Steven Love

How I enlarge Images and preserve their quality

A short time ago Bianca Thomas wrote a journal entry on “Stretching Pixels” and how enlarging an image can degrade it if you stretch the pixels too much.
.
In essence she is correct. When you increase the image’s physical print size while leaving the pixel dimensions the same the DPI will have to drop in order to stretch the image to the needed size. These stretching of the existing pixels creates gaps which can show up as distortions called “Artifacts”.
.
There is another method of enlarging an image while reducing the number of artifacts created. It is called “Interpolation”, also known as “Re-sampling”.
.
In Photoshop CS2 and CS3 you can Interpolate the image by using what is called the “Bicubic Method”. This method allows you to enlarge an image but retains the original DPI.
.
The Bicubic Method enlarges an image by using a very complex mathematical calculation similar to Fractal Geometry. The program basically looks at the image and then through Fractal geometry calculations “Predicts” where to add additional pixels so that you can increase the pixel dimensions while retaining the same DPI setting.
.
Photoshop has three Bicubic Methods;
If you are only enlarging the image a small amount, say 10% to 20% above its original size then the default Bicubic setting is good enough.
However, if you plan to increase size more than 20% then you’ll want to select “Bicubic Smoother”, which will help reduce distortions.
.
The third method called “Bicubic Sharper” is what you would use if you were reducing the size of an image. But since most people want to make their images larger the first two settings should be used.
.
There is a limit, however, on how large you can interpolate without distorting the image. For example; if you take a 4 megapixel image, which will measure 2272 by 1704 pixles, and boost it to an 8 megapixel image, which will measure 3264 by 2448 pixels, you will get some distortions.
.
There is a way to prevent these distortions and still be able to double the size of the image. First, do not try to enlarge the image to a huge size in one single step. The best way to enlarge an image and preserve most of its quality is to do it in two or more stages.
.
The following example will be an 8 mega pixel at 3264 by 2448 pixels at 300 dpi. This original size will print to 8.16 by 10.88 inches in Landscape orientation. But the procedure can also be applied to Portrait orientation as well.
.
First I open this image in Photoshop. On a side note, you should be doing this with either a RAW, PNG or TIFF format. JPEGs don’t interpolate well because of it being a lossy format to begin with.
.
Next, run the image through a noise reduction filter to reduce or get rid of any digital noise created by the camera. Then do what ever color, contrast, sharpness and exposure adjustments that are needed.
.
Now, go to the Image Tab and click on Image Size. You should see a display showing the current image size and resolution. Before you change the size make sure that the boxes labeled “Constrain Proportions” and “Re-Sample” are checked.
.
Next, we move on to change the document size. The document size is the size the image will physically print at. Do not alter the Pixel dimensions at this time.
.
As I mentioned earlier it is best to do this in incremental stages rather than one large step. For this example the current document size is 8.16 for height by 10.88 for width. You will now change the height to 12.8. By checking the box “Constrain Proportions” you should see the width size change automatically to 17.067 which is the correct width proportion for the changed height size. You will also see that the Pixel Dimensions also changed accordingly. That is why I told you not to tamper with the pixels. The program will do that for you.
.
Also notice that the resolution of 300 DPI did not change.
.
Before you click “OK” to complete the interpolation change the Method from the default Bicubic to Bicubic Smoother. After clicking “OK” you will notice that the size change will take some time because the mathematical calculations the program has to do are extensive.
.
Once this enlargement is complete you may notice some tiny distortions in area of solid color such as in the sky or in other mono-colored areas. The distortions may look like little grains. Once again use a noise reduction filter to remove or at least reduce the distortions.
.
Once you have reduced the noise again go back to the image size control and using the same procedure as before increase the height from 12.8 to 18.0. The width should automatically change to 24 inches.
.
Now look over the image again for noise and apply the noise reduction filters to remove or reduce the noise.
.
When you have finished save the image as a 16 bit PNG. This will preserve all color depth and clarity. But be warned that the file size will be huge and the saving process will take some time depending up your computer’s RAM capacity.
.
Use this version as your master image copy. If you need to make a copy to upload to the internet simply switch it back to 8 bit and save a JPEG copy of it.
.
On a final note, even though this will enable you to make some large high quality prints there will always be some images that will have a small amount of noise or distortions. The key is to make note of whether the noise is reduceable or not. If the noise gets worse as you enlarge then you know that the prior size was the maximum you can go with the image you have.
.
I’ve been able to successfully enlarge 8 megapixel 8 X 10 images so that they’ll print as big as 18 X 24 posters with little to no distortions. But for those of you with smaller cameras, such as 4 to 6 mega pixels, you may not be able to interpolate as large as I do without some distortions.
.
That’s how I enlarge my photos using Photoshop.
.
If any of you have any additional info on the subject feel free to post in the comments.

  • Deri Dority

    Deri Dority, 3 months ago

    I use OnOne’s Genuine Fractals. It is a Photoshop plug in, and in 2 steps you have your photo interpolated (it goes up to 1000%). If need be I then run it through Noise Ninja to get rid of any noise. Storing and working with your photo in TIFF or PNG (as you said above) is a great way to preserve it. Working with a photo in JPEG (or just plain opening and closing a JPEG file) leads to image degradation.
    Thanks for posting this topic Steve. This is one of those questions that is asked on a frequent basis.

  • Steven  Love

    Steven Love, 3 months ago

    I’ve heard of Noise Ninja but not that other Plug-In.
    I’ll have to check that out.

  • aphoto4you

    aphoto4you, 3 months ago

    yes Steven great info here….lots of people can and will learn from this and shoud learn…You and I will have lunch in AZ when i get there so we can shoot the breeze about lots of technics lol…

  • Sharon Perrett

    Sharon Perrett, 3 months ago

    Very interesting and very nicely explained Steve, thank you for taking the time to do that, I have PS7 which I haven’t used as yet.

  • Sean Crease

    Sean Crease, 3 months ago

    Thanks for the info, very handy.

    Sean

  • Kathie Nichols

    Kathie Nichols, 3 months ago

    Thanks Steven, most informative, I will save this to my favourites. There is so much to learn and RB has a fabulous community with members such as yourself sharing this valuable information. All I have to do now is buy myself the Photoshop progamme!

  • genevieve m

    genevieve m, 2 months ago

    great info thanks so much! will have to try this out.

  • Helene Kippert

    Helene Kippert, 2 months ago

    Thanks for the info Steven!

  • Eyal Nahmias

    Eyal Nahmias, 2 months ago

    I use the same plugin as deri. “OnOne’s Genuine Fractals.” it does a fantastic job in one step. If you’re a contributor to alamy for example or other stock agency they require you to use it. Version 5 now out, is well designed and does a fantastic job. The problem I have with the system you use is that every time you ran noise reduction you blur the image slightly and running it twice may create an irreversible quality drop. If it’s an image that needs to be sharpened later by a client than there may not be much left to do so. It will be interesting to sample an identical image image and process it in both ways and than examine at 100% the differences in sharpness, distortion and quality.

  • JayVee

    JayVee, 2 months ago

    Thanks so much Steven.

Add your comment

You need to login or signup to add your comment to this work.