alex g
Mu-43 All-Pro
Yesterday I unguardedly spent a few hours endangering my sanity by running some performance tests on Four-Thirds lenses.
What follows is a comparison between the E-M1 and E-M1 mark II of maximum burst ("sequential shooting") rates achieved when used in conjunction with a selection of three lenses, each representing a different generation of digital AF technology. The object of the exercise is to attempt to determine the extent to which the E-M1 mk II improves the shooting performance of legacy Four-Thirds lenses, with respect to its predecessor.
It seems sensible to me to divide the procedure into two parts: first, an assessment of comparative shooting speed and, second, an assessment of comparative autofocus accuracy. This post deals exclusively with the question of speed.
For those interested, there are some details of the actual testing procedure at the end of the post, but for now, here are the main points:
Lens selection
The following lenses were tested:
Shooting Settings
Each lens was tested on both bodies in a controlled, indoor environment, with various combinations of shooting settings.
Drive modes
Burst rates were measured in each of the five principle Olympus sequential drive modes:
Autofocus modes
Modes 1, 2 and 3 (the low speed modes) were tested exclusively with the camera autofocus set to C-AF (continuous), while modes 4 and 5 were tested in both S-AF (single) and C-AF.
It's worth noting at this point that the E-M1 mk II differs from the original E-M1 with regard to the relationship between the high speed drive modes and continuous autofocus mode.
When C-AF is selected with the camera in either Sequential High or High Silent drive mode, the behaviours are as follows:
It's necessary to bear in mind therefore, when comparing the respective frame rates for C-AF in either of the high speed drive modes, that the autofocus behaviour will differ between the E-M1 and the E-M1 mk II.
IBIS and aperture settings
Two further factors were taken into consideration, both of them affecting the final frame rate to a greater or lesser extent determined by the type of lens under test.
Results
A couple of points regarding the test results: I estimate the experimental margin of error to be about 0.5 fps, with one exception: due to a limitation of the method used for measuring frame rates, it was not possible to determine definitive values for the E-M1 mk II's fastest drive mode. So wherever a value of 40 fps is given in the following results, it should be taken to mean "at least 40" — in the current context, the exact figure isn't crucial. It's a bit like the counting system which birds are believed to have evolved for the purpose of evaluating the progress of their brood and thus the effectiveness of their parenting strategy, which goes something like: none, one, two, lots. So long as they have lots, greater precision is pointless.
While I was able to confirm values of at least 40fps in these cases — the true value may have been considerably higher.
With so many test results, it's not obvious how best to present the data. The least confusing approach seems to have been to divide it into four categories, based on the following camera settings:
1. IBIS on, aperture stopped-down:
2. IBIS on, aperture wide open:
3. IBIS off, aperture stopped-down:
4. IBIS off, aperture wide open:
Conclusions
Overall, I think it's fair to say that overall, the E-M1 mk II does improve on its predecessor with regard to burst rates when used in conjunction with Olympus Four-Thirds lenses. The exact amount ranges — to use my earlier analogy — from 'none' to 'lots', depending on the lens model and camera settings.
Unsurprisingly, the most modest improvements are seen in Low Sequential and Low Anti-Shock modes, both of which rely on fast aperture and autofocus mechanisms in order to achieve decent burst rates. Improvements in performance are therefore much more evident in the case of the modern, µ43 lens than in those of its less up-tp-date, Four-Thirds brethren.
However, the situation improves in Low Silent mode, where both the SWD and non-SWD lenses pick up a fair amount of speed. As has been discussed here and elsewhere, the electronic shutter of the newer camera is a much more viable option than its predecessor, and is perfectly usable in all but a few, specific situations, so the two Silent drive modes are of more interest than they may have been in the past. Even the non-SWD 150/2.0 lens clocks a very respectable 8.5 fps in full C-AF with IBIS off and the aperture wide open — nearly three times as fast as on the original E-M1.
In terms of their respective impacts on fps, the costs of IBIS and stopping-down the lens aperture appear to be cumulative. Frame rates suffer most when both the IBIS is active and the lens is stopped down.
Note that it's possible to have the camera automatically turn IS off during sequential bursts, by setting Custom Menu -> C2 -> ⧉ Image Stabilization -> Fps Priority. That will maximize fps while at the same time giving you a stabilized image in the viewfinder for framing etc. Clearly this is only a good idea in situations where it's possible to set a decently fast shutter speed, thus reducing the need for stabilization.
Moving on to the Sequential High modes, we begin to see some serious frame rate improvements, whilst not forgetting that to an extent we're comparing apples and oranges when looking at the C-AF figures, since only on the original E-M1 does C-AF remain active during the actual burst. Interestingly enough, the fastest mechanical shutter drive mode with active C-AF is delivered by the original E-M1, clocking a useful 6.5 fps and 6 fps for the 50-200 SWD and 150/2 lenses respectively, when using C-AF in Sequential High drive mode, with IBIS turned off.
