Hi All,
I just look over the review of 70D at DPR related to its focus, in sensor vs. phase detect. The difference looks quite exaggerated to me. I had misfocus issues with phase detect, but never that consistent. What do you think?
http://www.dpreview.com/previews/canon-eos-70d/7.
I have my misgivings too, but Imaging Resource expects on-chip PD AF to be very accurate. The basis for their reasoning http://www.imaging-resource.com/PRODS/ca...n-70dA.HTM is outlined as follows:
"In any phase-detect AF system, focus accuracy is a strong function of the "baseline" of the AF sensor. The greater the angle of the incoming light rays and the longer the distance between AF sensor elements, the more precise the focus will be. This is why many higher-end cameras have
special "high accuracy" central points, that kick in at apertures of f/2.8 or larger. Taking advantage of the wider light cone at very wide apertures, they use a second set of AF pixels spread further apart from each other to achieve greater accuracy.
With the Canon 70D's image sensor, the effective AF baseline can be made as wide as the lens will
allow, since there's no restriction caused by a fixed AF-sensor geometry. As the aperture gets wider (and you really need the added focus precision due to shallower depth of field), the camera can simply
select a more widely-separated group of pixels to use for its focus measurement. There's no limit to this, either. An f/2 lens will give higher AF accuracy than an f/2.8 one, and an f/1.2 lens will be even
more so. Combine this with the guaranteed alignment between the "focus sensor" and the image sensor, and we should see unprecedented levels of focus accuracy with very wide-aperture lenses."
But, as they say, the proof is in the pudding.
Quote:I have my misgivings too, but Imaging Resource expects on-chip PD AF to be very accurate. The basis for their reasoning http://www.imaging-resource.com/PRODS/ca...n-70dA.HTM is outlined as follows:
"In any phase-detect AF system, focus accuracy is a strong function of the "baseline" of the AF sensor. The greater the angle of the incoming light rays and the longer the distance between AF sensor elements, the more precise the focus will be. This is why many higher-end cameras have
special "high accuracy" central points, that kick in at apertures of f/2.8 or larger. Taking advantage of the wider light cone at very wide apertures, they use a second set of AF pixels spread further apart from each other to achieve greater accuracy.
With the Canon 70D's image sensor, the effective AF baseline can be made as wide as the lens will
allow, since there's no restriction caused by a fixed AF-sensor geometry. As the aperture gets wider (and you really need the added focus precision due to shallower depth of field), the camera can simply
select a more widely-separated group of pixels to use for its focus measurement. There's no limit to this, either. An f/2 lens will give higher AF accuracy than an f/2.8 one, and an f/1.2 lens will be even
more so. Combine this with the guaranteed alignment between the "focus sensor" and the image sensor, and we should see unprecedented levels of focus accuracy with very wide-aperture lenses."
But, as they say, the proof is in the pudding.
I'll wait and see how this technoligy translates in real life.
I will skip spuerlatives as - "we should see unprecedented levels of focus accuracy"
I think that theoretical/ marketing people forget something - What about RSA?
Quote:I'll wait and see how this technoligy translates in real life.
I will skip spuerlatives as - "we should see unprecedented levels of focus accuracy"
I think that theoretical/ marketing people forget something - What about RSA?
I just got 70D for myself. Yes, PD AF accuracy is indeed VERY high.
However, third party lenses are NOT supported.
Is there any system that can get over RSA in lenses
Quote:I just got 70D for myself. Yes, PD AF accuracy is indeed VERY high.
However, third party lenses are NOT supported.
<blockquote class="ipsBlockquote" data-author="thw" data-cid="24283" data-time="1377363697">
I just got 70D for myself. Yes, PD AF accuracy is indeed VERY high.
However, third party lenses are NOT supported.
Is there any system that can get over RSA in lenses
</blockquote>
Thanks for sharing.
Do you had any Canon xxD
before? If yes how it compares with them. I have 50d and I can say that AF
speed and accuracy is as good as it get. During my 4-5 year experience I cannot
say that I have seen any AF accuracy issue with my 50d. That is why I'm a
little bit confused how it is possible to make something significantly better
if it was already very good.
The only problem has been RSA with some fast glass, but as already mentioned, there
is no way to escape from it.
Thanks,
Miro
Quote:Is there any system that can get over RSA in lenses
Quote:I have my misgivings too, but Imaging Resource expects on-chip PD AF to be very accurate. The basis for their reasoning http://www.imaging-resource.com/PRODS/ca...n-70dA.HTM is outlined as follows:
"In any phase-detect AF system, focus accuracy is a strong function of the "baseline" of the AF sensor. The greater the angle of the incoming light rays and the longer the distance between AF sensor elements, the more precise the focus will be. This is why many higher-end cameras have
special "high accuracy" central points, that kick in at apertures of f/2.8 or larger. Taking advantage of the wider light cone at very wide apertures, they use a second set of AF pixels spread further apart from each other to achieve greater accuracy.
With the Canon 70D's image sensor, the effective AF baseline can be made as wide as the lens will
allow, since there's no restriction caused by a fixed AF-sensor geometry. As the aperture gets wider (and you really need the added focus precision due to shallower depth of field), the camera can simply
select a more widely-separated group of pixels to use for its focus measurement. There's no limit to this, either. An f/2 lens will give higher AF accuracy than an f/2.8 one, and an f/1.2 lens will be even
more so. Combine this with the guaranteed alignment between the "focus sensor" and the image sensor, and we should see unprecedented levels of focus accuracy with very wide-aperture lenses."
But, as they say, the proof is in the pudding.
