What is the best m43 for focus tracking?

Started Mar 31, 2013 | Questions
Anders W Forum Pro • Posts: 21,468
Re: See this review..

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
clengman
clengman Senior Member • Posts: 1,964
Re: See this review..

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

It occurred to me after I posted that the E-M5 might record it even if the e-pl1 doesn't. You have the 40-150mm zoom right? Do you see subject distance recorded in EXIF for photos taken with that lens? Just curious now.

 clengman's gear list:clengman's gear list
Olympus PEN E-PL1 Olympus E-M1 Olympus M.Zuiko Digital ED 14-42mm 1:3.5-5.6 Olympus Zuiko Digital ED 70-300mm 1:4.0-5.6 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 +6 more
Anders W Forum Pro • Posts: 21,468
Re: See this review..

clengman wrote:

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

It occurred to me after I posted that the E-M5 might record it even if the e-pl1 doesn't. You have the 40-150mm zoom right? Do you see subject distance recorded in EXIF for photos taken with that lens? Just curious now.

I haven't checked with that particular lens. But it would surprise me greatly if it's any different. I have checked for a considerable number of lenses by now, and it's in the EXIF for everyone of them. So it's just a matter of what the body wants to know.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
peevee1 Veteran Member • Posts: 6,247
Re: Another true believer

Anders W wrote:

PerL wrote:

I mention the G5 vs D4 as just one example of many very odd results. I can say a few more: Panasonic GX1 - best of all cameras in AF-C? Panasonic G3H - worst of all m43? Anders took out the OM-D values, because he thought they were flawed. That is 3 out 5 cameras in the m43 group with strange results.

Lets go to the DSLRs. Best of all - Canon 650D? The super cheap Canon 1100D - very close to 5DIII and Nikon D800, better than all medium DSLRs, including Canons own? Lets look at the SLT Sonys. The top of the APS-C line, Sony A77, slower than a NEX 7? The results are all over the place, contradicting AF system sophistication and processing power.

And finally - m43 CDAF vs PDAF DSLRs. Judging from this, the low end m43 Panasonics are now in the class of pro or semipro DSLR cameras in AF-C performance (Nikon D4, Nikon D800, Canon 5DIII, Nikon D600) and also better than all Sony SLTs and capable DSLRs like Nikon D7000 and Canon 60D.

You mention that the test may be a special case where CDAF works better - AF-C head on. Well, here good DSLRs perform excellent - at least on real targets. This is a series from a Nikon D300S (green indicates sharp)

I am getting out of this discussion now since it leads nowhere, too much waste of time and energy.

I am afraid that it is indeed a waste of time to argue with you. Having dealt with objections like those you now raise for the umpteenth time already in my first post on the subject as well as in several later posts, I have now lost any hope of having you understand why there is a need for elementary statistical tools like averages, standard deviations or correlation coefficients when dealing with data of the present kind. If you are ever to pass Statistics 101, you need to find a more patient teacher.

I don't see any averages and standard deviations in the list you cited, no confidence intervals. That is why it looks like the result of a single try, or, frankly, just BS.

peevee1 Veteran Member • Posts: 6,247
Re: what is the best m43 for focus tracking?

clengman wrote:

peevee1 wrote:

clengman wrote:

JamieTux wrote:

Hey everyone, since selling my Canon gear to move to m43 I've been shooting with just one top quality body (I have a gf2 as well buits really lacking in dr for my style of shooting so one useful camera only scares me!) I had the money put aside to get the new improved AF all singing all dancing Oly when its announced but all seems quiet on that front at the moment.  So I'd like to get a complementary camera to go with my om-d.  If that is the top at the moment then I guess I will need to hire where needed until the new one is announced and make a decision then.

I've even been looking at the GH3 despite its size, but I can't find anyone willing to put their neck on the line and call a winner!

Thanks in advance for your thoughts!

I see the discussions of "focus tracking" a lot and I wonder when this is actually useful. As Mick, Anders and axlotl pointed out there's a big difference between C-AF (which seems to be very good in all the newer m43 cameras), and C-AF with tracking.

To take the commonly cited examples; sports, kids playing, and flying birds, why is the tracking part really useful? Why would you want to keep the camera absolutely still and rely on the camera to follow the subject. Seems to me that it's preferable to track the subject by panning the camera. As long as you can keep the AF box on your subject,

And this is a real problem often.

an E-M5 (or even an E-PL3 or G3 and all subsequent cameras from what I understand) should do just fine in C-AF mode at keeping whatever is in the focus box in focus.

C-AF with tracking seems like a separate issue. Whether it's a CDAF or PDAF camera it still requires some sophisticated software dedicated to object recognition. (i.e. determining that a particular patch of light projected on the sensor represents a discrete, permanent object and then following it even as it's color, luminance, shape and size change.) This is more a machine vision problem and not a focusing problem. Are there really any cameras that are good at that?

True, and PDAF is not necessary for object recognition. Just like you can recognize objects with one eye closed (without depth information), a proper algorithm can do it too. 3rd dimension can help a little in difficult conditions.

So you think it's kind of a depth map added to the 2D projection that helps with object recognition and tracking? I was just trying to think about it conceptually to get an idea why an SLR would have an advantage for tracking over a mirrorless camera. I can understand why cameras with PDAF would have an advantage for C-AF (though it seems like CDAF systems are definitely closing the gap.)

Yes. But I don't think it is even the biggest advantage decent Canon and Nikon DSLRs (and not ALL DSLRs) have in tracking. I heard a lot of info that other brands are not as good. Having low res low noise high frequency input from color array is one hardware help in addition to PDAF array. But I suspect the most lies in the long developed tracking algorithms, and that is where most advantage lies.

