I believe I’ve written on this subject before. DPReview just exclusively announced the results of DxOMark testing of the Sigma 85mm f/1.4 “Art” lens, announcing it to have achieved a perfect 36 “perceptual megapixels” on the Nikon D810’s 36 megapixel sensor. It therefore sort-of-ties with the Sony FE 90mm f/2.8 macro lens, which achieved 42 perceptual megapixels on a 42 megapixel sensor.
Discussion immediately broke out on whether the Zeiss Otus 85/1.4 or Nikon 105/1.4 provided nicer bokeh and whether this was, in fact, more important than sharpness. The Zeiss, in particular, is almost tied with the Sigma for sharpness, and like it excels in many other technically measurable characteristics. The Nikon may be half a step behind, and in fact has a focal length of 94.8mm vs. the Sigma’s 79.9mm, so is more comparable than it sounds at first.
But back to my original point, which is the following:
That’s right, moiré. It is a physical certainty that if the sensor has no anti-aliasing filter on it, and the lens outresolves the sensor, you will see moiré on certain subjects (you can find technical details of moiré and aliasing on the Wikipedia pages, moiré pattern and Nyquist frequency). Many manufacturers have dropped the anti-aliasing filter to squeeze more resolution out of their images and as a small cost-saving, or installed a second filter that cancels the first (slight cost increase). Wisely, Canon added the 5Ds without the R to their line-up, which cuts back on moiré much better than the 5Ds R.
In Nikonland, it seems the highest resolving lenses and highest resolving sensors do not make a good pair. That being so, should we buy high resolution cameras at all, and what other choices do we have? Both the Nikon D600 and D610 have weak AA filters (link in Polish) – a bad idea for a lower-resovling camera as the range of lenses that outresolve it will be greater.
Pentax now releases its cameras with an AA filter that works up to shutter speeds of 1/1000s – presumably adequate for most portrait work, and hopeful that anything you’ll want to shoot above that shutter speed will be moving so fast that aliasing is not likely.
In spite of this, Pentax remains, as of this writing, a stalwart of compact lenses that value bokeh over resolution – partly, perhaps, owing to the fact that some of its lenses have been available for quite some time.
It’s well known that Tamron and Pentax have been sitting in a tree lately, and so it is fitting that Tamron somewhat recently launched a line of f/1.8 primes – more compact than the competition’s f/1.4 standard. DxOMark allows comparing Tamron’s 85mm lens with Zeiss’ Otus and Milvus, for instance, and there’s hardly a hair between them. Most surprising perhaps is the performance of the Milvus at less than half the price of the Otus.
Incidentally, while the three aforementioned lenses max out the D800E’s sensor at 36 perceptual megapixels, Sony’s new Gold Master 85mm lens reaches no more than the same value – 36 P-Mpix on an AA-filterless 42 megapixel sensor.
It’s worth asking therefore whether the trend for the second half of this decade is going to continue with lenses increasing in size, resolving power, and price, as exemplified by the Art and Otus, or a reconsideration of traditional values.* If the former, I hope consumers will be asking for strong anti-aliasing filters, and that camera makers, in spite of mobile phone cameras nipping at their heels, will grant that gift.
*An opportunity, perhaps, for Chinese lens makers trying to push into the market.
Update 7 February 2017: For differences between the Art and Otus, check out Roger Cicala’s blog post.