Sony Alpha 7r comparison in low light

March 24, 2014  •  4 Comments

11 I've had the chance now to use Sony's 36 megapixel powerhouse for a few weeks now for a wide range of work from settings in studio to both indoor and outdoor events. I'm still convinced that it's the best camera available right now for me but it's not perfect. It's certainly a unique camera and it's hard to put into words how it compares to my experience using Sony's other cameras. I decided to do some tests between the cameras paired with my own observations to put some perspective into where this camera sits amongst it's peers.

For special event use, I'm done with big cameras. The cameras, batteries, accessories, and lenses are all smaller on mirrorless cameras and it makes a huge difference. I can carry a lot more for the same weight or go minimalist and carry next to no weight and still have the essentials covered. Even with the battery grip and a zoom lens, the 7r fits nicely in a shoulder bag with room to spare. Add to that jockeying for position with press photographers and a half a dozen TV cameras on tripods and I have the advantage. The only down side to the 7r for events is when the light is autofocus speed.

To demonstrate the low light performance of these 3 options, I set up a controlled test in my living room at night using a focus calibration target for AF speed and ISO performance. All three cameras were tested with the same lens where possible and with the Metabones Speedbooster on the NEX-7. A light meter was used to calibrate the amount of light to a specific level.

Autofocus Performance

To test autofocus, I set up a a Spyder Lenscal focus calibration target and an LED video light. I turned the light down until my Sekonic meter couldn't register the light anymore and returned an error. The last reading I received before the errors was 0.5 seconds, f/2.8, ISO 6400. I used an f/2.8 lens on all cameras and shot at the target to see what would happen. For the a99 it was a 28-70 Tokina zoom and for the E mount cameras, I used the 16mm f/2.8 pancake lens. I also tested the 7r with the 24-70 f/4 zoom. I was worried the pancake lens, notoriously soft wide open, would hinder the autofocus n the E-mount cameras but it was the only f/2.8 lens I could use to compare with the a99. The results from this test were as follows:

a99: Hit focus for an average of 1 in 10
NEX-5R: Hit focus for an average of 9 in 10
NEX-7: Hit focus for an average of 10 in 10
a7r (f2.8): Hit focus for an average of 10 in 10
a7r (f4): Hit focus for an average of 8 in 10

The a99 was hopeless in this level of light and would only hunt and rack back and forth until giving up. I was very surprised by this! I found the 5r to hunt a little but found focus almost every time. The NEX-7 and 7r were surprisingly fast and hit focus every time, fairly quickly, and without a lot of hunting. Even with an f/4 maximum aperture the 7r performed surprisingly well with what I considered to be an acceptable hit rate. I switched the 7r to the 55mm f/1.8 Zeiss and kept turning down the light. The focus kept hitting 100% of the time even after the light was turned down to it's minimum and turned to be facing away from the target about 12 feet away. Only when I turned the light off and the target was illuminated with only a tv across the room was the autofocus unable to lock on at all. Despite the light being so dim, I was still able to focus manually using the super bright electronic viewfinder. Something that would never be possible with an optical viewfinder. This test is completely unscientific and had the benefit of a high contrast target. With a real world targets, the results might have been different. I was hoping that my light meter would allow me to measure the light right down to the lowest levels but that wasn't the case. Below are two photos. One taken with the f/2.8 lens with the video light and one taken at f/1.8 in near blackness to give you an indication of how dim it was. They're dark because I was hand-holding the camera and needed some shutter speed plus I wanted to show how it looked to my eyes. Through the viewfinder, the target was bright as day albeit noisy and distorted somewhat.

7r - 16mm f/2.8 Pancake - 1/25 - f/2.8 - ISO64007r - 16mm f/2.8 Pancake - 1/25 - f/2.8 - ISO6400At this level of light (0.5 sec, f/2.8, ISO6400 as per Sekonic meter), the 7r was able to hit focus 100% of the time while the a99 was barely able to hit 1 in 10. 7r - 55mm f/1.8 - 1/50 - f/1.8 - ISO64007r - 55mm f/1.8 - 1/50 - f/1.8 - ISO6400This level of light was from a TV across the room. The exposure settings were within 1/3 of a stop of the previous test (one stop faster shutter and at f/1.8 instead of f/2.8) and still at ISO6400. Autofocus wouldn't lock on but manual focus was still possible.

Up until now I always assumed that phase detect autofocus was king but I was dead wrong. While it can be faster than contrast detect in most conditions, it's not always better. This was one of the reasons I was considering keeping the a99 but now there is no advantage in my mind. Especially since I was having focus accuracy issues with my a99...I've seen no such problems with my other cameras.

 

ISO Performance

Next I wanted to test ISO noise between these cameras. I didn't want to get carried away and shoot dozens of photos at different settings. All I wanted to accomplish was to show the level of noise at high (but not crazy high) ISO levels and throw the Metabones Speedbooster into the mix. The following photos were shot at ISO6400 on the a99 and the 7r and ISO3200 on the NEX-7. I used one of the best lenses I have which will fit on all three cameras, a modified Leica R 50mm Summilux f/1.4. This lens has had it's bayonet replaced to a minolta/sony bayonet. All three cameras metered the calibration target the same at 1/3 of a second at f/5.6. 

7r - 50mm - ISO6400As you can see ISO6400 looks very good on the 7r.