But if you develop a sensor with full-area binning, it is going to have better, lower noise signal than the color array. Hundreds of on-sensor PDAF pixels give better, more defined 3d info compared to few PDAF sensors in DSLRs. And algorithm development should be continued, I hope Oly and Pana have enough money to keep mathematicians and programmers working on tracking employed, despite the losses they suffer. But obvious delay in introducing of on-sensor PDAF compared to Nikon, Canon, Sony and Fuji is troubling (of course they need it least of all, and still better without it than anything but Nikon 1, but still).

But what is needed is massive pixel binning, turning output of 16 mpix sensor to something like 0.01 mpix first (similar to what color array on focusing sensors in DSLRs have). No camera processor can do object recognition on 24 Mbytes of noise raw data in real time yet even if a sensor could output that (and it cannot or even close, so line skipping is used, leading to low-res noisy picture). You would need a supercomputer for that.

Anders W Forum Pro • Posts: 21,468
Re: Another true believer
1

peevee1 wrote:

Anders W wrote:

PerL wrote:

I mention the G5 vs D4 as just one example of many very odd results. I can say a few more: Panasonic GX1 - best of all cameras in AF-C? Panasonic G3H - worst of all m43? Anders took out the OM-D values, because he thought they were flawed. That is 3 out 5 cameras in the m43 group with strange results.

Lets go to the DSLRs. Best of all - Canon 650D? The super cheap Canon 1100D - very close to 5DIII and Nikon D800, better than all medium DSLRs, including Canons own? Lets look at the SLT Sonys. The top of the APS-C line, Sony A77, slower than a NEX 7? The results are all over the place, contradicting AF system sophistication and processing power.

And finally - m43 CDAF vs PDAF DSLRs. Judging from this, the low end m43 Panasonics are now in the class of pro or semipro DSLR cameras in AF-C performance (Nikon D4, Nikon D800, Canon 5DIII, Nikon D600) and also better than all Sony SLTs and capable DSLRs like Nikon D7000 and Canon 60D.

You mention that the test may be a special case where CDAF works better - AF-C head on. Well, here good DSLRs perform excellent - at least on real targets. This is a series from a Nikon D300S (green indicates sharp)

I am getting out of this discussion now since it leads nowhere, too much waste of time and energy.

I am afraid that it is indeed a waste of time to argue with you. Having dealt with objections like those you now raise for the umpteenth time already in my first post on the subject as well as in several later posts, I have now lost any hope of having you understand why there is a need for elementary statistical tools like averages, standard deviations or correlation coefficients when dealing with data of the present kind. If you are ever to pass Statistics 101, you need to find a more patient teacher.

I don't see any averages and standard deviations in the list you cited, no confidence intervals.

I count on people's ability to count to the extent that they need to. But within limits, I try to give them a helping hand with the math. Here's one example from a prior post:

"Consider the fact that in the FNAC report of two years ago (the one where they tested the GH2 that you have personal experience with), the average value for the mirrorless camera tested was 16.25 meters. Two years later, that figure is down to 11. For MFT cameras specifically, the value two years ago was 13.5. By now, the average is down to 7.  In the meantime, the results for DSLRs haven't changed much. PDAF is a rather mature technology. CDAF has been developing rapidly in the last few years and is now catching up."

http://www.dpreview.com/forums/post/51207299

That is why it looks like the result of a single try, or, frankly, just BS.

That's what it may look like to people who can't, or don't want to, count, and therefore, as an unfortunate result, speak BS.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Anders W Forum Pro • Posts: 21,468
Re: what is the best m43 for focus tracking?

amtberg wrote:

Anders W wrote:

amtberg wrote:

Anders W wrote:

Second, I have excluded the results for the Olympus E-PM2 and E-M5 since the results indicate that the lab made the mistake of shooting both cameras in high-speed burst mode where they won't even try to AF between shots. The buffer has then filled rapidly, which explains the slow rate at the end of the series.

Seems like you disregard the bad result for two  cameras while ignoring a related problem in this whole test.  It wasn't just the EM5 and EPM2 where bad settings/buffer size affected the results.  Many of the cameras apparently ran out of buffer before closest approach.

What would the examples be and how do you deduce that they did?

I think that is giving some of the DSLR's artificially bad results.  And in general, the results are not as fine-grained as the numbers suggest insofar as there seems to be a +/- difference of at least 3-4m depending on when the burst was started and thus when the last exposure was snapped in relation to the object distance.

Again, on what basis do you draw that conclusion? Just want to hear how you reason before saying what I think about it.

More generally, there may well be a margin of error of the kind you suggest (although I'd say it is hard to tell how big it is) which is why I suggest looking at the broader pattern and not pay too much attention to the results for each individual camera. Preferrably, they should have run this test a number of times per camera and reported the mean as well as standard deviations.

But this is about the only systematic AF-C test that I have been able to come up with and I think that, if interpreted in terms of broader patterns, it does tell us something. Let me know if you or anyone else has a better data source.

I'm just looking at their diagrams showing when the photos were taken in relation to the proximity of the approaching focus target.  In many cases, especially with the DSLRs, the buffer seems to have filled up before the target got very close, so the frequency of pics slowed down.  If the last pic in the series was snapped at 20m, that's obviously not going to be an accurate reflection of that camera's C-AF performance.  Look at the Sony A77, for example.  The buffer ran out at about 19m, then one more pic was taken at about 15'.  Could it have AFd at 5'?  Who knows?  In the case of the 5D, the closest shot attempted was at about 8.5m.  With the 600D no pics were taken inside of 11m.

