[joze]

Hallo,



habe das Gefühl, dass mein neues EF24-105 leichten Frontfocus hat. Angehängtes Bild zeigt eine rauhe Wand, fokussiert habe ich auf die Stelle, an der der Nagel in die Wand geht. Man sieht's auch an dem roten Kästchen.



Der scharfe Bereich der Wand liegt leicht vor dieser Stelle. Allerdings ist die Abweichung relativ klein. Andererseits habe ich das Gefühl, dass beim Portrait-Fotografieren die Gesichter dadurch oft ganz leicht unscharf sind.



Meine Frage -- ist das schon Frontfocus, oder bin ich zu pingelig.







Anmerkung:

Das Bild wurde mit einer mittleren Brennweite (ca. 60mm)und Blende natürlich ganz offen aufgenommen. Der Nagel ist real ca. 5 mm hoch. Das raw-Bild liegt auf http://www.zellner.org/EF24-105/IMG_3226.CR2.

[Brightcolours]

[quote name='joze' timestamp='1294791766' post='5432']

Hallo,



habe das Gefühl, dass mein neues EF24-105 leichten Frontfocus hat. Angehängtes Bild zeigt eine rauhe Wand, fokussiert habe ich auf die Stelle, an der der Nagel in die Wand geht. Man sieht's auch an dem roten Kästchen.



Der scharfe Bereich der Wand liegt leicht vor dieser Stelle. Allerdings ist die Abweichung relativ klein. Andererseits habe ich das Gefühl, dass beim Portrait-Fotografieren die Gesichter dadurch oft ganz leicht unscharf sind.



Meine Frage -- ist das schon Frontfocus, oder bin ich zu pingelig.







Anmerkung:

Das Bild wurde mit einer mittleren Brennweite (ca. 60mm)und Blende natürlich ganz offen aufgenommen. Der Nagel ist real ca. 5 mm hoch. Das raw-Bild liegt auf http://www.zellner.org/EF24-105/IMG_3226.CR2.

[/quote]

Not a good test, the AF sensor point is bigger than the square shown, and the camera may well focus on the wall part just in front of the nail.



Of course, yes, your lens/camera combination may well front focus. But the way you want to test it is not a good method at all.

[PuxaVida]

I don't know how it is in Canon's but the squares are also a bit larger on the AF sensor than seen in the viewfinder. So if you check the angle of the wall and the nail perpendicular to it, you'll see that the head of the nail seems to be in focus. And the head of the nail is "almost" on the same focal plane with the in-focus wall part. I agree with BC, you'd better test it in another way.



Serkan

[joze]

OK -- I made a better test with the [url="http://www.peleng8.com/how-to-detect-back-focus.html"]peleng8[/url] pattern:

• The front focus is almost not visible at f=24mm
  
  
• front focus is biggest somewhere @ f=60mm
  
  
• front focus is slightly visible @ f=105mm
  
  

Here's the result for f=60 mm, where the front focus is most significant. The image shows 6 different images. The focus field lies well on the checkerboard pattern which is perpendicular to the optical axis.



The scale appears to be sharp somewhere between 0 and 1, so the focus is wrong by about 0.5 which is about half of the the depth of focus (DOF) (I assume the lens to have no field curvature).



Again my question: is this significant front focus or am I too picky?

[wim]

This is frontfocus, be it only slight, and AFAIAC, within spec.



A few things to take into consideration, however.



1) Canon lenses are calibrated for AF at 50 X FL. At close distances and longer distances the results may therefore vary. From my experience when focusing closer than 50X FL the tendency with well calibrated bodies and lenses, especially zoom lenses, is to get slight frontfocus, be it within spec, and with longer distances slight backfocus, although in the latter case DoF tends to make up for this.



2) AF zoom lenses are not parfocal, which means the camera needs to refocus after zooming every single time, and the tolerances required in the focusing mechanism in order to make AF possible may account for differences in focus across the zoom range for the same subject and subject distance. Often this is not linear either, from my experience it varies up and down across the zoom range, very likely coinciding with the variable movements of the groups and elements with changing distances and zooming.



3) The size of the actual AF sensors is three times the size of the red square in the view finder, linear, or 9 times for area. This means you need to choose a contrast transition which is large enough for the AF sensor selected to get accurate AF results. Furthermore, the AF point may not be exactly in the spot that the red square indicates, it may be offset slightly due to unavoidable tolerances. This is something to experiment with to find out where exactly it is most sensitive.



4) For lenses with maximum apertures smaller than F/2.8 (larger f number IOW), AF accuracy according to Canon specs is within 1 f-stop of aperture selected, and therefore the resutl you seem to be getting is within spec (AF is just within DoF), as the 0, which I assume is yoru focusing point, is still within the DoF zone. Note that with lenses with a maximum aperture of F/2.8 and faster, this is 1/3 of DoF, provided you use one of the fast AF points (with most non professional models, which includes even the 5D II, this is the center AF point). Of course, with the 24-105L this doesn't help you, as it will only work with the AF-points for slower lenses (effectively, a fast AF point will become a slow one with a slow lens, they are generally implemented as dual AF-points).



