant to create a perfect image. Just one that is more acceptable. Anyone will notice a difference.

  • Shawn

    Don’t forget that SmillaEnlarger also run’s from the command line… for Batch processing :D

    Try this out for fun…

    Draw a 3×3 grid in paint (Take 9 different colors)

    Increase it with SmillaEnlarger (3000%) and try the different settings you’ll love the effects that it will generate. Add this with Scale2x and ImageMagick and you’ve got one heck of a powerpack to make a small picture look like a amazing work of art.

    Shawn

  • http://www.rchsd.org Tom Williams

    I’m not much of a graphics dude (dangerously curious) but I would like to see another app as another aspect of this review. An app that can shrink an image without distortion and all while being applied as an average non-graphics type expert. I use paint.net to do some of this stuff and I do appreciate its functionality but I also refer a ton of noobs (or peeps more noob than I) to this site for a quick and easy free solution to their everyday challenges.

    Thank you very much for continuing your great site.
    Been using and promoting this site for about 3 years now.

    Regards,
    Tom Williams

    • Samer

      @ jasray: the metaphor you use — that the output is like a counterfeit bill that an expert can identify — is actually somewhat of a compliment. Of course the enlarging is imperfect; the only way to get a perfect image is to go out there and take one (or find another source or buy the hi-res version or whatever).

      @ Carbonize: yes that is what the program does is apply filters. But it’s a nice, all-in-one place for these and will let you create and save filtering profiles and quickly test them out.

      @ Tom Williams: thanks for your kind words about my site!

      @ Shawn: I haven’t tried what you were describing but would love to see a screenshot of what you were doing.

      Thanks

  • Anonymous

    First of all, thanks are due for presenting and clearly describing this useful free tool.

    Now, as regards Carbonize’s assertion and the reply given that the tool only applies filters after simple resizing, I quote its description on the site linked:
    “Based on original algorithms, this standalone software uses a number of filters that are designed to remove some of the typical artifacts generated by the enlarging process and “fill in” image information…”
    Thus, it seems to not only use filters but also fill in image information according to algorithms, as done by other such tools of the premium variety.

    As regards similar free tools, I know of easyResize, but its site seems dodgy, as it’s now blocked by my online guard, so I do not want to link to it.

  • Corrina

    Thank you for this. It will come in very handy in my classroom when the kids need a pic for a project and a “one-touch” solution is all we have time for!

  • Pingback: Photo Card Place » Blog Archive » How to enlarge pictures without distortion

  • Yvonne

    Just tried your program: seems to be what I’ve been looking for.
    (One cavil, not about the program but the licence: is the font *really* copperplate, or just my computer’s next-best approximation? ‘Cos it rendered it totally unreadable!)