First, note that the reason why I excluded the results for the E-M5 and the E-PM2 was not that their buffers filled. As I pointed out from the outset, the reason that I excluded them is that they were shot in a burst mode where they won't even attempt to AF except for the first shot in the burst. This in turn means that the test doesn't, for these two cameras, test what it purports to test.

Second, note that the test is a joint test of AF-C and burst-mode (including buffer) capabilities, and that the outcome is measured in two different regards: Number of sharp shots and subject distance for the last sharp shot. These preconditions are the same for all cameras tested, whether mirrorless or DSLR. Further, if some of them do worse when it comes to subject distance because their buffer filled as a result of high fps, they are likely to do better with regard to the number of sharp shots fired than they would if the fps had been lower and the buffer never filled.

Third, I went through the entire set of cameras looking for instances where it seemed likely that the camera had been hampered at the end due to the buffer being fill. I found only two likely candidates: the Samsung NX1000 on the mirrorless side and the A77 on the DSLR side.

Fourth, you cannot assume, as you seem to do, that a lower fps at the very end is always due to a buffer-full condition being reached. As you approach the end of the test, it gets more and more difficult for the AF system to keep up, i.e., keep the subject in focus. As you know, the distance the AF mechanism has to travel for every meter the subject distance shortens is much, much larger at the end of the test than at the beginning. There are two different ways for the camera to handle that predicament: It can keep up the fps at the risk of the shots being unsharp or it can try firing at a slower rate or stop firing alltogether when it finds that it doesn't make sense to do so. As the results show, different cameras are programmed to behave in different ways here: Some keep firing but fire unsharp shots and others slow down or stop firing altogether.

In every case the closest attempt depends entirely on the timing of the burst.  So, for example, with the Nikon D800 the last pic was snapped at 4.5m and the second to last at 8m.  What if the last shot had been at 6m instead of 4.5m?  Might it have been in focus?  Maybe....

While there might well be a stochastic element involved here, which is why I prefer to look at these data in a statistical way, you strongly exaggerate that element when you claim that the closest attempt depends entirely on the timing of the burst. What we know in the case of the D800 is that its true result is somewhere between 8 and 4.5 m. I prefer to record, for this camera as for others, the last sharp shot. In this case, the result of 8 m is still sufficiently good to place the D800 towards the top end of the distribution. If all cameras had reacted by firing one or more unsharp photos at the end, an alternative method would have been to record a point mid-way between the last sharp shot and the next one. But since some cameras quit firing instead of firing unsharp shots, that method cannot be used across the board.

If the GX1's buffer had been started a hair later might not have taken that list pic and it could have scored an 8 instead of a 2.

The remarkable thing about the GX1 shot at two meters is not that it was fired but that it was sharp. The D4, D800, and D600 fire at pretty short distances too, but the shots they fire are no longer sharp.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Anders W Forum Pro • Posts: 21,468
Re: See this review..

clengman wrote:

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

It occurred to me after I posted that the E-M5 might record it even if the e-pl1 doesn't. You have the 40-150mm zoom right? Do you see subject distance recorded in EXIF for photos taken with that lens? Just curious now.

OK. So I did check with my 40-150/4-5.6 R, and, yes, focus distance is reported in the EXIF in that case too, just as with all other native MFT Oly and Pany lenses I have looked at.

One additonal thing to note here: There are actually two different measures of focus distance reported in the EXIFs. One is called FocusStepCount and the other FocusDistance. The latter is reported in meters (with up to three decimals). Presumably, the first is what is actually reported/measured and the second just a conversion. But one small mystery here is that whereas I have found the FocusStepCount to behave predictably (e.g., by reporting very similar values on repeated AF trials on the same target and with the mean value changing predictably as you change the focus distance), the FocusDistance value is somewhat suspect. For example, even if the FocusStepCount is unchanged from one shot to another, the FocusDistance value might differ. I have still to understand why this is the case. Perhaps it's just a firmware bug of one kind or another. But when I experiment with the AF system in a way that makes use of this information, it is the FocusStepCount I pay attention to for the time being.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Anders W Forum Pro • Posts: 21,468
Re: See this review..

clengman wrote:

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

It occurred to me after I posted that the E-M5 might record it even if the e-pl1 doesn't. You have the 40-150mm zoom right? Do you see subject distance recorded in EXIF for photos taken with that lens? Just curious now.

OK. So I did check with my 40-150/4-5.6 R, and, yes, focus distance is reported in the EXIF in that case too, just as with all other native MFT Oly and Pany lenses I have looked at.

One additonal thing to note here: There are actually two different measures of focus distance reported in the EXIFs. One is called FocusStepCount and the other FocusDistance. The latter is reported in meters (with up to three decimals). Presumably, the first is what is actually reported/measured and the second just a conversion. But one small mystery here is that whereas I have found the FocusStepCount to behave predictably (e.g., by reporting very similar values on repeated AF trials on the same target and with the mean value changing predictably as you change the focus distance), the FocusDistance value is somewhat suspect. For example, even if the FocusStepCount is unchanged from one shot to another, the FocusDistance value might differ. I have still to understand why this is the case. Perhaps it's just a firmware bug of one kind or another. But when I experiment with the AF system in a way that makes use of this information, it is the FocusStepCount I pay attention to for the time being.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Anders W Forum Pro • Posts: 21,468
Re: See this review..

clengman wrote:

clengman wrote:

Anders W wrote:

clengman wrote:

olliess wrote:

clengman wrote:

olliess wrote:

Anders W wrote:

olliess wrote:

I was asking about that exact post. You state that the system only shifts the sensor in rotation about the horizontal axis (which I take to be the optical axis). So does this system compensate pitch and yaw only using vertical/horizontal translation?