If you're happy to settle for S-AF during the burst, however, the E-M1 mk II achieves some truly stupendous frame rates, even with the ten-year-old lenses. Even with IBIS turned on, both lenses reach 10+ fps in both Sequential High and High Silent drive modes (although in the case of the 150/2, the aperture must be wide open). It's when you turn IBIS off that the fun really starts, however. Using the mechanical shutter, the 50-200 SWD delivers 14fps stopped down and over 15fps wide open, and the 150/2, though still limited to 6fps when stopped down, equals the performance of the 50-200 SWD when opened up.
The big surprise to me was the performance of the Four-Thirds lenses in High Silent mode — I totally did not expect to see them operate at the same bonkers frame rates as the latest µ43 lenses. The E-M1 mk II delivered at least 40 fps with both the 50-200 SWD and the 150/2 — even when either of them was stopped down. Very impressive.
I should mention one relatively minor point regarding performance with the Four-Thirds lenses — when using C-AF in either of the high-speed drive modes, the crazy high frame rates only kick in from the second frame of the burst. The interval between the first and second frames is rather longer, presumably to allow the lens to get its act together before gunning the throttle. I can't see this being an issue in all but the most demanding of situations. To be clear, this behaviour is only present in C-AF mode — when using S-AF, full frame rate is reached from the first frame of the burst.
Stepping back for a moment, I find the change in autofocus behaviour we see in the newer camera's high speed drive modes quite interesting, in that it favours a slightly different shooting strategy. Despite the fact that AF is locked for the duration of each burst, focus appears to be re-acquired very quickly if the camera is in C-AF mode, enabling you to both take advantage of the greatly improved C-AF and tracking performance of the E-M1 mk II to keep the lens "primed" and ready to shoot, and at the same time make use of the extreme burst rates to capture short but exceedingly dense clusters of images in response to the action as it unfolds. For example, two lions fighting may cover a fairly large area as they chase each other around, but the actual moments of contact — those which are likely to yield the best images — are typically short, intense frenzies of swiping paws, claws and snapping jaws, which despite their violence are relatively static with respect to the camera position. Each frame of a half-second, twenty shot burst fired off at such a moment will likely capture a different facial expression or aspect of the action, but the focus distance will probably not change enough for the fact that it's locked to be an issue.
I think that about wraps it up - please shout if you spot any glaring errors or evidence of gross stupidity.
Testing Notes
For reference, here are a few notes regarding the test procedure.
Each burst of frames was shot handheld with the camera facing a LensAlign target at a distance of about 3 metres. The lens was initially focused on the top (furthest end) of the sloping LensAlign ruler and, once the burst had started, panned vertically downwards so as to reach the bottom (nearest end) of the ruler about three seconds later. By inspecting the recorded images, it was possible to confirm that the autofocus had been doing roughly what it was supposed to and not losing the plot entirely. Each sequence of images was then divided into groups of equal timestamp (in other words, one group per second of elapsed time), and the first and last (potentially incomplete) groups were discarded. Finally, the frame rate was obtained by counting the members of the (typically) two remaining (complete) groups, and taking the mean value.
Camera settings:
What follows is a comparison between the E-M1 and E-M1 mark II of maximum burst ("sequential shooting") rates achieved when used in conjunction with a selection of three lenses, each representing a different generation of digital AF technology. The object of the exercise is to attempt to determine the extent to which the E-M1 mk II improves the shooting performance of legacy Four-Thirds lenses, with respect to its predecessor.
It seems sensible to me to divide the procedure into two parts: first, an assessment of comparative shooting speed and, second, an assessment of comparative autofocus accuracy. This post deals exclusively with the question of speed.
For those interested, there are some details of the actual testing procedure at the end of the post, but for now, here are the main points:
Lens selection
The following lenses were tested:
- The Olympus 150/2.0 — a good example of first gen Four-Thirds lens technology. It uses a micromotor to drive the AF, and has a relatively slow, heavy iris mechanism.
- The Olympus 50-200/2.8-3.5 SWD — a popular example of second gen Four-Thirds lens tech. It has an ultrasonic AF motor instead of the older micromotor, and a somewhat faster aperture design.
- The Olympus 40-150/2.8 PRO — this µ43 lens represents the current state of the art, by way of a control. This lens has an extremely fast, dual AF system and a a light and responsive aperture mechanism. Being roughly a decade younger than its brothers, its control electronics may also be assumed to be correspondingly more sophisticated.
Shooting Settings
Each lens was tested on both bodies in a controlled, indoor environment, with various combinations of shooting settings.
