3 Way Holiday Camera Shootout Part 6, Performance and Speed

D5200, 18-200mm lens. You can see the orange in this photo is bolder than in the photo below. The one below is probably more accurate.

G6, 14-140mm lens. Different aspect ratio from the M43 camera. Apart from this and the different color rendition I see little substantial difference between these photos.

 

Published Camera Tests  Many published camera tests just regurgitate manufacturer specifications and present this as an evaluation of camera performance. For instance I was recently reading a comparison between two cameras, one of which could shoot continuously at 9 fps, the other at 5.3 fps. The faster one was declared "better". But the reviewer just got this information from the spec sheets without testing to discover if 9 fps actually delivered any benefit over 5.3 fps.  Delving further in to the specs, it appears that both cameras claim to be able to do continuous AF, Continuous Drive at 4.2 fps. But again we have no idea how effective either camera might be in terms of the data which really matters, namely the percentage of frames in sharp focus.

Real World Performance Testing  I subjected all three kits to a series of tests, evaluating the relevant ones by checking the resulting photos at 100% on screen.

General remarks  All three cameras (Except the D5200 in Live View Mode)  delivered very good performance. I think most users would be well pleased with any of them. They each switch on and operate in a brisk and responsive fashion. A missed shot will rarely be due to any deficiency in the equipment.

Autofocus Speed  For this test I focussed back and forth between a near object and a far object, acquiring focus with a half press of the shutter button and waiting for the beep plus green AF area. I used the central area in each case. I did not fire the shutter.

Camera	Outdoors, dull overcast focus actions per second	Indoors low light Focus actions per second
D5200, OVF View	1.5	1.1
D5200, Monitor (Live) View	0.3	0.2
G6	2.2	1.3
V2	2.5	1.0

 Comment on Autofocus Speed   Outdoors the V2 is the AF speed champ, just beating the G6 by a small margin. The D5200 is by no means slow and will please most users but it can't keep up with the two MILC's.

Each camera slowed indoors but the V2 was slowed the most by low light. I believe this is because it switches from PDAF to CDAF in low light and Nikon CDAF is slower than Nikon PDAF.

Of the three the G6 was almost fastest outdoors and was the fastest indoors.

Turning to the D5200 in Monitor (Live) view mode we discover that the word "speed" does not apply in any sense at all. The camera switches from Phase Detect (PD) to Contrast Detect (CD) which does the "5 step dance" on every shot even when the subject distance has not altered. It goes eee...errr...eee...errr...beep (Yay! look at me, I did it !!) Presumably exhausted by this effort, the screen goes black for 2-4 seconds before the camera is finally ready to make the next shot. That might be just acceptable for landscape and similar work but the Monitor AF is error prone as well as slow. If this sounds bad it is better than the last DSLR which I owned which was a Canon EOS 60D. This thing did an eight step dance to achieve Live View focus and still made mistakes. I suspect that in most cases pre-setting focus distance by scale would be more satisfactory for Live View photography.

You may read opinions on the net by various photo sages that DSLR's focus faster than MILC's as if this were a statement of indisputable fact. My testing shows this is not so.

You may also read that CD autofocus is slower than PD autofocus. This is only true to the extent that CD AF in current model DSLR's  is slower than PDAF in those same cameras. CDAF in current model M43 cameras is very fast indeed, faster in fact than the PDAF in many DSLR's. It is also in my experience and testing, more consistently accurate.

D5200 with 18-200mm lens

G6 with 14-140mm lens

 

Shot to Shot Times  For this test I pointed the camera at a fixed target and in Single AF, Single Drive Mode, center AF Area, repeatedly pressed the shutter to make an exposure. The camera did not have to refocus between shots but it did have to confirm correct focus.

Camera	Shot to Shot time, seconds
D5200, OVF View	0.65
D5200, Live View	4.1
G6	0.5
V2	0.25

 

Comment on Shot to Shot Times  Like it's namesake in Peenemunde in 1944, the V2 goes like a rocket.  The G6 and D5200 in OVF Mode are by no means disgraced however and will suit most purposes just fine. The D5200 in Live View Mode is best suited to subjects firmly bolted down and immobile.

Continuous AF, Continuous Drive, Predictive Autofocus  For this round of tests I used cars driving along a suburban road and a person walking towards the camera at close range. The cars were in mixed light varying from sunny to shaded. The walker was in partly open shade  requiring a high ISO for adequate shutter speed.