Pitch and yaw are corrected by shifting the sensor up-down and left-right. All in-body IS systems do this and work like that.

Thanks. I understood previous in-body systems to work this way, but I had wondered (from the name?) if the Olympus system actually corrected for multiple rotation axes. It seems not.

I don't think you're understanding correctly.

Sorry, I should have said "corrected using multiple rotation axes." You can see from the previous post that I was talking about pitch/yaw compensation (two of three rotation axes) via sensor translation.

Sorry. My mistake.

I think the biggest misunderstanding of the E-M5 IBIS is that it's the 5-axis part that makes it great. The additional two axes are only valuable in limited circumstances and in fact those corrections can't be applied with most lenses. The lens has to be able to communicate its current focus distance to the camera in order for X and Y-axis translation correction to be done right. None of the m43 lenses that I have do that and obviously no adapted lens will either.

Not disputing your general point that two of the five axes are of pretty limited importance most of the time. The main point is that the new IBIS is very efficient in correcting the other three.

I believe it. I'm looking forward to getting my hands on an E-M5 at some point down the line. It sounds like the improved IBIS will be nice to use.

But I have strong reasons to think you are wrong when you say that all/most native MFT lenses don't report focus distance to the body and therefore can't even enable correction for vertical/horizontal shift. If they don't do that, why would information on focus distance appear in the E-M5 EXIFs, as it does for native, electrically connected lenses (possibly adapted FT lenses too, although I am not sure about that since I don't have any)?

I didn't say anything about all or most of the native lenses. I can only comment on the two native, electronically-coupled lenses I own, the mkI 14-42mm zoom and the non-R version of the 40-150mm zoom. I'm sure that some lenses (maybe most? I don't know.) report this value.

I've looked for the focus distance in the EXIF data for pictures taken with both these lenses and I have not seen a value for subject distance recorded anywhere.

If I'm wrong I don't mind being corrected.

Note that I was talking about the E-M5 here. I am not saying every MFT camera retrieves the info on focus distance from the lens and reports it to the body. But, judging by its EXIFs, the E-M5 does. And the E-M5 is the only camera that needs that information for full IBIS functionality.

It occurred to me after I posted that the E-M5 might record it even if the e-pl1 doesn't. You have the 40-150mm zoom right? Do you see subject distance recorded in EXIF for photos taken with that lens? Just curious now.

OK. So I did check with my 40-150/4-5.6 R, and, yes, focus distance is reported in the EXIF in that case too, just as with all other native MFT Oly and Pany lenses I have looked at.

One additonal thing to note here: There are actually two different measures of focus distance reported in the EXIFs. One is called FocusStepCount and the other FocusDistance. The latter is reported in meters (with up to three decimals). Presumably, the first is what is actually reported/measured and the second just a conversion. But one small mystery here is that whereas I have found the FocusStepCount to behave predictably (e.g., by reporting very similar values on repeated AF trials on the same target and with the mean value changing predictably as you change the focus distance), the FocusDistance value is somewhat suspect. For example, even if the FocusStepCount is unchanged from one shot to another, the FocusDistance value might differ. I have still to understand why this is the case. Perhaps it's just a firmware bug of one kind or another. But when I experiment with the AF system in a way that makes use of this information, it is the FocusStepCount I pay attention to for the time being.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
peevee1 Veteran Member • Posts: 6,247
Re: what is the best m43 for focus tracking?

clengman wrote:

peevee1 wrote:

clengman wrote:

JamieTux wrote:

Hey everyone, since selling my Canon gear to move to m43 I've been shooting with just one top quality body (I have a gf2 as well buits really lacking in dr for my style of shooting so one useful camera only scares me!) I had the money put aside to get the new improved AF all singing all dancing Oly when its announced but all seems quiet on that front at the moment.  So I'd like to get a complementary camera to go with my om-d.  If that is the top at the moment then I guess I will need to hire where needed until the new one is announced and make a decision then.

I've even been looking at the GH3 despite its size, but I can't find anyone willing to put their neck on the line and call a winner!

Thanks in advance for your thoughts!

I see the discussions of "focus tracking" a lot and I wonder when this is actually useful. As Mick, Anders and axlotl pointed out there's a big difference between C-AF (which seems to be very good in all the newer m43 cameras), and C-AF with tracking.

To take the commonly cited examples; sports, kids playing, and flying birds, why is the tracking part really useful? Why would you want to keep the camera absolutely still and rely on the camera to follow the subject. Seems to me that it's preferable to track the subject by panning the camera. As long as you can keep the AF box on your subject,

And this is a real problem often.

an E-M5 (or even an E-PL3 or G3 and all subsequent cameras from what I understand) should do just fine in C-AF mode at keeping whatever is in the focus box in focus.

C-AF with tracking seems like a separate issue. Whether it's a CDAF or PDAF camera it still requires some sophisticated software dedicated to object recognition. (i.e. determining that a particular patch of light projected on the sensor represents a discrete, permanent object and then following it even as it's color, luminance, shape and size change.) This is more a machine vision problem and not a focusing problem. Are there really any cameras that are good at that?

True, and PDAF is not necessary for object recognition. Just like you can recognize objects with one eye closed (without depth information), a proper algorithm can do it too. 3rd dimension can help a little in difficult conditions.

So you think it's kind of a depth map added to the 2D projection that helps with object recognition and tracking? I was just trying to think about it conceptually to get an idea why an SLR would have an advantage for tracking over a mirrorless camera. I can understand why cameras with PDAF would have an advantage for C-AF (though it seems like CDAF systems are definitely closing the gap.)

