Focus problems with SAL 50mm f1.4 (on A350)

[Chris Malcolm]

I thought at first this lens was naturally rather soft at apertures wider than f2.8m but I've discovered that was mostly inaccurate focus. In my testing so far I get the impression that the central focus point is more accurate than the selectable edge ones. It also seems to be the case that AF on the central focus point is not quite accurate enough for the very shallow DoF this lens has at wide apertures and distances of around twelve to fifteen feet. If I keep focussing and shooting on exactly the same thing at f2 or wider some of the images turn out impressively sharp and some are clearly out of focus, usually by focussing a bit further out.

These problems seem worse when the light is dim and there isn't a sharp high contrast edge to focus on. In poor conditions when it sometimes fails to focus or takes a bit of tracking back and forth to find a focus it seems at least sometimes to be willing to settle on a poorer focus.

I've tried putting the A350 on continuous focus and tracking the focus point slowly backwards and forwards in distance to see how much latitude it has in what it thinks is a sharp focus. If change of focus is always indicated by a bit of audible activity from the focus motor then it does seem to be the case that the range of what the camera thinks is acceptably sharp focus at least overlaps and possibly exceeds the DoF of this lens at that kind of range and wide apertures.

I've had a shot at manual focussing, but it's hard to do better than the AF with a small optical viewfinder image, a live view that won't magnify in RAW (why on earth not?), and a focus ring with a very small travel so that the tiniest of nudges has -- for this kind of critical focussing -- a large effect.

It's clear from my lucky accidents of very good focus that at wide apertures this is a remarkably good lens, but the problem is how to focus it when neither AF nor manual focussing provision seem to be quite up to the precision required.

Is this a general problem with properly working A350s and this lens? If so, any tips about how to coax a good sharp focus from the system at wide apertures?

[David Kilpatrick]

Chris, first of all the live view is not related to raw or jpeg, it is just a video feed sub-VGA resolution like a webcam, and can not be magnified because there is no information there to magnify.

Secondly, the 50mm f/1.4 has a focus shift related to aperture, and the A350 lacks an f/2.8 centre AF sensor - all the sensors at approx f/6.3. This means you are really focusing at f/6.3, with all the depth of field that implies, then expecting to shoot at f/2 or f/1.4 with much less depth of field - the AF simply is not accurate enough, the focus shift adds to this, and the colour and level of light affects it too.

The A700 and A900 have f/2.8 centre sensors and work much better with the 50mm f/1.4, but what is really needed is a redesign like the Nikon AF-S G 50mm f/1.4 with nearly zero aperture related focus shift. Answer - bracket your focus, manually. It's just like the old argument between short base and long base rangefinders, in theory the Hi-Matic 7S II was not capable of focusing its 40mm f/1.7 lens accurate enough to use full aperture and the Minolta CLE was not accurate enough to use a 50mm f/1.4 Leica lens - etc.

David

[TomB]

I have similar experience with Minolta 50 mm F1.7 on A350. My impression was it has something to do with camera vibrations too as test shots on a tripod at 1 m were almost perfect but I had back focus on hand-held images. I get 10 cm back focus at 2-3 m when shooting with the center AF sensor and about 1 cm using the left AF sensor. I'm slightly disappointed and I consider canceling my order of Sigma 30 mm F1.4. I'm not sure if it will be useful on my camera.

Maybe there could be a similar solution for adjusting the AF system of A350 like here for A100: http://www.flickr.com/groups/sony_alpha ... 1269119583

Tom

[David Kilpatrick]

I have not used the Sigma 30mm f1.4 on the A350, or any Sony. I have used it on Canon 300D and Sigma SD9, and on the basis of that experience, I would not buy it. It has very poor illumination at f1.4 (dark corners) and does not focus accurately enough to be useful wide open. The Sigma 50mm f1.4 is likely to be much better, but more expensive and huge. The focus error has nothing to do with vibrations, focus should be OK at around 1.9m with the 1.7 but if it is not, the manual correction using the AF adjustment screws is possible.

My A350 has been very accurate with most lenses but not with the 16-80mm CZ at 16mm.

David

[Chris Malcolm]

I have carried out some focussing experiments. I find that at distances between one and two metres (all I've tested) and in good daylight that I get very consistent repeatable focussing. At f5.6 and wider the sharpest DoF area lies a few cm past the the nominal focus point, e.g. if focussing on a frontal nose the nose would be slightly out of focus and the ears close to the sharpest. At f8 and smaller apertures the point of nominal focus comes within the front edge of the expanding range of DoF.

I also find that if I switch to manual focus and use AF "green square" focus confirmation, that as these close distances the span of focus ring adjustment which is confirmed as being in correct focus is wide enough to span at least most of the DoF.

If these results extend to further distances, and to other lenses, then it should be possible to adjust this out in the camera.

[David Kilpatrick]

Using the MF confirm light is useful - if your lens tends to back focus, try setting MF at the closest point the light comes on.

David

[TomB]

I checked properly focusing of my A350 with both lenses. My first tests were not exact enough, so I thought of vibrations influence first (I didn’t express it clearly). I adjusted the AF module by instructions from flicker.com, finally I turned all three screws 20 ° clockwise in three steps.

Now the focus is better with all lenses (incl. set lens) even though I didn’t experience any focusing problems prior to purchasing the 50 mm, as it was pretty good focused in DoF. The focus is not perfect at F1.7 as you can see on the right but it shifts with higher aperture values and this adjustment fits better the 16-105 on various focal lengths. Now I'm happy with the 50 mm for portraits.

