PDA

View Full Version : Hybrid AF on the new E-M1


pdk42
2nd September 2013, 10:45 PM
It seems most reports about the E-M1 say it'll have hybrid AF. I was curious about how hybrid AF works so did a bit of googling (as you do). I found an excellent article on DPReview which went some way to explaining the basics:

http://www.dpreview.com/news/2010/8/5/fujifilmpd

The talk in this article was about putting the AF sensels only into the centre of the sensor but I assume there's no reason why they couldn't be placed in several locations to give a reasonable spread of focus points. I'm still unsure how many sensels are needed for PDAF and whether it'll cause image artefacts. Maybe the JPEG engine can correct if so, but will it be detectable in the raw files? I can only assume that Oly know what they're doing (!).

My bigger concern from what I read whilst googling though is two things:

- Reports of poor hybrid AF performance in other brands (e.g. Canon 650d)

- That lenses designed for CDAF don't necessarily support PDAF (e.g some Sony E-mount lenses such as the 35/1.8)

The second may be the cause of the first of course.

I can see that CDAF and PDAF place different demands on the lens's focus mechanism - PDAF needs to move quickly to a pre-calculated focus point (open loop system), whilst CDAF needs to be able to hunt quickly back and forth whilst reporting its position to the camera as it in turn assesses contrast (closed loop system). I can imagine that the electronics and drive motors could have different designs according to the intended usage.

This leads me to two potential conclusions:

- That the older 4/3 lenses may work very well on the new camera using PDAF but that they will revert to slow mode if the camera switches to CDAF (e.g. low light, face recognition, arbitrary focus point position).

- That the newer u4/3 lenses may not support PDAF at all. This means that we won't be birding or shooting sports with our 35-100s or 75-300s after all. I can see this being a BIG disappointment for many if so!!

Views?

Zuiko
3rd September 2013, 12:20 AM
I'm keeping an open mind, at least until the camera is available. It has been acknowledged by Olympus for a long while that they have been working towards a solution of the problems of using lenses designed for PDAF on MFT bodies. They haven't exactly rushed to bring a product to market and this suggests that they have been more concerned with getting it right, rather than just going for a quick fix. Indeed, it had been hinted that there might have been one more Four Thirds DSLR if they were not able to resolve the issue in time for the E-5 upgrade cycle. The fact that they are now on the verge of launching a camera with the promised solution bodes well.

However, how successful it will be in practice depends upon what level of performance Olympus has deemed to be acceptable and whether continuous AF focus tracking has been implimented to the same standard as S-AF. The benchmark could well be focus performance to match that of the E-5, which to some will still not be good enough, particularly for subjects such as birds in flight. There is also the issue of whether the focusing improvements are restricted to lenses designed for PDAF or will they also apply to current MFT lenses designed to focus by contrast detection?

Matching the existing E-5 focusing performance wouldn't be too much of a disaster, because although it might not satisfy serious birders it would prove more than adequate for the vast majority of photographers. However, there is little point speculating upon this until the camera is available. The proof of the pudding as they say.........

Dick Bowman
3rd September 2013, 06:18 AM
[... deleted ...] The benchmark could well be focus performance to match that of the E-5, which to some will still not be good enough, [... deleted ...]
I've seen this comment a couple of times now, and it bothers me.

Are Olympus subtly managing expectations here?

For me, an absolute requirement would be superior performance to the E-5 - across the board. A "new" body after several years isn't very compelling if the design aim is "to match the abilities of the older one" - unless it does so at a dramatically reduced price.

OM USer
3rd September 2013, 10:41 AM
I'm still unsure how many sensels are needed for PDAF and whether it'll cause image artefacts. Maybe the JPEG engine can correct if so, but will it be detectable in the raw files?

Perhaps this will have its own RAW format with known positions for the sensels (and as you say, the values at these positions will be ignored by the JPEG engine).

...That the newer u4/3 lenses may not support PDAF at all. This means that we won't be birding or shooting sports with our 35-100s or 75-300s after all. I can see this being a BIG disappointment for many if so!!

Yes, that would be a dissapointment. Hopefully though the PDAF sensels will be used to at least start CDAF searching in the correct direction and maybe limit the range that CDAF will hunt before starting to search the other way.