Drive modes
Burst rates were measured in each of the five principle Olympus sequential drive modes:
- Sequential Low (mechanical shutter)
- Low Anti-Shock (Electronic First Curtain shutter)
- Low Silent (fully electronic shutter)
- Sequential High (mechanical shutter)
- High Silent (fully electronic shutter)
Autofocus modes
Modes 1, 2 and 3 (the low speed modes) were tested exclusively with the camera autofocus set to C-AF (continuous), while modes 4 and 5 were tested in both S-AF (single) and C-AF.
It's worth noting at this point that the E-M1 mk II differs from the original E-M1 with regard to the relationship between the high speed drive modes and continuous autofocus mode.
When C-AF is selected with the camera in either Sequential High or High Silent drive mode, the behaviours are as follows:
- E-M1: exposure is locked for the duration of the burst, but focus is not. C-AF continues to operate.
- E-M1 mk II: exposure and focus are both locked for the duration of the burst.
It's necessary to bear in mind therefore, when comparing the respective frame rates for C-AF in either of the high speed drive modes, that the autofocus behaviour will differ between the E-M1 and the E-M1 mk II.
IBIS and aperture settings
Two further factors were taken into consideration, both of them affecting the final frame rate to a greater or lesser extent determined by the type of lens under test.
- Each combination of drive mode and autofocus mode was tested both with and without image stabilization (IBIS) activated. At higher shutter speeds, such as that used for the tests (1/1000 sec), IBIS is of debatable usefulness and since it has a negative impact on frame rate, it is often wise to turn it off. Added to which, the absurdly high frame rates which the E-M1 mk II is capable of are only available when IBIS is turned off.
- Each of the above combinations of settings was also tested both with the lens aperture set wide open, and with it stopped down a little. Shooting with the lens wide open has a significant impact on the maximum frame rates obtainable from legacy lenses, since it effectively eliminates the need for the camera to open and close the aperture blades between each frame of the burst.
Results
A couple of points regarding the test results: I estimate the experimental margin of error to be about 0.5 fps, with one exception: due to a limitation of the method used for measuring frame rates, it was not possible to determine definitive values for the E-M1 mk II's fastest drive mode. So wherever a value of 40 fps is given in the following results, it should be taken to mean "at least 40" — in the current context, the exact figure isn't crucial. It's a bit like the counting system which birds are believed to have evolved for the purpose of evaluating the progress of their brood and thus the effectiveness of their parenting strategy, which goes something like: none, one, two, lots. So long as they have lots, greater precision is pointless.
While I was able to confirm values of at least 40fps in these cases — the true value may have been considerably higher.
With so many test results, it's not obvious how best to present the data. The least confusing approach seems to have been to divide it into four categories, based on the following camera settings:
- IBIS on, lens stopped-down.
- IBIS on, lens wide open.
- IBIS off, lens stopped-down.
- IBIS off, lens wide open.
1. IBIS on, aperture stopped-down:
Subscribe to see EXIF info for this image (if available)
2. IBIS on, aperture wide open:
Subscribe to see EXIF info for this image (if available)
3. IBIS off, aperture stopped-down:
Subscribe to see EXIF info for this image (if available)
4. IBIS off, aperture wide open:
Subscribe to see EXIF info for this image (if available)
Conclusions
Overall, I think it's fair to say that overall, the E-M1 mk II does improve on its predecessor with regard to burst rates when used in conjunction with Olympus Four-Thirds lenses. The exact amount ranges — to use my earlier analogy — from 'none' to 'lots', depending on the lens model and camera settings.
Unsurprisingly, the most modest improvements are seen in Low Sequential and Low Anti-Shock modes, both of which rely on fast aperture and autofocus mechanisms in order to achieve decent burst rates. Improvements in performance are therefore much more evident in the case of the modern, µ43 lens than in those of its less up-tp-date, Four-Thirds brethren.
However, the situation improves in Low Silent mode, where both the SWD and non-SWD lenses pick up a fair amount of speed. As has been discussed here and elsewhere, the electronic shutter of the newer camera is a much more viable option than its predecessor, and is perfectly usable in all but a few, specific situations, so the two Silent drive modes are of more interest than they may have been in the past. Even the non-SWD 150/2.0 lens clocks a very respectable 8.5 fps in full C-AF with IBIS off and the aperture wide open — nearly three times as fast as on the original E-M1.
In terms of their respective impacts on fps, the costs of IBIS and stopping-down the lens aperture appear to be cumulative. Frame rates suffer most when both the IBIS is active and the lens is stopped down.
Note that it's possible to have the camera automatically turn IS off during sequential bursts, by setting Custom Menu -> C2 -> ⧉ Image Stabilization -> Fps Priority. That will maximize fps while at the same time giving you a stabilized image in the viewfinder for framing etc. Clearly this is only a good idea in situations where it's possible to set a decently fast shutter speed, thus reducing the need for stabilization.