Well,
the base distance in PD is determened by prism angle. Yes the different prism angle in PD requires different AF sensor but this has never been an issue for middle and pro range SLR. I still puzeling what is the advantafe of such system.
The only one that I can see is lowering the cost and reduce machanical requirements.
Quote:Well,
the base distance in PD is determened by prism angle. Yes the diferent prism angle in PD requires diferent AF sensor by this has never been an issue for middle and pro range SLR. I still puseling what is advantafe of such system.
The only one that I can see is lowering the cost and reduce machanical requirements.
In order for the camera to take a picture or make a movie, the mirror must be up (unless it's the Sony SLT system, with the permanent mirror). During that time, "normal" PD AF does not work. So, it does not allow for continuous AF in movies, and even when taking pictures, if you shoot series, as long as the mirror is up, AF cannot track. So even for photography, when continuous tracking is desired, such a new system can provide benefits.
In theory at least, the new system now has many many more AF sensors compared to the traditional PD AF, which is limited to selected points determined by where the AF sensors are located.
Quote:In order for the camera to take a picture or make a movie, the mirror must be up (unless it's the Sony SLT system, with the permanent mirror). During that time, "normal" PD AF does not work. So, it does not allow for continuous AF in movies, and even when taking pictures, if you shoot series, as long as the mirror is up, AF cannot track. So even for photography, when continuous tracking is desired, such a new system can provide benefits.
In theory at least, the new system now has many many more AF sensors compared to the traditional PD AF, which is limited to selected points determined by where the AF sensors are located.
Thnaks,
Tracking moving object could be beter at least in theory. AF cover area could be advantage for someone.
Honestly Im tired to read canon advertizing about this revolutionary??? system. They sounds like the Nikon 1 advertisement one year ago.
Back to the really techical part. I have questions
1. what is the base distance in canon 70d hibrid PD?
2. Is it constant across focus
Quote:Thnaks,
Tracking moving object could be beter at least in theory. AF cover area could be advantage for someone.
Honestly Im tired to read canon advertizing about this revolutionary??? system. They sounds like the Nikon 1 advertisement one year ago.
Back to the really techical part. I have questions
1. what is the base distance in canon 70d hibrid PD?
2. Is it constant across focus
Yes, indeed, the new sensor is technically something really new (not just a shrunk dSLR, like the Nikon 1). For daily use
of a dSLR, it perhaps makes not such a big difference nowadays, since there are plenty of good performing cameras out there.
But technically, yes, it is very novel, and has the potential for further innovations. If its PD AF gets as good as the viewfinder PD AF, then the days of the mirror box are counted, provided good EVF come along. Further, the dual pixel technology has also the potential to increase dynamic range dramatically, if different ISOs can be applied to the two pixels during imaging (like the Magic Lantern dynamic range expansion recently introduced for some Canon cameras).
According to Canon, the area of the sensor that can be used for AF is about 80% horizontal and 80% vertical.
http://www.canon.com/news/2013/jul02e.html
http://www.learn.usa.canon.com/app/pdfs/...uality.pdf
http://www.canon.com/camera-museum/tech/report/2013/08/
So, you can calculate the distances, i.e. 22.5 x 15mm. 80% is 18mm x 12mm. if a point at the edge of this 80% is to be calculated to be in focus, the maximal spread of the out of focus signal could be 3mm (vertical), the point at the edge at 80% is 1.5mm away from the edge, so the out of focus phase signal could be spread +/- 1.5mm. In the center of course, bigger spreads are possible.
This sensor area will of course not change, irrespective of lens or focus distance. What is only relevant is the angle of the incident light, which can change of course depending on focal length, aperture, and focus distance (with simple old lens designs, the lens is just moved further away from the sensor, so the angles will become more narrow).
Canon says, apertures up to f11 can work with the new PD.
I guess one can try to calculate how much out of focus something can be to work, if one uses the 1.5mm out of focus blur above as starting point. Say, a 50mm lens, at 1:10 (0.1x) magnification, a point at ~50cm distance could be maximally blurred to 15mm (radius), if the lens is out of focus. At the bottom here, http://en.wikipedia.org/wiki/Circle_of_confusion it shows how one can calculate the blur radius c, so one could run a couple of different examples through to get some idea how well it works. Anyway, as pointed out, in the center, larger blur radius would work for lenses such as f1.2 or f1.4. One notes, that in most cameras, for such large aperture lenses, also conventional PD AF is usually limited to central areas.
Thanks for the input,
I agree with your and Canon explanation about advantages of their new system. While such
system offers
1.huge AF cover area - nice feature for video
2.solved/reduced split prism blackout problem – it allows to work with A=11. Nice feature for video
However such advantages should comes at certain cost. In this case I suspect the they reduced the maximum “baseline distance” used in their PD. If this is true – the new system offers the potential of bad AF accuracy esp with fast glass :-)
E.g Canon 10D,20d,30d…..60d offer 2,8 max baseline only for AF central point.
Pro camera
has 2,8 AF area only around image center and they have huge AF sensor
There is no camera that offers baseline distance higher that 2,8. That is why all SLR
suffer somehow with focus accuracy for 1,4 and faster lenses. E.g. Canon and
Nikon specifies very good the AF accuracy with such fast glass. E.g. Canon 1D III+
50/1,2 L
There are some physical lows – You cannot make sharp AF point at image corner with big baseline
/2,8/.
I still puzzling the new canon split focus design - what is the output of such convex desing insted of rounded prism
Theory, theory….. The better way to prove it is to search for real life experience. E.g wait for Dpreview
With Kind Regards,
Miro
|
