Enlarging Digital Files

mickyates Digital, Mick's Photo Blog, Nikon, Photography, Print, Processing, Software Leave a Comment

I was asked to take a few prints along to the Frome Wessex Camera Club last Friday. Phil Taylor was making an excellent presentation on ‘seeing’,  as opposed to over-fixation with technology. He had liked a couple of the archive images I had posted using early digital cameras, and asked me to show some prints (and the cameras) to add to his presentation.

In 1999, around the time I bought Nikon’s landmark D1 professional digital camera, I also had a Coolpix 950, a consumer breakthrough. The former boasted 2.7 megapixels, with a  file size of 2000 x 1296 and the latter 1.92 megapixels, with a file size of 1600 x 1200. The other camera we had, Kodak’s DC290 had 2.11 megapixels, and a  file size of  1792 x 1200.

The E950 was a quirky design, but a clever one as the rotating body allowed the use of a good internal zoom in an overall thin body footprint. It also had an excellent range of features, and is still quite sought after on EBay today.

Mick Yates. 2020. Kodak DC290 & Nikon Coolpix E950

Early 2000, I had taken some images in Mumbai with the E950, and when Phil asked me for prints, I was in the process of going though the original files and processing them. Sadly, I had only shot in JPEG, although the camera is capable of high quality TIFFs, essentially RAW. Each file was in the range of only 400 – 800 KB in size, tiny by today’s standards. Yet the colour is superb, and there is a reasonable amount of detail.

Those early digicams all were set against Kodak’s colours as the ‘gold standard’, emulating their best films, and the Mumbai images needed no colour adjustment. After some tidying up in Lightroom on contrast, highlights / shadows etc, I exported the file to an enlarged size of 2400 px on the longest edge at high quality, yielding a 2.5 MB image.

Mick Yates. 2000. Mumbai. Nikon Coolpix E950

This was one of the prints I showed at the club. And here is another:

Mick Yates. 2000. Mumbai. Nikon Coolpix E950

These images have been reduced to 1000 px on longest side, for the web. Click on the image to see a larger version.

For the meeting, I took the 2400 px images and printed at A4. Club members were impressed with the quality and in particular the smooth graduation of the blues in the sky. Phil’s point had been proven, I think.

I then decided to do a  little more practical research on re-sizing of images.

Enlarging of images in Photoshop is usually very successful, with the ‘Preserve Details 2.0’ process in the latest software (when you click on Image/Image Size). I usually increase the DPI to 300, which increases the image size, and then choose the edge size that I want.  I took the files and enlarged to 3600 px on the longest side, the norm for my web site and subsequent printing. This yielded a 3.5 MB file. In previous years I would have used Genuine Fractals software, but the new Photoshop process is extremely good.

I also used Topaz Gigapixel, to create a file of similar size. The Topaz suite claims to use AI, though I suspect that it is really just some very sophisticated machine learning algorithms. I don’t expect Topaz to become sentient! The Auto setting on Topaz is usually good enough, saving to best quality JPG, although you can also manually fine tune both noise suppression and removal of blur which is especially useful with older images of uncertain heritage. You can zoom in and see the side by side effects, although rendering can take a while.

Here is the side-by-side comparison.

Mick Yates. 2000. Mumbai. Nikon Coolpix E950

Whilst Gigapixel does accentuate the lens’s chromatic aberration (see the two birds, on the left), the detail in the brick work is indeed better than Photoshop. Click on the image to open a larger version of the test. If I was going to print this, I would use Photoshop to reduce that aberration effect.

It would seem to me that we should not be afraid of enlarging ‘old’ images with modern software, and it is amazing the results that can be achieved. And, used judiciously. I recommend Gigapixel.