For myself I'm hoping that the larger apertures of the new MFT "pro" lenses that may be coming our way will improve BIF on the E-M5. It would be interesting to find out from our Oly DSLR users the ratio of people using the slower 75-300mm F4.0-5.6, the faster 50-200mm F2.8-3.5, or the SHG lenses (for those with a bigger budgets) for BIF and getting good results. The current MFT offering of 75-300mm F4.8-6.7 is very poor in the specification stakes by comparison and maybe this is where the problem actually lies.

Nick Temple-Fry
3rd September 2013, 11:31 AM
Perhaps this will have its own RAW format with known positions for the sensels (and as you say, the values at these positions will be ignored by the JPEG engine).


I can see the opportunity for Ian to run a competition with some great prizes, it will be a version of "Spot the Ball". A little careful cropping will stop anyone cheating by using the soon to be published sensor locations.

SPOT The PDAF PIXELS and WIN AN E-M1 (key fob)

As to the rest of the thread topic, I'm husbanding my disappointment. I'm sure there will be a lot to get really upset about (if we must) when we've got some decent review testing.

Nick

Ross the fiddler
3rd September 2013, 01:07 PM
For myself I'm hoping that the larger apertures of the new MFT "pro" lenses that may be coming our way will improve BIF on the E-M5. It would be interesting to find out from our Oly DSLR users the ratio of people using the slower 70-300mm F4.0-5.6, the faster 50-200mm F2.8-3.5, or the SHG lenses (for those with a bigger budgets) for BIF and getting good results. The current MFT offering of 75-300mm F4.8-6.7 is very poor in the specification stakes by comparison and maybe this is where the problem actually lies.

The 75-300mm F4.8-6.7 is only 1/2 f stop darker than the 4/3's 70-300 lens & may have some limits, but on the E-M5 with the good ISO performance, it is quite usable with nice results.

Ross the fiddler
3rd September 2013, 01:13 PM
As to the rest of the thread topic, I'm husbanding my disappointment. I'm sure there will be a lot to get really upset about (if we must) when we've got some decent review testing.

Nick

Some are already whinging about the video specs mentioned on the rumour sites. :rolleyes:

brian1208
3rd September 2013, 01:15 PM
Some are already whinging about the video specs mentioned on the rumour sites. :rolleyes:


Its going to shoot video too? What will they come up with next *chr

Olybirder
3rd September 2013, 02:38 PM
Some are already whinging about the video specs mentioned on the rumour sites. :rolleyes:
Yes, according to those comments it is the error of the decade by Olympus and the (yet to be announced) camera is pretty much a non starter. :rolleyes: Some very irrational comments on those sites.

Ron

Zuiko
3rd September 2013, 03:37 PM
I've seen this comment a couple of times now, and it bothers me.

Are Olympus subtly managing expectations here?

For me, an absolute requirement would be superior performance to the E-5 - across the board. A "new" body after several years isn't very compelling if the design aim is "to match the abilities of the older one" - unless it does so at a dramatically reduced price.

Hi Dick, I'm just musing the possibilities and have no information regarding the benchmark that Olympus may have set themselves; it's not an informed comment, merely speculation of one of several scenarios and one which I hope is incorrect. However, I think we must also be aware that this is the first implimentation by Olympus of a solution to a long-standing problem and as such it may need further tuning to reach the optimum potential.

Ulfric M Douglas
3rd September 2013, 06:44 PM
- Reports of poor hybrid AF performance in other brands (e.g. Canon 650d)
Poor on-sensor PDAF performance : Canon 650D, Sony NEX6
Good on-sensor PDAF : nikon1 System.
Probably alright on-sensor PDAF : Canon 70D
Probably good on-sensor PDAF : Olympus E-M1

Cross yer fingers!

pdk42
4th September 2013, 11:25 AM
Poor on-sensor PDAF performance : Canon 650D, Sony NEX6
Good on-sensor PDAF : nikon1 System.
Probably alright on-sensor PDAF : Canon 70D
Probably good on-sensor PDAF : Olympus E-M1

Cross yer fingers!

Let's hope you're right Ulfric. Would be great to have an E -M5 that delivers AF that can track moving subjects.