Yes. But I don't think it is even the biggest advantage decent Canon and Nikon DSLRs (and not ALL DSLRs) have in tracking. I heard a lot of info that other brands are not as good. Having low res low noise high frequency input from color array is one hardware help in addition to PDAF array. But I suspect the most lies in the long developed tracking algorithms, and that is where most advantage lies.

But if you develop a sensor with full-area binning, it is going to have better, lower noise signal than the color array. Hundreds of on-sensor PDAF pixels give better, more defined 3d info compared to few PDAF sensors in DSLRs. And algorithm development should be continued, I hope Oly and Pana have enough money to keep mathematicians and programmers working on tracking employed, despite the losses they suffer. But obvious delay in introducing of on-sensor PDAF compared to Nikon, Canon, Sony and Fuji is troubling (of course they need it least of all, and still better without it than anything but Nikon 1, but still).

But what is needed is massive pixel binning, turning output of 16 mpix sensor to something like 0.01 mpix first (similar to what color array on focusing sensors in DSLRs have). No camera processor can do object recognition on 24 Mbytes of noise raw data in real time yet even if a sensor could output that (and it cannot or even close, so line skipping is used, leading to low-res noisy picture). You would need a supercomputer for that.

peevee1 Veteran Member • Posts: 6,247
Re: Another true believer
1

Anders W wrote:

peevee1 wrote:

Anders W wrote:

PerL wrote:

I mention the G5 vs D4 as just one example of many very odd results. I can say a few more: Panasonic GX1 - best of all cameras in AF-C? Panasonic G3H - worst of all m43? Anders took out the OM-D values, because he thought they were flawed. That is 3 out 5 cameras in the m43 group with strange results.

Lets go to the DSLRs. Best of all - Canon 650D? The super cheap Canon 1100D - very close to 5DIII and Nikon D800, better than all medium DSLRs, including Canons own? Lets look at the SLT Sonys. The top of the APS-C line, Sony A77, slower than a NEX 7? The results are all over the place, contradicting AF system sophistication and processing power.

And finally - m43 CDAF vs PDAF DSLRs. Judging from this, the low end m43 Panasonics are now in the class of pro or semipro DSLR cameras in AF-C performance (Nikon D4, Nikon D800, Canon 5DIII, Nikon D600) and also better than all Sony SLTs and capable DSLRs like Nikon D7000 and Canon 60D.

You mention that the test may be a special case where CDAF works better - AF-C head on. Well, here good DSLRs perform excellent - at least on real targets. This is a series from a Nikon D300S (green indicates sharp)

I am getting out of this discussion now since it leads nowhere, too much waste of time and energy.

I am afraid that it is indeed a waste of time to argue with you. Having dealt with objections like those you now raise for the umpteenth time already in my first post on the subject as well as in several later posts, I have now lost any hope of having you understand why there is a need for elementary statistical tools like averages, standard deviations or correlation coefficients when dealing with data of the present kind. If you are ever to pass Statistics 101, you need to find a more patient teacher.

I don't see any averages and standard deviations in the list you cited, no confidence intervals.

I count on people's ability to count to the extent that they need to. But within limits, I try to give them a helping hand with the math. Here's one example from a prior post:

"Consider the fact that in the FNAC report of two years ago (the one where they tested the GH2 that you have personal experience with), the average value for the mirrorless camera tested was 16.25 meters. Two years later, that figure is down to 11. For MFT cameras specifically, the value two years ago was 13.5. By now, the average is down to 7.  In the meantime, the results for DSLRs haven't changed much. PDAF is a rather mature technology. CDAF has been developing rapidly in the last few years and is now catching up."

http://www.dpreview.com/forums/post/51207299

That is why it looks like the result of a single try, or, frankly, just BS.

That's what it may look like to people who can't, or don't want to, count, and therefore, as an unfortunate result, speak BS.

Count what? Averaging mft cameras, very few of them, does not help. Even the parameter itself, the closest distance, is not very useful, but just with a single try is not useful at all. And in the light of things like using the cameras with locked focus and different shooting speeds leading to buffer overflow, just shows complete incompetence of the testers.

BTW, I am not a m43 detractor, I like E-M5 very much and often even shoot sports (gymnastics) with it. And know the limitations of its AF-C and AF-C+Tr very well by now, and how to get around them for best results. The discussion or tests of the performance, without mentioning specific lenses, light levels, apertures, picture modes, sizes of the subjects, and especially "Rls priority" settings are laughable as the parameters greatly affect the performance of AF-C and tracking.

Anders W Forum Pro • Posts: 21,468
Re: Another true believer

peevee1 wrote:

Anders W wrote:

peevee1 wrote:

Anders W wrote:

PerL wrote:

I mention the G5 vs D4 as just one example of many very odd results. I can say a few more: Panasonic GX1 - best of all cameras in AF-C? Panasonic G3H - worst of all m43? Anders took out the OM-D values, because he thought they were flawed. That is 3 out 5 cameras in the m43 group with strange results.

Lets go to the DSLRs. Best of all - Canon 650D? The super cheap Canon 1100D - very close to 5DIII and Nikon D800, better than all medium DSLRs, including Canons own? Lets look at the SLT Sonys. The top of the APS-C line, Sony A77, slower than a NEX 7? The results are all over the place, contradicting AF system sophistication and processing power.

And finally - m43 CDAF vs PDAF DSLRs. Judging from this, the low end m43 Panasonics are now in the class of pro or semipro DSLR cameras in AF-C performance (Nikon D4, Nikon D800, Canon 5DIII, Nikon D600) and also better than all Sony SLTs and capable DSLRs like Nikon D7000 and Canon 60D.

