It's interesting it appears to be a whole a new design, as the existing Sony 50/0.95 and the equivalent 35mm/0.95 for APS-C mirrorless systems already perform extremely well
considering the limitations of such a fast optic. You can't open a lens up that far without sacrificing some technical quality, of course, but Mitakon capitalised on that by going for rendering rather than 'lab' quality, and the resulting lenses are really very good. (Especially the 35mm mark II; I bought Christopher Frost's copy off him and it's not been removed from the Fuji body I put it on since I got it.)
It's hard to imagine there is much more they could do to improve on the optical formula, and simply moving to the Canon mount doesn't
require new optics; they could have just stuck the Sony formula on the larger EF mount. I would guess that the larger front element and filter thread is to reduce vignetting, which is one of the few genuine faults of the existing designs, but that also makes the corners harder to master. Seems like a lot of work for a lens which I doubt anybody really expects to be a
technical A+ anyway.
For what it's worth, though yes these lenses are manual focus only and f/0.95 may seem daunting at first, it's really not
that much different from focusing an f/1.4 lens (you get diminishing returns on apertures after f/1.6, so f/1.4 to f/0.95 is only really about half a stop difference in terms of both depth of field and light gathering) and the other Mitakon lenses' focus rings are nearly perfect; the only manual lenses I've used with better focus rings are the top Zeiss lenses and the Samyang XP 85mm, and those are only
slightly ahead of the Mitakon 50 and 35. They are
far better for manual focus than any Canon, Nikon, Sony, Sigma, or Tamron lens.
Of course, this lens could turn out to be a huge mess, we don't know, but at least on paper and given how the current similar Mitakons have turned out, this is a lens that 50mm prime fans should be every excited for.
ahsanford said:
But please educate me for a moment here, as this phenomenon was unknown to me until the 85 f/1.4L IS was released. The D-shaped bokeh is less about the rear baffle and more about the sheer wide open aperture actual size being bigger than the mirror box, right? What in particular does the baffle have to do with this?
It's the mirror box. The rectangular shield on the back of this lens is actually to combat flare. Several Canon FD and early EF lenses also use this, and it does not interfere with the rendering of out of focus areas, as they are engineered so the image circle coming out is still more than large enough (and round).
The reason that cutting shapes out of card and the camera's own mirror box can interfere with the rendering is because they are blocking out light from the image circle. This shielding doesn't, or, at least,
shouldn't.
But as you're well aware,
the mirrorbox of Canons will clip the image circle anyway, so it's basically irrelevant. The only way to get a lens faster than f/1.4 on a Canon
without the mirrorbox clipping would be to move to a shorter focal length, something like 22mm or so if my rough, late-night maths is correct. 22/0.95 results in an aperture half the size of 50/0.95, and that should be able to
just about project an image circle
just escaping the mirrorbox. (Though the angle the edges would come in at would be so broad, I'd expect the corner quality to be shockingly bad.)
One advantage of going all the way to f/0.95, though, and especially with the even, 'easy' rendering that Mitakon use, is that everything gets so blurred-out you don't really make out any particular background shapes anyway. You'll only notice the clipped highlights if you're basically doing a YouTube-esque 'bokeh test' shot with christmas lights against a stark backdrop, and, well, I doubt many people are actually shooting that as part of their 'real' photos. Similar to how the Canon 50mm and 85mm f/1.2s are kind of awful lenses in strict testing, there's more to a lens than simply its ability to pass the disconnected, arbitrary testing of most reviewers. The rendering is far more important for lenses like this, and clipped highlight shapes haven't really been a problem for the Canon 85mm f/1.2, and that lens has been producing them for decades now.