Follow up to the "focal reducer vs. full frame" article (plus another take on Fuji RAW conversion)
The latest article caused, it seems, huge interest among readers, reaching thousands of hits in less than a couple of days. It also raised a few questions and I thoughts some clarification is due. Here goes:
1) Let me repeat that the article was in no way a comparison between camera brands, neither between camera systems. In my mind, it's totally clear that each system has its strengths and weaknesses. I remarked that I'd prefer Fuji, because of a number of features I find exciting about the system; the ability to easily use manual lenses with such an adapter being only one of them.
2) A focal reducer will not magically transform your crop-sensor camera into a full-frame one. Besides the obvious -you can primarily use them in manual mode, AF enabled examples are very limited in number and performance- there are other considerations. As far as crop sensor cameras have gone, they apparently can't match same generation/technology FF ones in a number of areas. Under the right conditions, the FF camera should have the potential for broader DR and color depth, as well as a lower noise signature at higher ISO settings. Please notice the careful use of the term "potential"; indeed this topic of "potential" vs. "practical" IQ is something I'll try to cover in a future blog. But the above statement on FF virtues still stands.
3) An advantage of the way a focal reducer works, may have escaped some readers. I'll try to elucidate on how this works. A focal reducer virtually offers complete practical equivalence with the larger format.
For the sake of argument, let's suppose we have two systems, one FF, one APS-C, which have the exact same metering system and are of the same tech generation. The first part is important, as we shall see. In practice, it doesn't work like that, but it's close enough for our discussion.
For a particular photo, a FF camera may shoot with a 50mm lens at f/4, 1/100" and ISO 400.
To match this, the APS-C camera with a FR, works like this:
The lens "becomes" ~35mm in FoV (this varies between FR models, by a couple of mm), which, multiplied by the crop factor is equivalent to 53-54mm in FF; close enough for all purposes, if we need to match the frame, from (almost) the same distance.
The lens gains one stop of light transmission, which makes it equivalent to a f/2.8 lens, in "APS-C terms". The DoF is also that of a "APS-C specific" f/2.8 lens. This is important to remember. Again, multiplied with the crop factor, the lens has the same DoF as a f/4 lens on FF.
The trick is, the gain in light gathering still stands. As a result, our crop sensor camera will need either one stop higher shutter speed (1/200", in our example) or, more importantly, one stop lower ISO (ISO200 in the same example). This, I hope, explains why I noted that I used ISO200 for the Sony in the original article, since it wouldn't get high enough shutter speeds in all cases, if I didn't.
As a result, in a number of cases, the crop sensor camera can eliminate most of the ISO advantage a FF camera should have over it (which, typically, is closer to one stop, between same sensor technology cameras). This depends on a number of parameters, but I hope you get the general idea.
The issue of Fuji RAW conversion was brought forward once again in the article. Blog reader Hans Aspenberg took the time to contact some experiments on the original files from the article on RAW conversion (thanks mate!).
For starters, he used the latest version of Lightroom, to see if things improved over earlier versions:
Short answer: no, they haven't. At least not spectacularly so.
Here is one of his efforts on Iridient Developer 3.0.2 on the Mac, exported as TIF then imported into LR CC:
As he notes: "Quite different. Lot more details, but also a bit "harder" in its expression. I'm not sure whether I like it.". I tend to agree, although, from what I've seen in other Iridient conversions, careful manipulation may give excellent results, so it obviously depends on textures, shadow detail, etc, of the scene. Definitely seems to resolve more "real" detail even with default settings.
Finally, here is a take using RAWTherapee:
As per my original findings, Hans seems to favor the amount of detail and general rendering, although the program remains pretty demanding to use.
I also agree with his final conclusion: for most purposes, LR is "good enough" (I'd say even better than that), particularly if you consider its overall characteristics. Should one need the outmost in detail (e.g. for photos to be printed in larger sizes), then one of the other options discussed would (still) be a safer bet.