Focus accuracy on A700 with 50mm f1.4 and 100mm Macro

[Jonathan K]

Hi David,

I remember you stating that the 50mm f1.4 (I still have the 1st generation Minolta) would focus much more accurately on the A700 than on the 7D. I remember you posting a picture of some leaves and feathers on a forest ground.

I am asking because my (two weeks old) A700 seems to backfocus a bit, with 50 f1.4, but above all, with the 100mm Macro (2nd generation Minolta), which has become my favourite lens.
I can easily and consistently reproduce the effect at 2.8. for subjects between 4 and 6 meters with this lens.
With the 50mm, I am not sure whether the lens is faulty (I sent my 7D for calibration because of Backfocus issues, but the 50mm never worked well, even after the calibration. I hardly ever used it, and when I did, I came out pretty frustrated) or whether I am moving after focussing.

So what are your experiences with the 50mm f.1.4, the 100mm Macro and the A700?

Another thing: I already talked to a Sony service and they would do the calibration in 7 days, but the only need one lens ??!!?!? Any lens!!!
With the 7D I had to send in all lenses. Any ideas?

When the AF-assist light is used, the effect is even stronger.

Thanks in advance for your views,

Cheers
Jonathan

[David Kilpatrick]

I used four A700s in the first month of the camera's release. The final camera I bought did not have as accurate a focus with fast, shorter lenses as the test sample. I found this out at the Cirque du Soleil shoot in January, when the 50mm consistently backfocused at around 30 ft distance. Later on, I decided to adjust the AF setting screws very carefully using the 28mm f2 and 50mm f1.4 as a reference. I'm still not entirely sure that it is focusing accurately with all lenses at all distances - for example, closest possible shots with the 70-300mm seem a touch back focused to me compared to what I'm seeing through the finder.

David

[Jonathan K]

That is what I thought, after having read so many threads on DPR about focus inaccuracies. So the question is:
Does it make sense to have it calibrated by Sony or do you believe there is a danger it will be returned worse than it is now?
At this moment, I believe I can live with the inaccuracies... It is a bit of a disappointment because I hoped to use the 50mm much more, and especially since I love to use the Macro lens for anything, don't ask me why, but I just enjoy the results.
I don't believe I will try it myself, at least not as long as there is warranty...

Thanks,

Jonathan

[David Kilpatrick]

What matters most is infinity calibration on the 50mm f1.4. Sony will set it up for perfect results at around 2 metres. First, check and see how accurate it is at 2 metres - if it's good at this distance, don't bother sending it away. If you get a significant error at 2 metres, send the camera and lens in for Sony to calibrate.

David

[Javelin]

I find with my A700 and the 50 1.7 and 100mm as well as some others that when it focuses ona subject it initially nails it but if it has to refocus a small distance it will sometimes not get it quite right.

[Jonathan K]

Javelin, I did not experience that. Usually the first focus and a refocus are exactly the same. But I found, that sometimes, the camera's first "attempt" to focus looks accurate, but then it moves on (it hardly ever focus in one "move") and gives me the green light when the picture looks out of focus again in the viewfinder... So the "correction" the camera makes seems to be a correction for the worse...

But anyway: I did some hopefully proper testing: I put the camera on tripod, SSS off, and I took pictures of a series of books with a front large enough to cover the spot circle. I used f1.6 since at 1.4 the whole picture is foggy and blurry (I suppose my sample is not the very best... I've seen lenses perform better at 1.4)

I got:
At 1 m distance: focus more or less accurate
At 1,50 m distance: 6 cm BF
At 2 m distance: 8-10 cm BF
At 2,50 m distance: 12 cm BF.

(Sorry for measuring in m/cm, but I am sure I will convert incorrectly when I try )

The 100 mm macro does a good job until 2 - 2,5 metres, from there on it is a mess.

So I suppose: off it goes to a Sony service.... I think I will send along anything fast: 50 f1.4, the 100 f2.8 and the 28-75 D f2.8.

Thanks for your advice...

Cheers Jonathan

[David Kilpatrick]

Just to throw a spanner in the works, most 50mm f1.4 lenses suffer from focus shift as you stop down. You are focusing at f1.4, but in that slightly soft mix there may be a dominant tendency to focus in front. This means that on the focusing screen you will see a certain apparent result based on the transmissive aperture of the focus screen itself - around f4.5. If you had a true dead matt focus screen, the image would look out of focus, and it also does to the sensor. The AF module also has an effective aperture beyond which it can not 'see' any of the image forming light. The centre line sensor on the A700 is an f2.8 sensor, and can 'see' the lens image forming rays up to f2.8 - but that is only 1/4 of the image forming light. The other sensors can only 'see' f5.6.

What this means in practice is that your 50mm f1.4, autofocused, may appear to be sharply focused at f5.6 (or perhaps f2.8 if the centre sensor is used) but if you actually use f1.4, you are using anything up to 16X the level of image forming light which is not necessarily focusing in the same plane.

This may sound complicated but anyone who is familiar with the tendency of fast lenses to shift their apparent focus point when stopped down will understand how this reacts with the optics of both the focus screen and the AF sensors.

The only answer is a lens with a very high order of corrections and no focus shift with aperture change.

David

[Jonathan K]

Hi David,

well it took me some time to digest that one...
I have about 1001 questions for you right now, so let me stick to the practical things:

Do I get this right that there is an discrepancy between the view through the viewfinder (which is „altered“ by the optical qualities of the focus screen) and the actual image the sensor is „seeing“? Could that be the reason why the DOF previewed through the viewfinder sometimes gives totally different results than what I see on the final image?