Moving on to the Sequential High modes, we begin to see some serious frame rate improvements, whilst not forgetting that to an extent we're comparing apples and oranges when looking at the C-AF figures, since only on the original E-M1 does C-AF remain active during the actual burst. Interestingly enough, the fastest mechanical shutter drive mode with active C-AF is delivered by the original E-M1, clocking a useful 6.5 fps and 6 fps for the 50-200 SWD and 150/2 lenses respectively, when using C-AF in Sequential High drive mode, with IBIS turned off.
If you're happy to settle for S-AF during the burst, however, the E-M1 mk II achieves some truly stupendous frame rates, even with the ten-year-old lenses. Even with IBIS turned on, both lenses reach 10+ fps in both Sequential High and High Silent drive modes (although in the case of the 150/2, the aperture must be wide open). It's when you turn IBIS off that the fun really starts, however. Using the mechanical shutter, the 50-200 SWD delivers 14fps stopped down and over 15fps wide open, and the 150/2, though still limited to 6fps when stopped down, equals the performance of the 50-200 SWD when opened up.
The big surprise to me was the performance of the Four-Thirds lenses in High Silent mode — I totally did not expect to see them operate at the same bonkers frame rates as the latest µ43 lenses. The E-M1 mk II delivered at least 40 fps with both the 50-200 SWD and the 150/2 — even when either of them was stopped down. Very impressive.
I should mention one relatively minor point regarding performance with the Four-Thirds lenses — when using C-AF in either of the high-speed drive modes, the crazy high frame rates only kick in from the second frame of the burst. The interval between the first and second frames is rather longer, presumably to allow the lens to get its act together before gunning the throttle. I can't see this being an issue in all but the most demanding of situations. To be clear, this behaviour is only present in C-AF mode — when using S-AF, full frame rate is reached from the first frame of the burst.
Stepping back for a moment, I find the change in autofocus behaviour we see in the newer camera's high speed drive modes quite interesting, in that it favours a slightly different shooting strategy. Despite the fact that AF is locked for the duration of each burst, focus appears to be re-acquired very quickly if the camera is in C-AF mode, enabling you to both take advantage of the greatly improved C-AF and tracking performance of the E-M1 mk II to keep the lens "primed" and ready to shoot, and at the same time make use of the extreme burst rates to capture short but exceedingly dense clusters of images in response to the action as it unfolds. For example, two lions fighting may cover a fairly large area as they chase each other around, but the actual moments of contact — those which are likely to yield the best images — are typically short, intense frenzies of swiping paws, claws and snapping jaws, which despite their violence are relatively static with respect to the camera position. Each frame of a half-second, twenty shot burst fired off at such a moment will likely capture a different facial expression or aspect of the action, but the focus distance will probably not change enough for the fact that it's locked to be an issue.
I think that about wraps it up - please shout if you spot any glaring errors or evidence of gross stupidity.
Testing Notes
For reference, here are a few notes regarding the test procedure.
Each burst of frames was shot handheld with the camera facing a LensAlign target at a distance of about 3 metres. The lens was initially focused on the top (furthest end) of the sloping LensAlign ruler and, once the burst had started, panned vertically downwards so as to reach the bottom (nearest end) of the ruler about three seconds later. By inspecting the recorded images, it was possible to confirm that the autofocus had been doing roughly what it was supposed to and not losing the plot entirely. Each sequence of images was then divided into groups of equal timestamp (in other words, one group per second of elapsed time), and the first and last (potentially incomplete) groups were discarded. Finally, the frame rate was obtained by counting the members of the (typically) two remaining (complete) groups, and taking the mean value.
Camera settings:
- Release Priority in C-AF was set to 'On'. A few random spot tests were made with Release Priority turned 'Off', and the impact on frame rate under the test conditions was found to be negligible.
- The display frame rate on both bodies was set to "Normal".
- Images were saved as RAW files only. In the case of the E-M1 mk II, images were saved to both SD card slots simultaneously. [As a sidenote: during one session, the latter shot 3281 frames on a single battery charge, completely filling a pair of 64GB cards.]
- Shooting parameters were as follows: aperture priority mode; lens aperture either wide open or stopped down one stop, as required; s/s 1/1000 second; Auto ISO (ranging from about 500 to 2000). The two zoom lenses were both set to a focal length of 150mm, to match that of the 150/2.
- Periodic breaks were taken for the purpose of consuming revitalizing and sanity-preserving beverages.
maybe you should see if decreasing sanity alters the flow of time and effects the frame rates.
Please report any findings you think may be of interest if you get a chance — you can never have too much data! 