The task for each camera/lens kit was to hold focus on the moving subject at 4 or 5 frames per second, and in the case of the V2, 15 frames per second.  The V2 has a huge buffer of 72 RAW frames at 5 fps. The G6 can manage 6 RAW frames before slowing, the D5200, 4 frames. So I used Fine JPG capture for all tests. With JPG files the G6 and D5200 could run off 50+ frames per run, more than enough for my purposes. I used single center AF Area, AF Continuous and a 95 MB/Sec card.  The D5200 and V2 were set at 5 fps, the G6 at M Burst which gave 4 fps. I ran the V2 at 15 fps in a separate test. The V2 is as far as I am aware the only camera at any price which can do predictive AF at 15 fps.

I set each lens at or near it's longest zoom position. I used the eye level viewfinder for all shots.

I used cars and the walker to be reasonably sure that each camera had the same task to perform in the same conditions, making the test fair. I did not test sports as this would introduce many uncontrolled variables.

I downloaded the resulting many hundreds of files and inspected each at 100% on the computer screen. I rated each as being Sharply in focus, Just out or Unsharp.

The User Experience  In reasonably bright light outdoors, the most pleasing camera to use in Continuous AF is the V2.  It has a muted shutter sound. The EVF appearance is that of continuous streaming video even though the camera is actually making stills. There is no perceptible EVF blackout. The image preview appears steady in the EVF.   The RAW buffer is huge. If you want, the camera can run silently at 5 or 15 fps with AF on every frame.

Next best to use is the D5200 which operates just like other DSLR's with a short blackout between each shot as the mirror flips up and down.

The G6 has the longest viewfinder blackout after each frame. It is usable and with practice the subject can be maintained in frame easily enough. It would be nice however if the Panasonic guys got the system in the V2.

Results 

Camera	Cars, Frames sharply in focus average over several runs	Walker, Frames sharply in focus. Average over several runs
D5200, 5fps	86%	95%
G6, 4fps	85%	93%
V2, 5fps	66%	76%
V2, 15 fps	65%	Not tested

 Comment on Predictive AF Results  Each of these consumer level cameras performed surprisingly well at predictive AF. All three cameras got a majority of frames in sharp focus.

The V2 is the absolute speed king however when we look at the percentage of frames in focus the other two perform a little better.

I found no significant difference between the D5200 and the G6 both of which performed at a level which would have been considered excellent in an expensive professional DSLR just a few years ago, and maybe even today.

You may read in the online photographic commentariat that

a) DSLR's do predictive AF better  than MILC's

b) MILC's are useless for predictive AF

c) Phase Detect AF is better for predictive AF than Contrast Detect AF

Each of these assertions is incorrect as to fact.

I think what is correct is that

a) Contrast Detect AF in DSLR's with PDAF optimised lenses is unsatisfactory for predictive AF. These lenses have  helical focussing mechanisms which were once considered fast but are really very slow compared to the direct drive AF mechanisms in modern M43 lenses.

b) The latest MILC cameras and lenses, particularly those for the M43 system can do predictive AF very well indeed. However M43  EVF refresh technology needs to be upgraded  to match that of the Nikon 1 system.

Performance Ranking  This is a bit difficult to determine due to the mixed results achieved.

The V2 is the nicest to use and is capable of the fastest shot to shot times and frame rates in good light outdoors where most predictive AF work will be done. But the percentage of sharp frames is not up to the other two.  In addition its speed drops substantially in low light indoors.

The D5200 works well and delivers a good result in all the test conditions (with eye level viewing).

The G6 does predictive AF  much better than some commentators and testers would have you believe. Some of these testers may have set the Autofocus  Mode on M43 cameras to "AF Tracking" which does NOT produce effective predictive AF.  Continuous AF is available with the monitor swung out, a handy feature not available on the other two.

In the end I have decided not to call a winner of the Performance /Speed criterion. Each does a good job but each has a different spectrum of capabilities.

The very fact that I am not able to call a winner with any confidence is actually very significant. Two years ago any mid range DSLR from Canon or Nikon would have easily outperformed any MILC.  It is intriguing to see how the march of technology has altered the balance between DSLR's and MILC's, or at least the Nikon 1 and M43 systems.

I have little doubt that MILC technology and performance/speed will continue to advance. But the DSLR has just about come to the end of it's evolutionary journey.  The main arena for performance advances in DSLR's lies in the Live View side of their operation. But any advances in that direction will potentially be of even more use to MILC's. I am thinking of Canon's incorporation of PDAF on every pixel of the imaging sensor, being introduced in Australia as I write on the EOS 70D. Canon wants it to improve smooth continuous AF during motion picture recording. But if the idea really works well why bother with all the mechanical paraphenalia of a DSLR ?  It might be even more suitable for a MILC.  

Next: Ergonomics