With all of the recent sound and fury on camera forums pertaining to the Sony A7r, and after many weeks of scrutinizing online samples, reading wildly varying opinions, and finally taking on extensive testing with an A7r of my own (both with native lenses and in combination with several different adapters) I found myself lost in an existential camera crisis. What is the true value of this camera? What purpose does it serve? Is it worth it?
Everyone of course will have their own opinions on how to answer such questions. But to borrow a phrase from Søren Kierkegaard, “I must find a truth that is true for me.”
With that, I decided to focus first on what occupies a large portion of my shooting: Canon’s TS-E lenses. When the Sony A7r was announced, there was quite a stir among those of us who use the marvelous TS-E optics (specifically the 17mm TS-E and 24mm TS-E II.) These are the two lenses that all these years have kept me tethered to the Canon system. Nikon’s D800 camera had a wondrous Sony sensor with more resolution and amazing dynamic range, but it wasn’t enough to convince me to give up my TS-Es. So I stuck with my trusty 5DII, certainly a capable camera, and decided to wait for the day that Canon would inevitably catch up in the sensor department.
But then came the A7r, and suddenly I was presented with the possibility to match my TS-E lenses to a Sony 36mp sensor—albeit with an adapter between them. The downside is that by adding more lens-to-camera interfaces, you increase the likelihood of misalignments of the lens to the sensor plane. As Roger Cicala brought to the attention of many photographers, There is no Free Lunch. (Or, as I was thinking as I was shooting the photo at the top of this post, There is no Free Parking.) The question is how noticeable will those misalignments be in real world photos?
I had that nagging concern, but still, the temptation was too great. (And perhaps just as crucially to my purchasing decision, it had been far too long since I had bought a shiny new piece of technology.) So armed with an A7r and a Metabones adapter, I decided to transpose my existential questions into practical questions. If I mount a TS-E lens to my A7r, will I get better results? Are there any circumstances I will ever get a worse result?
What I did was shoot scenes with my 5DII and A7r, and compare the results by upscaling the 5DII files (to see where the A7r files are better) as well as by downscaling the A7r to the 5DII files (to check if the A7r files were ever worse). Yes, I know some people will say “It’s not fair to the 5DII to interpolate the images.” But that’s the whole point in my view of comparing it to a camera with more resolution. It’s not supposed to be a level playing field. In theory, the higher megapixel sensor is supposed to deliver better results than the camera you have to upscale. I wanted to know if using the adapter would wipe out the benefit of the extra resolution.
After a number of tests I’ve come to the following conclusions:
- There is indeed a discernible gain in resolution when using the A7r. It may not be huge, but it is noticeable, especially in the center.
- The resolution gain is still apparent even when the lenses are shifted.
- The difference in resolution starts to diminish as the aperture gets stopped down and diffraction creeps in, but more resolution is still noticeable in my opinion at least up to f/13.
- The corners will get soft on the A7r, but they do not seem to ever get worse than the 5DII, even when shifted (and in a couple of circumstances even stayed very slightly better). The discrepancy between sharp center to soft corner is more noticeable on an A7r (and more so at wider apertures)…but of course if this affects “real world” photos depends on the level of detail in the corners.
- When the A7r files are downsampled to 5DII size, in some circumstances the resolution benefit can still be noticed.
- The A7r with the adapter is vulnerable to strange flare effects, likely due to internal reflections. (Masking the inside of the adapter can solve this issue but I haven’t had a chance to try this myself yet.)
Now comes the point where I present a few of the examples. First, we start with the 17mm TS-E unshifted, stopped down to f/13. I was interested if there would still be any resolution benefit after diffraction took its toll. The 5DII holds up pretty well, but the A7r does indeed show a little better. Full scene and crops follow below (click on the crops for 100% magnification).
Now here are the results at maximum shift (I shifted to the side instead of shifting up, since shifting up would have produced only empty sky with no detail for comparison). Center areas of the image are still showing better resolution on the A7r. Top left corner seems about the same; bottom left seems a little bit better with the A7r.
Now, here is a set with the 24mm TS-E, shot at f/8. At this aperture, the advantage of the A7r seems a little more pronounced.
In this scene, I did see some softness appear at the extreme edges of the frame, though again in absolute terms it doesn’t seem worse than the 5DII. This particular example shows how the A7r can go from noticeably more resolution to about the same as the 5DII, apparent in both 5DII upscaled and A7r downscaled scenarios:
As seen before in the 17mm samples, upper corners on the A7r become soft to the point it’s about the same resolution as the 5DII…the center of the frame stays sharper (for brevity only the downscaled to 5DII sized images are posted here):
So, have I resolved my existential camera crisis?
The resolution question for adapted lenses, I think I’ve at least put to rest in my own mind. Yes, the corners get soft with the wide angle TS-Es, but not soft enough to discourage me from using the camera. Yes, the relative softness to sharpness discrepancy is more than what I’d see using my native Canon body. But, the native body is also going to have less overall resolution, and its corners are going to be soft as well. If I were to go through my portfolio at 100% zoom on a computer screen, most of my photos would not look critically sharp in the corners, and this hasn’t bothered me before. So I don’t think this is really going to ruin any A7r images. The only real downside that irks me is the flare issue…but knowing that there is a relatively easy fix for that, I’m merely annoyed instead of stressed out by it.
And overall, I just have to comment that I really like using the A7r and working with its files. I don’t want to get into lengthy discussions on dynamic range or post processing at this point, since I just want to spend some more time working with more files in a greater range of scenarios … but I will say that my initial impressions are positive. In terms of overall image quality, I’m happy so far.