7r - 50mm - ISO64007r - 50mm - ISO6400 - cropped a99 - 50mm - ISO6400a99 - 50mm - ISO6400 - Cropped NEX7 - 50mm - ISO3200 with SpeedboosterNEX7 - 50mm - ISO3200 with Speedbooster - cropped

You can judge for yourself from the extreme crops of the bubble level on the base of the calibrator. I feel that the 7r looks very good at ISO6400. More importantly, the NEX-7 looks a tiny bit better to me at ISO3200 and with the Speedbooster than the a99 does at ISO6400. I always suspected this to be the case and they're certainly close but the NEX-7 is a very good APS-C camera and I knew that with the speed booster it could come close to full frame performance. The newer Sony mirrorless APS-C cameras like the a6000 have better ISO performance than the NEX-7 and very good autofocus in a package that's going to be less than $700 and will still fit in a coat pocket. I can't see any reason to keep the a99 on this front either when the performance combination of the NEX-7 and Speedbooster are just as good. I'm afraid this test is the nail in the coffin for my beloved a99. It's been a great camera and I've had no regrets owning it opposed to a Canon 5D mkII or mkIII but Sony is innovating in a way nobody can keep up with right now so I have to say goodbye.

 

Conclusion

There are many things to consider when comparing cameras for real world use and tests like this rarely tell the whole story. What these tests do provide however is an opportunity to dig deep into the finer details. This has allowed me to notice the little things that really make the difference in the end. I could write a whole post about it but some of these things include the ergonomics of the cameras, their weight, how I don't really miss not having a top LCD and buttons for ISO, drive, white balance, etc. in the 7r. The fact that the battery door on the 7r battery grip is flimsy but the ergonomics are clean and simple and much better than the complex set up on the a99. I'll likely miss the fully articulated screen of the a99 but it always felt flimsy to me as well...the screen on the E mount cameras all feel very solid.

I've read my fair share of reviews on these cameras as I always do before I buy but the only thing I've learned is that everybody is looking for something different. To some the shutter noise in the 7r is a problem for street photography and for others the flash sync speed is an issue for flash photography. Some don't like the 'plasticy' feel of the a99 and some don't like the tri-navi controls on the NEX-7...which I love BTW and am sad to see go away. All I can tell you is that this is the second time a Sony camera has caused me to stop, think, and immediately send my best, favourite, camera to KEH. I hope it doesn't become a habit but at the rate Sony is going, it won't be long before they've changed the game again.


Comments

Dave Tameling
Adam: I've always likened ISO performance to pixel size. Smaller pixels generally make for less than stellar high ISO performance. This is why the results surprised me so much. I didn't think 36MP could trump 24MP. Having shot with the 7r for a while longer now, I can honestly say that it's the best ISO performer I've ever used. There are of course cameras which are better but not in this size or price point or with so many pixels. I've shot some commercial work in a factory setting in a couple of instances, where ISO 6400 was required, shooting both stills and video, and the results were perfectly usable. I never felt happy with 6400 on the a99 regardless...it was an emergency go-to setting at best.

Unfortunately it was a poor test so the comparisons are especially subjective. I might be biased towards the NEX-7 a little as well...I've done a lot of great photography with that camera and it's only now starting to show it's age a little. ISO 3200 is certainly the limit of it's usable ISO range.
Dave Tameling
KURSAD: Thanks for the reply. I agree that phase detect should be better. The results surprised me. Now that I've been using the 7r for more time I can say that it's certainly slower and more likely to hunt than the a99 in nominal light levels. I'm very curious to see what the new 7s is capable of. It's autofocus is rated to function with 4 stops less light than the 7r.

As for the lens, if it were just the one lens, I'd agree with you...but to have issues in 4 different lenses isn't right and not a battle I had time to take on at the time. I disagree with you however that these Tokina's are low end. In the case of the f/2.8 zooms, when they were new, they were excellent lenses often used by professionals. Dated now due to better autofocus and lens coating, image and build quality was never an issue...these lenses were built like tanks and looked virtually new despite their age. Now that I'm 100% mirrorless I don't miss the bulk or weight. Sony's new 70-200 E-mount lens is a bit pricey at the moment so I'm going to pick up something manual focus. Likely the Nikkor f/4. Much smaller and lighter with little sacrifice in image quality and can be found for about 10% the cost of the Sony right now.
KURSAD(non-registered)
The focus accuracy on A99 may be because of the lens (Tokina in this case) quality. Different lenses tend to give different behaviors. With this in mind, one should also consider that Tokina is a low end lens.

PDAF is better than contrast detect AF, period. This is why all manufacturers are trying to implement it to their sensors nowadays.
Adam(non-registered)
Interesting test....
Though a few comments on the ISO test, where I think you are underestimating the A99:
Assuming you were shooting all the test shots from the same distance, the A99 is effectively more cropped than the A7r (because it has more resolution and is therefore a larger image). Under such circumstances, it's not surprising that the A7r would look even better. (I expect better noise performance out of the A7r regardless, but these crops exaggerate the difference).
I also can't agree that the NEX-7 looks better than the A99 -- I see a lot more chroma noise in the NEX-7 shot, and that is only at ISO 3200.
No comments posted.
Loading...

Archive
January February March April May June (5) July (1) August September October (2) November (3) December (1)
January February March April May June July August September October (1) November December (3)
January (2) February (1) March (2) April May (1) June July August September (1) October November (1) December
January (1) February March April May June July August September October November December
January February March April May June July August September October November December
January February March April May June July August September October November December