You mention that the test may be a special case where CDAF works better - AF-C head on. Well, here good DSLRs perform excellent - at least on real targets. This is a series from a Nikon D300S (green indicates sharp)

I am getting out of this discussion now since it leads nowhere, too much waste of time and energy.

I am afraid that it is indeed a waste of time to argue with you. Having dealt with objections like those you now raise for the umpteenth time already in my first post on the subject as well as in several later posts, I have now lost any hope of having you understand why there is a need for elementary statistical tools like averages, standard deviations or correlation coefficients when dealing with data of the present kind. If you are ever to pass Statistics 101, you need to find a more patient teacher.

I don't see any averages and standard deviations in the list you cited, no confidence intervals.

I count on people's ability to count to the extent that they need to. But within limits, I try to give them a helping hand with the math. Here's one example from a prior post:

"Consider the fact that in the FNAC report of two years ago (the one where they tested the GH2 that you have personal experience with), the average value for the mirrorless camera tested was 16.25 meters. Two years later, that figure is down to 11. For MFT cameras specifically, the value two years ago was 13.5. By now, the average is down to 7.  In the meantime, the results for DSLRs haven't changed much. PDAF is a rather mature technology. CDAF has been developing rapidly in the last few years and is now catching up."

http://www.dpreview.com/forums/post/51207299

That is why it looks like the result of a single try, or, frankly, just BS.

That's what it may look like to people who can't, or don't want to, count, and therefore, as an unfortunate result, speak BS.

Count what?

You remarked that you didn't see things like averages in the list I cited. I responded that I thought people would be able to compute those themselves.

Averaging mft cameras, very few of them, does not help.

Exactly why wouldn't it help to average, for example, mirrorless cameras, MFT cameras, and DSLRs and then compare those averages?

Even the parameter itself, the closest distance, is not very useful

Exactly why wouldn't that be useful?

, but just with a single try is not useful at all. And in the light of things like using the cameras with locked focus

What do you mean by locked focus and why wouldn't it be useful to lock it. In this test, the AF point is on the target throughout the test. What's wrong with that?

and different shooting speeds leading to buffer overflow,

See my response to amtberg here

http://www.dpreview.com/forums/post/51220164

where I deal with that matter in some detail.

How do you suggest the test should cope with the fact that cameras have different fps and different buffer sizes?

just shows complete incompetence of the testers.

So what data sources would you suggest we use instead?

And in what regards are their tests better?

BTW, I am not a m43 detractor, I like E-M5 very much and often even shoot sports (gymnastics) with it. And know the limitations of its AF-C and AF-C+Tr very well by now, and how to get around them for best results. The discussion or tests of the performance, without mentioning specific lenses, light levels, apertures, picture modes, sizes of the subjects, and especially "Rls priority" settings are laughable as the parameters greatly affect the performance of AF-C and tracking.

Apparently you didn't read the test description.

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
OP JamieTux Veteran Member • Posts: 4,079
Re: what is the best m43 for focus tracking?

Don't forget that PDAF does have one very strong trump card...  It knows whether something is in front of or behind the current focus point.  Otherwise I agree - simplify everything so that the computer has less work to do and it will be faster

 JamieTux's gear list:JamieTux's gear list
Nikon 1 J5 Fujifilm X-E3 Nikon 1 Nikkor 18.5mm f/1.8 Zeiss Touit 50mm F2.8 Nikon 1 Nikkor VR 10-30mm f/3.5-5.6 PD-Zoom +6 more
clengman
clengman Senior Member • Posts: 1,964
Re: what is the best m43 for focus tracking?
1

peevee1 wrote:

clengman wrote:

peevee1 wrote:

clengman wrote:

JamieTux wrote:

Hey everyone, since selling my Canon gear to move to m43 I've been shooting with just one top quality body (I have a gf2 as well buits really lacking in dr for my style of shooting so one useful camera only scares me!) I had the money put aside to get the new improved AF all singing all dancing Oly when its announced but all seems quiet on that front at the moment.  So I'd like to get a complementary camera to go with my om-d.  If that is the top at the moment then I guess I will need to hire where needed until the new one is announced and make a decision then.

I've even been looking at the GH3 despite its size, but I can't find anyone willing to put their neck on the line and call a winner!

Thanks in advance for your thoughts!

I see the discussions of "focus tracking" a lot and I wonder when this is actually useful. As Mick, Anders and axlotl pointed out there's a big difference between C-AF (which seems to be very good in all the newer m43 cameras), and C-AF with tracking.

To take the commonly cited examples; sports, kids playing, and flying birds, why is the tracking part really useful? Why would you want to keep the camera absolutely still and rely on the camera to follow the subject. Seems to me that it's preferable to track the subject by panning the camera. As long as you can keep the AF box on your subject,

And this is a real problem often.

an E-M5 (or even an E-PL3 or G3 and all subsequent cameras from what I understand) should do just fine in C-AF mode at keeping whatever is in the focus box in focus.

C-AF with tracking seems like a separate issue. Whether it's a CDAF or PDAF camera it still requires some sophisticated software dedicated to object recognition. (i.e. determining that a particular patch of light projected on the sensor represents a discrete, permanent object and then following it even as it's color, luminance, shape and size change.) This is more a machine vision problem and not a focusing problem. Are there really any cameras that are good at that?

True, and PDAF is not necessary for object recognition. Just like you can recognize objects with one eye closed (without depth information), a proper algorithm can do it too. 3rd dimension can help a little in difficult conditions.

