LensTip.com

Lens review

Sigma 4.5 mm f/2.8 EX DC CIRCULAR FISHEYE HSM

24 November 2009
Szymon Starczewski

4. Image resolution

The first test category and first problems. Our charts are printed on special high definition foil. When testing normal lenses the lamps are positioned in such a way to prevent the light, which is partially reflected in the chart, fall back into the lens. With a circular “fisheye” it is simply impossible. We decided to position the lamps so the reflexes on the charts fall in those places that are not used for any measurements.

Dealing with a „fisheye” it is, of course, impossible to put the chart in such a position that it fills the lens’s field of vision entirely because the chart should have had infinite dimensions. As we strived to meet the ideal settings we decided to test the lens exceptionally on one chart only and we chose the biggest one we possess. Even in that case, though, during the test the lens was so close that its front element was a dozen or so centimeters from the centre of the chart…

There are huge deformations due to the lens’s projection characteristics and they cause one more problem. Compared to normal lenses, the resolution measurement areas move away significantly from the frame centre and this fact might cause further differences in comparison with other tests.

- - - - - - - - - - - - - - - - - advertisement - - - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Let’s get down to facts, though. The MTF50 values graph, measured on the strength of RAW files from a Canon 20D is presented in the picture below.

Sigma 4.5 mm f/2.8 EX DC CIRCULAR FISHEYE HSM - Image resolution

At first glance you can notice that, apart from the maximum relative aperture, the tested Sigma’s results are sensational, brushing record levels by f/5.6 and f/8.0. Can we trust our measurements, though? It seems that in reality the results could be several percent lower than those, presented on our graph. The result we got for f/16 can prove it – it amounts to 32 lpmm whereas for all wide angle lenses, tested on a Canon, the analogical result by that aperture value barely exceeded 30 lpmm. However even lowering the values, presented on the graph, by 1-2 lpmm wouldn’t change our very positive opinion about the lens’s frame centre performance.

The situation is a lot worse at the edge. If we take into account the extreme field of vision, though, we still should consider the results to be at least satisfactory.

Sigma 4.5 mm f/2.8 EX DC CIRCULAR FISHEYE HSM - Image resolution