BTW I discovered, that the left AF sensor is slightly shifted forward in comparison with others but with the new adjustment it is not as noticeable as it better fits in DoF. I don't care for it.

[David Kilpatrick]

The left and right screws adjust that edge/corner sensors in that direction, front and back screws adjust the other way - the one in the corner of the L shape is the 'main' adjustment and the other two are 'yaw' and 'pitch'. You can with care fix the whole sensor array.

David

[Brad Smith]

David I'm not sure what you are saying with regard to the 3 screws. I'm concerned I may have adjusted my A300 incorrectly. I had a back focus issue but the problem was constant across the sensor plane. I adjusted all 3 allen bolts the same amount. Was this the correct thing to do or should I have only adjusted that one in the bottom left?

Cheers - Brad

[David Kilpatrick]

That was the correct thing do to. If you only adjust one, you skew the AF sensor plane.

David

[Hakanie]

Using the MF confirm light is useful - if your lens tends to back focus, try setting MF at the closest point the light comes on.

David

I own a KM5d, does it have the MF confirm light?

Hakanie

[Chris Malcolm]

Since my posting on the topic of back focus with the SAL 50mm f1/4 lens, I've acquired the 500mm f8 reflex lens. It too back focuses, and has such a shallow DoF that focusing is extremely critical. Using manual focus there is no perceptible range of movement where the correct focus indicator comes on. The slightest touch in either direction drops it out of focus. Since it has fixed aperture there are no focus shift problems, so I decided to use the adjustment screws around the tripod bolt to correct the focus on this very critical lens, and see where that left focus with the 50mm.

I found it took roughly 20 degrees of clockwise rotation on all three screws to get focus spot on at close distances (about 12 feet) with the 500mm f8. At distances between 3 feet and 12 feet with the 50mm at f1.4 this has removed the back focus and placed the point of focus within the DoF close to the front boundary, which seems a good place for it to be for portraits.

It'll take a while to check out how these lenses now perform at longer distances, especially since the 500mm requires good bright light and unusually clear and calm atmospheric conditions to resolve high detail at distances of a few miles. So this is an interim report of encouraging progress in correcting back focus.

Incidentally while carrying out these focus tests on a tripod (Benbo) I noticed that with the 500mm lens mirror or shutter vibration (or both) caused easily noticeable detail smearing at a shutter speeds of 1/30th sec. Very annoying considering that there's no mirror lock possible on the A350. My tripod may also not be stable enough for this very critical lens, since the sharpest results came from the highest shutter speeds.

[David Kilpatrick]

Using the MF confirm light is useful - if your lens tends to back focus, try setting MF at the closest point the light comes on.

David

I own a KM5d, does it have the MF confirm light?

Hakanie

Yes, the dot in the finder between the arrows is the focus confirm light.

David

[Chris Malcolm]

Since my posting on the topic of back focus with the SAL 50mm f1/4 lens, I've acquired the 500mm f8 reflex lens. It too back focuses, and has such a shallow DoF that focusing is extremely critical. Using manual focus there is no perceptible range of movement where the correct focus indicator comes on. The slightest touch in either direction drops it out of focus. Since it has fixed aperture there are no focus shift problems, so I decided to use the adjustment screws around the tripod bolt to correct the focus on this very critical lens, and see where that left focus with the 50mm.

I found it took roughly 20 degrees of clockwise rotation on all three screws to get focus spot on at close distances (about 12 feet) with the 500mm f8. At distances between 3 feet and 12 feet with the 50mm at f1.4 this has removed the back focus and placed the point of focus within the DoF close to the front boundary, which seems a good place for it to be for portraits.

It'll take a while to check out how these lenses now perform at longer distances, especially since the 500mm requires good bright light and unusually clear and calm atmospheric conditions to resolve high detail at distances of a few miles. So this is an interim report of encouraging progress in correcting back focus.

I had to abandon the idea of using shots of distant targets (such as a clock tower) a few miles away over the city rooftops. Convection currents from sun shining on rooftops, and from sundry chimneys, heating and cooling systems, etc. made good distortion free visibility over that kind of terrain and distance much too rare for testing purposes. Using more precise closer targets I found some slight readjustments were needed to the 500mm focus. The result of this more precise focus adjustment on my 50mm f1.4 lens is that at apertures of f2 and smaller the AF always puts the point of focus inside the DoF, but at wider apertures it drifts forwards out of the DoF. At f1.4 the point of focus is at least the width of the DoF ahead of the region of sharp DoF, and the image is clearly badly focussed.

Since the lens is much sharper at f2 than f1.4, and much sharper at f2.8 than f2, and bringing the f1.4 focus point inside the DoF would throw out the f2 focus, the current adjustment seems the most rational compromise for the aperture related focus drift of this lens on my A350's AF.

An unexpected improvement may also have occurred in the focusing of my 18-250mm zoom. I thought this slight adjustment, so important for very critical shallow DoF lenses and apertures, would be insignificant for the zoom, since its focus point was inside the much larger DoF anyway. But it looks as though the rather soft images at 250mm, which I had taken to be a feature of this lens, was in part due to backfocus, because at 250mm it seems to have sharpened up a bit.

[Brad Smith]

Hi Chris, I've found that 1/30th is just about the worst shutter speed there is for mirror vibration effects. Sony's IS doesn't help in this situation either sadly