So you think it's kind of a depth map added to the 2D projection that helps with object recognition and tracking? I was just trying to think about it conceptually to get an idea why an SLR would have an advantage for tracking over a mirrorless camera. I can understand why cameras with PDAF would have an advantage for C-AF (though it seems like CDAF systems are definitely closing the gap.)

Yes. But I don't think it is even the biggest advantage decent Canon and Nikon DSLRs (and not ALL DSLRs) have in tracking. I heard a lot of info that other brands are not as good. Having low res low noise high frequency input from color array is one hardware help in addition to PDAF array. But I suspect the most lies in the long developed tracking algorithms, and that is where most advantage lies.

But if you develop a sensor with full-area binning, it is going to have better, lower noise signal than the color array. Hundreds of on-sensor PDAF pixels give better, more defined 3d info compared to few PDAF sensors in DSLRs. And algorithm development should be continued, I hope Oly and Pana have enough money to keep mathematicians and programmers working on tracking employed, despite the losses they suffer. But obvious delay in introducing of on-sensor PDAF compared to Nikon, Canon, Sony and Fuji is troubling (of course they need it least of all, and still better without it than anything but Nikon 1, but still).

I think I understand a little better. THanks.

But what is needed is massive pixel binning, turning output of 16 mpix sensor to something like 0.01 mpix first (similar to what color array on focusing sensors in DSLRs have). No camera processor can do object recognition on 24 Mbytes of noise raw data in real time yet even if a sensor could output that (and it cannot or even close, so line skipping is used, leading to low-res noisy picture). You would need a supercomputer for that.

 clengman's gear list:clengman's gear list
Olympus PEN E-PL1 Olympus E-M1 Olympus M.Zuiko Digital ED 14-42mm 1:3.5-5.6 Olympus Zuiko Digital ED 70-300mm 1:4.0-5.6 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 +6 more
Anders W Forum Pro • Posts: 21,468
Re: what is the best m43 for focus tracking?

JamieTux wrote:

Don't forget that PDAF does have one very strong trump card...  It knows whether something is in front of or behind the current focus point.  Otherwise I agree - simplify everything so that the computer has less work to do and it will be faster

I'd say that this trump card has already been overcome. Otherwise, ordinary static CDAF wouldn't be as fast as PDAF. However, PDAF might currently have the edge when it comes to determining in which direction a non-static target is moving and at what speed. Some ideas about how that advantage might be overcome are discussed here:

http://www.dpreview.com/forums/post/50628843

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
clengman
clengman Senior Member • Posts: 1,964
Re: what is the best m43 for focus tracking?

JamieTux wrote:

Don't forget that PDAF does have one very strong trump card...  It knows whether something is in front of or behind the current focus point.

This was the reason for my initial question. I understand that PDAF systems have a definite advantage with regard to C-AF because can calculate at each instant whether the subject is in front of or behind the focus point. That allows the camera to react more quickly to subject motions in the z-axis. Though, apparently the differences in C-AF performance are shrinking.

It doesn't explain to me though why a PDAF system would be better at tracking C-AF where the problem is tracking subject motion in the x and y directions. I still don't understand entirely why a PDAF system would have much of an advantage for that task.

I see a slight advantage in that if the camera sees a blob of light that seems to move as a unit in all three axes, it's likely to be a discrete object. The CDAF system can look at "blob" movement in two axes, but it's more difficult (but not impossible) to "see" the blob move in the z-axis. (A CDAF system could look at the changing size of the blob for instance, or it can record a value for local contrast across the blob along with a value for focus distance and use that information to plot a z-axis trajectory in which the blob has maximum contrast.)

Otherwise I agree - simplify everything so that the computer has less work to do and it will be faster

I guess this is the compelling argument, but certainly, someone at Olympus must know that if they do some sort of binning operation before they attempt the tracking operation that they can produce a faster tracking algorithm. Perhaps they wanted to develop their system to take advantage of a more complex set of information knowing that it would not be as fast or reliable in early iterations and just count on faster processing in the future to eventually bring the speed of the system up to par?

 clengman's gear list:clengman's gear list
Olympus PEN E-PL1 Olympus E-M1 Olympus M.Zuiko Digital ED 14-42mm 1:3.5-5.6 Olympus Zuiko Digital ED 70-300mm 1:4.0-5.6 Olympus M.Zuiko Digital ED 40-150mm 1:4-5.6 +6 more
OP JamieTux Veteran Member • Posts: 4,079
Re: what is the best m43 for focus tracking?

Hi Anders, I was just talking in relation to tracking AF - and I was thinking that if you take 2 readings and the second is further than the first then you know very quickly the direction of motion.  I wasn't talking about it being THE trump card for everything

I think that the biggest factor for trakcing is probably the amount of data that needs to be processed - and not having separate processors for that job as far as I am aware in CDAF sensors - about to reply to Celngman with more thoughts.

Haven't had a chance to test the GH3 yet by the way

 JamieTux's gear list:JamieTux's gear list
Nikon 1 J5 Fujifilm X-E3 Nikon 1 Nikkor 18.5mm f/1.8 Zeiss Touit 50mm F2.8 Nikon 1 Nikkor VR 10-30mm f/3.5-5.6 PD-Zoom +6 more
OP JamieTux Veteran Member • Posts: 4,079
Re: what is the best m43 for focus tracking?

clengman wrote:

JamieTux wrote:

Don't forget that PDAF does have one very strong trump card...  It knows whether something is in front of or behind the current focus point.

This was the reason for my initial question. I understand that PDAF systems have a definite advantage with regard to C-AF because can calculate at each instant whether the subject is in front of or behind the focus point. That allows the camera to react more quickly to subject motions in the z-axis. Though, apparently the differences in C-AF performance are shrinking.

