hubsand Offline Upload & Sell: Off
|
p.4 #2 · Sigma 50mm: Marmite Bokeh | |
Having a fascinating time comparing the Minolta 58/1.2 with the Pentax 50/1.4, Sigma 50/1.4 and the two fast Canon 50's . . . in truth, the lenses I have on loan from Canon (which I assume are therefore in tip-top condition) are superb. In fact, resolution-wise, there's little to separate any of these at any aperture. The Sigma resolves better in Zone C than any of them, but the 50L is absolutely fantastic in every way, and outperforms the Sigma everywhere apart from the corners at f1.4-f2.8 at distance.
The Sigma works better at distance than close up, whereas the 50L is perfect (optically) whatever you ask it to do.
These tests have been a good way of getting to grips with the bokeh question . . . I'm coming to the conclusion that 'good' or 'bad' bokeh is surprisingly elusive to define, and highly subjective. What remains as mysterious as the Mona Lisa's smile is the uncannily beautiful shapes drawn by the Minolta 58mm – a lens that 'technically' doesn't have textbook perfect bokeh, if we assume the benchmark to be 'buttery-ness'. And yet . . .
Both the Sigma and the Canon L have clearly been designed to give a specific rounded-aperture blade look to the defocused areas, so both are very smooth, but the Sigma is smooth to the point of being complete featureless. Is this a good thing? Well, if your intention is to isolate the subject from the background, then the Sigma will do the job better than anything else at this focal length . . . . but what if the background to the subject adds relevance or value to it? Surely then, a more detailed presentation is advantageous. It's hard to see how it can be quantified.
If we assume a completely 'blended' background to be the benchmark, the Sigma is 'better' than the Canon L or any other 50mm for bokeh. But in some circumstances the Canon L, or the Zeiss 50 will make a better picture. If you have both in your kit bag, you'd need a pretty intimate knowledge of how they both draw to deploy them to best effect. Detailed review coming . . .
|