Can you really spot the difference between an image taken at 20 mega pixels on my Pen-F or at 80 mega pixels on the same camera using Hi Res Shot or the original 20 mega pixel image scaled up to 80 mega pixels in Topaz Gigapixel AI?
Above is the original 20mp image taken on my Pen-F with my M.Zuiko 14-150mm F4-5.6 II set at 150mm taken at f5.6, 1/125th, ISO 640 and processed through Perfectly Clear with the Rich Colour preset.
Above is the same file (image) having been up-resed through the very latest version of Gigapixel AI v5.2.3, can you see the difference with the first image at the very top?
As before in the very top image this picture was also taken from the identical position on my Pen-F with my M.Zuiko 14-150mm F4-5.6 II set at 150mm taken at f5.6, 1/125th, ISO 640 but this time using the 80mp Hi Res Shot function and again processed through Perfectly Clear with the Rich Colour preset to match the look. Once more can you see any differences amongst the three picture in terms of their detail and resolution?
Well, here's the thing. In order to load the three images onto this blog, the images have firstly to converted into jpegs and secondly large images down sized to fit a preset limit so the big ones are only half their original size. Plus, sneakily, I have deliberately mixed the three up to leave you guessing despite the captioning! So can you really tell which is which? I doubt it. Don't bother.
What can I tell you from my own observations, in the real world, however? Hmm. On the 5120x2880 (5K/UHD+ - Ultra High Definition Plus) screen of my 27" iMac when I move from image to image displayed at full size I can see no discernible difference amongst the three, either as the full sized jpegs (before they were down sized to be loaded into this blog) nor when printed at A3+.
I also repeated the exercise with a 20mp raw file (ORF) developed as a 16 bit tiff again in Perfectly Clear and once more the results were up-reased to 80mp using Gigapixel AI
Once again can you see any difference between the two pictures displayed on a full resolution 5K screen or printed at A3+? Well once more you will need to take my word for it but no you cannot. This time the subject was moving in the wind so Hi Res Shot could bot be used - see below for an explanation.
So what was different, if it wasn't visual?
First Hi Rest Shot is great on a non moving subject as it joins multiple 20mp images together taken by moving the sensor around to make the final composite 80mp shot, so it has limited to use on subjects that are completely still as movement of any element of an image causes significant blurring of the moving part. You are therefore probably better off using Gigapixel, I think?
Second Gigapixel AI itself is only useful when you want to either make prints beyond A3+ up to A1 size or large (twice normal, whatever that means to you) crops of parts of your original image.
Third file sizes. The original jpeg image at 20mp was 13.3Mb whilst its up-resed 80mp image occupied 110.1Mb. The tiffs ended up being 121.4Mb for the original 20mp image and 242.9Mb for the 80mp up-resed version, similarly for the 80Mb Hi Res Shot images. These big file sizes for as single image are starting to get very silly, especially given that in normal use you can't make out any visual differences in the end results when comparing like with like i.e. Jpegs with each other and tiffs with each other.
Fourth processing power and time. Chucking all these big files around for 80Mb images needs lots of processing power, takes a lot longer and means ever bigger computers with oodles more disk space to boot = bundles of cash.
If you are considering an upgrade to something with a bigger megapixel count being lured by some nefarious marketing, something around the 50mp+ mark say? Don't bother. From these experiments I have concluded it is an expensive and pointless waste of time and money. To double the resolution of a current 20mp chip you would need 80mp anyway (the squares law applies, 2x resolution is 2x2 = 4x current number of mp) and you can achieve that at any time you like when needed in post with a £70 piece of software when needed and it avoids all that extra processing power, time, storage space and expense when you don't. Result.