Apparently so - although in my own experience erratic motion (even in a single direction) is better with a PDAF system (I think!)

It doesn't explain to me though why a PDAF system would be better at tracking C-AF where the problem is tracking subject motion in the x and y directions. I still don't understand entirely why a PDAF system would have much of an advantage for that task

I see a slight advantage in that if the camera sees a blob of light that seems to move as a unit in all three axes, it's likely to be a discrete object. The CDAF system can look at "blob" movement in two axes, but it's more difficult (but not impossible) to "see" the blob move in the z-axis. (A CDAF system could look at the changing size of the blob for instance, or it can record a value for local contrast across the blob along with a value for focus distance and use that information to plot a z-axis trajectory in which the blob has maximum contrast.)

I disagree a bit here...  Taking your blob example - as a point gets more out of focus it grows (and gets lower contrast) so that might make it appear to be moving towards you (and only in the z axis)

To split the blob up and look at the contrast on the blob you have to first define the correct blob as the one that has smaller blobs of detail - doing that with the amount of data that the sensor is giving would require huge amounts of analysis - on a PDAF system it's easier because they are dealing with a much smaller data set - they are simply looking at what the colour and brightness is of the blob is and where it's movign to in the x and y axis (this was the game changer when Nikon stole a march with the D3/D300s AF module).

Then you have the fact that for a contrast AF system to know that it is in focus it must have to overshoot (until contrast is reduced) and then jump back - you can see this happening if you put your camera in AF-C and focus on a stationary object.

So I would think the real break through will be when someone works out the algorithm to make the blobs and then runs the AF algorithms on just those.  Standard binning based on zones of the camera wouldn't work I don't think, it would need to be image specific.
To me its like using fourier analysis to turn a piece of music into a set of midi instruments - the advantage we have is that we only use that analysis for a specific job instead of trying to recreate the whole image.

Otherwise I agree - simplify everything so that the computer has less work to do and it will be faster

I guess this is the compelling argument, but certainly, someone at Olympus must know that if they do some sort of binning operation before they attempt the tracking operation that they can produce a faster tracking algorithm. Perhaps they wanted to develop their system to take advantage of a more complex set of information knowing that it would not be as fast or reliable in early iterations and just count on faster processing in the future to eventually bring the speed of the system up to par?

I think I gave my thoughts as to why that is above  Hoping that the information processing will keep up or overtake the information gathering seems like a risky strategy to me - I would think that if they were working that way the latest 12mpix bodies would have tracked better than the first 16mp ones - but maybe the difference would have been insignificant and gone unnoticed.

I still think that the speed improvements will be coming from reducing data intelligently.

 JamieTux's gear list:JamieTux's gear list
Nikon 1 J5 Fujifilm X-E3 Nikon 1 Nikkor 18.5mm f/1.8 Zeiss Touit 50mm F2.8 Nikon 1 Nikkor VR 10-30mm f/3.5-5.6 PD-Zoom +6 more
Anders W Forum Pro • Posts: 21,468
Re: what is the best m43 for focus tracking?

JamieTux wrote:

Hi Anders, I was just talking in relation to tracking AF - and I was thinking that if you take 2 readings and the second is further than the first then you know very quickly the direction of motion.  I wasn't talking about it being THE trump card for everything

We are probably thinking along roughly the same lines. But I find it interesting to consider the details of when and to what extent the ability of PDAF to tell whether the subject is in front of or behind focus and the direction and speed of the movement translates into a clear advantage. Consider to begin with the following examples.

Example 1: Suppose that for a single shot (not a burst), we press the shutter fully, without having half-pressed to prefocus, while aiming at a new moving target. In this case, CDAF is likely to do just as well as PDAF, perhaps better. The circumstances are essentially the same as when focusing on a static target.

Example 2: Suppose we instead try to prefocus on a moving target and then expect the camera to keep that target in focus while we wait for the proper moment to fire. In this case, a known difference between CDAF and PDAF is that CDAF has to adjust focus a little all the time in order to even know whether there is any motion that brings the subject out of focus whereas PDAF doesn't have to do that. On the other hand, PDAF lenses have difficulties moving in very small steps with great precision so it might be more difficult for PDAF to fine-tune focus as the subject moves. Furthermore, the inability of CDAF to determine by means of a single AF reading whether the subject is in focus, behind focus, or in front of focus can be compensated for by "learning from experience". If the subject is regularly moving in one direction, an intelligent CDAF system would quickly discover that fact by trial and error and then try the expected focus direction before the unexpected in its attempt to keep the subject in focus, thus doing better than it otherwise would. So it doesn't seem to me that PDAF has a major advantage in this case either.

There are then, of course, the more complicated scenarios of burst-mode shooting, but I thought we might take the simple ones first.

In the above examples, I am just thinking aloud, and I might have missed something important. So feel free to try to correct me if you think I have missed something important.

I think that the biggest factor for trakcing is probably the amount of data that needs to be processed - and not having separate processors for that job as far as I am aware in CDAF sensors - about to reply to Celngman with more thoughts.

Yes. Data processing capabilities are of considerable importance here. But with those capabilities continuing to rise more or less in agreement with Moore's law, there is hope on the horizon.

Haven't had a chance to test the GH3 yet by the way

But you have it already or what?

 Anders W's gear list:Anders W's gear list
Panasonic Lumix DMC-G1 Olympus OM-D E-M5 Olympus E-M1 Panasonic Lumix G Vario 14-45mm F3.5-5.6 ASPH OIS Panasonic Lumix G Vario 7-14mm F4 ASPH +28 more
Keyboard shortcuts:
FForum MMy threads