Nikon D70s real–life noise performance
When determining how a lens or a camera performs with respect to certain aberrations photographers generally like examining digital files on the screens of their computers at 100% magnification. Whereas this approach, of course, is legitimate, its major drawback is that it is rather difficult to reliably judge how what you see on the screen translates into real–life prints and whether a certain level of performance will be acceptable for the types of photographs that one intends to create with the piece of equipment under consideration. For one thing, if two different aberrations appear acceptable at 100% magnification on the screen it does not necessarily mean that both will be tolerable in actual prints.
Recently I needed to print several dozens photographs from files created with the Nikon D70s camera at different ISO settings. Considering the number of prints I thought that this was a good opportunity to see how the camera performs in terms of noise at different ISO settings in real life. I converted and post–processed all (RAW) files in the usual way (no noise–reduction software was used), recorded them onto CDs and took them to a professional Fuji laboratory where they were printed at 300dpi on Fujicolor Crystal Archive Paper for Frontier with the use of a Frontier350 minilab.
Quickly reviewing the prints for the first time had me quite impressed—knowing that many photographs were taken at relatively high ISO values I did not notice as much noise as I expected to see. I then inspected them more closely and did the following test: without referring to technical data or 100% on–screen magnification I examined the prints at normal viewing distance and divided them into three groups—where noise was invisible, where it was visible at close examination yet was not obtrusive and would not be noticed by non–photographers, and where it is obvious. I then checked at what ISO value each photograph was taken and got the following conclusion:
ISO200 through to ISO500—noise basically invisible;
ISO640 and ISO800—noise visible at close examination but is unlikely to be noticed by non–photographers;
ISO1250 and ISO1600—noise very visible.
(As any photographer would inevitably do I did check the files on the subject of noise during post–processing. Although this created certain expectations as to what I would see in actual prints, considering the number of photographs and the time gap between processing the files and reviewing the prints I simply could not remember all the details. From this perspective, this methodology was sufficiently objective for my purposes.)
This result was somewhat inconsistent with my previous impression that one should refrain from using ISO values higher then 400, which once again goes to say that only examining files at 100% magnification does not always suffice. And of course, the prints could be further improved by utilising noise–reduction software.
Photo on the left: taken at ISO800
100% crop of the area marked in red on the left; no sharpening applied.
By today's standards this image would normally be considered quite noisy. In real–life prints, however, the noise is almost unobtrusive.
There is a certain difference between examining files at 100% magnification and viewing real–life prints; although the former is a very useful approach photographers should exercise the latter as often as possible, too. As far as the D70s and real–life prints are concerned, I do not hesitate using the camera at ISO800 and lower ISO settings; ISO1250 and ISO1600 call for the use of noise–reduction software.
Related article: Nikon D70s—what could be further improved?