Some focus hits, a lot of misses. Is this normal for f/1.4? (with example images)

[dubcat]

I recently calibrated my Sigma 50mm f/1.4 lens with FoCal. I've never had any success with this lens and I am at least getting some hits now. However, I don't know if my hit rate is lower than I should expect so I did this experiment using my Canon 50d and Sigma 50mm f/1.4 lens set to f/1.4 aperture.

I did this in my well lit conservatory where my daughter decided to pose for the camera and sit still. I upped the ISO to 200 just to make sure camera shake was not a factor and shot at 1/640s. I fired away being really careful to aim the active (centre) focus point at her left eye, the one on the right in the pictures. I tried to aim for where the white of the eye and the iris meet. I did not focus and recompose. I had the shutter release button set to do metering, focus and shutter release.

I had a few hit's and a lot of misses. I seem to either get a focus lock on the hair in front of her forehead or the eye itself. I have no idea if this hit rate is normal or not at this aperture. I have posted 100% crops at:

http://s284.beta.photobucket.com/user/aak_linkedphotos/library/Forum/Focus Lock Testing Sigma 50mm

I would appreciate your comments on the focus hit rate and general quality of image. There were no post processing edits carried out other than crop. I converted from RAW in Lightroom 4. Is this to be expected or should I expect better? If there is an issue is it likely to be my lens, the camera body or something else?

 

[tdodd]

I don't know which version of Focal you have, but I have the Pro version and some of the graphs produced are quite eye opening with respect to focus consistency (or lack of it). My Canon 50/1.4 is notoriously imprecise and inconsistent. Of course f/1.4 reveals errors more readily than slower lenses, but for shallow DOF photography I would classify it as unreliable and safety shots would be required. Here's a typical result of running fully automatic calibration. There is no relationship at all between the AFMA value and the sharpness produced by my lens. It is garbage.

For comparison here is the curve for my 85/1.8 when paired with the same body....

Repeated tests have produced similar results over and over for both lenses.

As I understand it the Sigma should be a whole lot better due to superior AF motor engineering, so once calibrated consistency ought to be pretty good. Unfortunately I think that unless you eliminate the human element you will struggle to pin down the performance of the equipment itself. Even if your daughter is completely static - a feat in itself, what about you? Can you be certain that you are not wavering even a couple of mm between focusing and firing?

If you consult DOF tables they won't help at all when pixel peeping. For critical analysis - 100% crops - you'll find the DOF really is wafer thin on a lens like this when used at headshot/portrait distances and it will take almost nothing to yield a poor result. It would be interesting to see a crop with the AF point shown overlaid onto the image, just as you would get from DPP. That way we can see if there is the potential for the AF point to pick up an eyebrow, for example, instead of the eye.

Also, in the case of the Canon, it is a little bit "dreamy" used wide open and stopping down does sharpen the image up nicely. I understand that you probably bought an f/1.4 lens with use at f/1.4 in very much in mind, but would you be happier with the results at maybe f/1.6 or f/1.8?

 

[dubcat]

By the way here are some from the Canon 100mm f/2.8 wide open as a comparison. I asked a friend and he said they look sharper but they don't really 'pop' as he thought they would. I wonder if they only even look more in focus because they are 2 stops more depth of field....

http://s284.beta.photobucket.com/user/aak_linkedphotos/library/Forum/Focus Lock Testing Canon 100mm

Tim - i posted before reading your post. I will take a look at it and respond. I value your input and didn't want you to think i was ignoring you - quite the opposite in fact. I was hoping you would respond.

 

[Willid1]

FoCal have just released the results of collating the data people send them after calibrating their lenses using the FoCal software (feel a bit mean now having always dismissed the "upload data" box)

Some quite interesting results.

http://www.reikan.co.uk/focalweb/index.php/online-tools/lenscamera-information/

Shows which lenses have the most unreliable focus among other things.

And I would echo tdodd's comments re the EF 50 f1.4, I only wish it had been a proper USM motor in the focus mechanism, would have been heaps better.

David

 

[dubcat]

So should I expect better shots than what I got though? This is a real life typical scenario for me - taking pictures of my kids. Would you have expected better consistency and better sharpness on the pictures that are more in focus?

I used focal pro (beta for mac).

 

[tdodd]

I think it would need someone with specific experience of that model of lens to comment, or at least regular experience of shooting at f/1.4, which is not me I'm afraid. I just know that shooting living subjects at f/1.4 is not going to be as easy as doing the same job at f/2.8. It's not just the burden on the gear; It's the burden on the photographer too. Both have to be good.

EDIT : It's also worth commenting on the importance of a bit of sympathetic PP. Here's a 100% crop from my 50/1.4 at f/1.4 on my 1D3, which has pixels which are far more tolerant of error than the 50D. As you will see from the original raw nothing looks sharp, yet there is no problem with focus. At 1/8000 shake is absolutely not the problem, but the image is very soft. Slap a bit of sharpening on it and BAM!

This is what I mean by the "dreamy" look of Canon's lens when used wide open.

 

[4wd]

It can help to use AI Servo mode, your camera might be perfectly still but your model will still be breathing I would think.
Also just because it goes down to F1:4 doesn't mean you have to use that, F:2 will still give you nice bokeh and be a tad more forgiving.