Sorry to sound like a schoolboy here, but how exactly does a acute matte focus screen work? I remember learning that „modern“ focus screens are actual Fresnel lenses on one side... So what is a „true dead matt focus screen“ and why can you see the correct focus there?
Ans in how far does the focus screen influence the AF-system? As far as I understood, the focus screen has no function in the AF-system??

Practically speaking: Can I expect the A700 (with its central dual cross-line sensor) to perform accurately in combination with a lens which has f2.8 as a max aperture? Would sound logical to me...

Last question (just to make sure): Is that „focus shift“ a matter of bad quality, or is it inherent to the construction of a fast lens?

You don't know any link to a good an detailed online explanation to all of this?

Thanks a lot in advance and sorry for bothering you with millions of questions,

Cheers Jonathan

[David Kilpatrick]

I don't have time to explain in full, but for the focusing screen explanation, see our other website dPhotoexpert:

http://www.dphotoexpert.com/2007/09/21/ ... iewfinder/

Focus shift is inherent in fast lenses but can be designed out. Modern focusing screens tend to compensate for stop-down focus shift, as they only use an effective f2.8 to f5.6 of the lens aperture, and these are commonly used f-stops. AF systems also use a restricted pupil, and therefore can not focus an f1.4 lens accurately. The special f2.8 sensor in the A700 can focus an f2.8 lens accurately, wide open. The other sensors can not; they 'see' the lens as if it was stopped down to f5.6. The f2.8 sensor will not work any faster, or in any lower light, with an f1.4 lens rather than an f2.8. Nor will you see any gain in viewfinder brightness (though many users swear that they can).

This subject is a 'can of worms' for the industry, and a technical writer's minefield as the optical calculations needed to understand it are not readily available. But - DSLRs are not WSYWYG through the prism viewfinder, and AF systems do not actually function best with faster lenses.

That's one reason why Sony issues a lens like the 70-300mm G SSM in an f4.5-5.6 maximum aperture. It's actually better matched to the range of Alpha bodies than a faster one would be.

David

[picman]

What matters most is infinity calibration on the 50mm f1.4. Sony will set it up for perfect results at around 2 metres. First, check and see how accurate it is at 2 metres - if it's good at this distance, don't bother sending it away. If you get a significant error at 2 metres, send the camera and lens in for Sony to calibrate.
David

There I was happily shooting away not aware that anything might be wrong (an occasional flower or bee was not focused properly and I attributed that to user-error, now I wonder), but seeing all these reports about back- and frontfocusing I'm going to test my A700 and CZ24-70.

I found this on the internet http://www.focustestchart.com/chart.html

Is there another better way of testing? I guess I will have to test various mm e.g 24mm, 40mm and 70mm? Any pointers David?

Thanks, cheers, Bob.

[Javelin]

IMHO Sony should call on us here and over at Dyxum to get samples of every single Minolta lens brought into their service centres and gather enough info to make an adjustmet table they can use in the firmware of the cameras. woudn't they be miles ahead that way ? I see a new menu item called lens correction on/off. They probably have all the info they need to do this to begin with just from Minolta's archives.

I don't have time to explain in full, but for the focusing screen explanation, see our other website dPhotoexpert:

http://www.dphotoexpert.com/2007/09/21/ ... iewfinder/

Focus shift is inherent in fast lenses but can be designed out. Modern focusing screens tend to compensate for stop-down focus shift, as they only use an effective f2.8 to f5.6 of the lens aperture, and these are commonly used f-stops. AF systems also use a restricted pupil, and therefore can not focus an f1.4 lens accurately. The special f2.8 sensor in the A700 can focus an f2.8 lens accurately, wide open. The other sensors can not; they 'see' the lens as if it was stopped down to f5.6. The f2.8 sensor will not work any faster, or in any lower light, with an f1.4 lens rather than an f2.8. Nor will you see any gain in viewfinder brightness (though many users swear that they can).

This subject is a 'can of worms' for the industry, and a technical writer's minefield as the optical calculations needed to understand it are not readily available. But - DSLRs are not WSYWYG through the prism viewfinder, and AF systems do not actually function best with faster lenses.

That's one reason why Sony issues a lens like the 70-300mm G SSM in an f4.5-5.6 maximum aperture. It's actually better matched to the range of Alpha bodies than a faster one would be.

David

[David Kilpatrick]

The firmware already contains an adjustment table. It only covers one lens, the 50mm f1.7, and it only corrects for an aperture of f7.1 and a working distance of 1.93 metres. One reason the fast lenses like the 50mm f1.7 may misfocus at full aperture is because the adjustment table is set to assume f7.1 as the working aperture, and any focus shift occurring will be compensated for at that aperture - but wide open will not be perfect, as the focus shift will be assumed to be present when it's not.

Agreed - they need a far more complex adjustment table, covering the aperture set, the focal length and the focus distance. Maybe the Alpha 900 will have it.

David

[Wes Gibbon]

Presumably this is an industry wide problem - but how do other manufacturers compare? Are these problems more effectively addressed by semi-pro/pro cameras that cost £1000 or £2000+?

[Jonathan K]

I believe the Canon/Nikon high-end cameras have some option to adjust every single lens....
Don't ask me more, I was told by somebody who was told by somebody

Jonathan

[Glyn R]

The dphoto expert comments on live view only apply cameras where the Live view image is taken directly from the main sensor. The Sony system is only really a view of the viewfinder image so focus errors will creep in. This was apparent on my E-330 which had both systems, manual focusing using live view was only accurate with the main sensor viewer.