5) AF depends on lighting. The 1D series cameras (1Ds, and 1D) have much more advanced AF-systems, and generally work well in all kinds of light. All amateur cameras up to and including 500D, 40D, and 5D have AF-ssystems targeted for a single spectrum band of light, and may therefore have problems with AF especially in artifical lighting, such as fluorescent light. The newer cameras from 50D, 5D II onwards use two bands, and are more accurate in this regard. IOW, the 7D and 60D are also more accurate. I am not sure about the 550D, however, although from my experience it seems to be a bit better too



6) AF is about statistics, due to the tolerance of lenses and bodies. When a body-lens combo reaches focus by AF, every time you press the AF button focus will be slightly different, within a certain band, even for static objects and focusing from a tripod with a cable release. However, it should always reach focus conform Canon specs (1/3 of DoF with fast lenses and fast AF points, 1 stop of DoF for all other).



7) Optimal sharpness at pixel level is only required at very large magnifications (like 50% or 100% on screen). For anything else the standard in-DoF sharpness obtained from calibration in spec will suffice for normal shooting, especially if you do some post-processing, either in-camera or on RAW files.



8) AF is a result of the body and lens working together, which is why you really need to send in both when you want a lens calibrated, and ideally you should send in all of your lenses and bodies together for optimal results.



I suggest you try shooting a few normal, real life subjects at different distances, and see how the camera-lens combo fares for AF in those instances. Also test different FLs, not just 60 mm. Really go through the range, 24, 35, 50, 70, 85, 105 mm. You may find it is the same as the results you describe at 60 mm, but it may just as well be better at other FLs. You will liekly find that teh results are ok. There is a big difference between testing these things and actual real life shooting, and i am talking from experience here. You only need to take notice, if the results from real life shooting are not what you exppect. Even so, just like it was in the days we didn't have AF, a number fo shots will be out of focus. Also ote that the best AF system lacks the intelligence of the average person, when it comes to picking the correct contrast transition for focusing, even if, generally speaking modern cameras with AF generally perform way better than we ever did ourselves in the past with manual focus and split wedge or prism collar focusing screens. There is therefore no shortcut in this regard to exactly learn how your camera and lens combo or combos behave under different circumstances in order to get optimal results from them, just as there wasn't in the past with our MF systems.



If you are still unhappy with the AF focus point positioning of your lens-camera combo after the above test, and you don't have micro focus adjust on the body you are using, I suggest you have it calibrated by Canon. If you are not a CPS member, you have to take quite a delay into account, however. Over here in the Netherlands it takes anywhere from 3 weeks to more than 2 months, depending on how busy it is at the repair centre(s). I don't know what the situation in Germany is, however.



HTH, kind regards, Wim

[miro]

HelloWim,

Thanks for valuable input. I didn’t know that canon calibrate lenses at 50xFL.

I have one Tokina 100macro that front focus at longer distances. Up until 3meter is ok. At 5meter front focus a little bit but acceptable. At 20 meters front focus seriously.

I end-up with micro-adjustment at 5meters. Unfortunaely If I try to take group porthret from let say 20meters at 2,8 I have blurred faces.

This combined with severe glare problems makes me to think to sell this lens.

BTW this lens suffers from huge LoCA too, but they can be avoided by postprocessing somehow.



greetings,

Miro

[joze]

[quote name='wim' timestamp='1294903493' post='5464']



4) For lenses with maximum apertures smaller than F/2.8 (larger f number IOW), AF accuracy according to Canon specs is within 1 f-stop of aperture selected, and therefore the resutl you seem to be getting is within spec (AF is just within DoF), as the 0, which I assume is yoru focusing point, is still within the DoF zone. Note that with lenses with a maximum aperture of F/2.8 and faster, this is 1/3 of DoF, provided you use one of the fast AF points (with most non professional models, which includes even the 5D II, this is the center AF point). Of course, with the 24-105L this doesn't help you, as it will only work with the AF-points for slower lenses (effectively, a fast AF point will become a slow one with a slow lens, they are generally implemented as dual AF-points).



[/quote]



wim: thanks for your input, this helps a lot.



Does 4) mean, that a F/2.8 24-70 L Zoom focuses more accurate and/or faster? (I've a 60D body).



Does the focus capability scale further with the speed of the lens, i.e. does a F/1.4 lens autofocus even more accurate and faster that a F2.8 lens?

[wim]

[quote name='joze' timestamp='1294920275' post='5474']

wim: thanks for your input, this helps a lot.



Does 4) mean, that a F/2.8 24-70 L Zoom focuses more accurate and/or faster? (I've a 60D body).



Does the focus capability scale further with the speed of the lens, i.e. does a F/1.4 lens autofocus even more accurate and faster that a F2.8 lens?

[/quote]

In principle, an F/2.8 lens does indeed focus more accurately as a result, provided you use an F/2.8+ AF point, and in principle an F/1.4 lens also, or even more accurately, but that is much of a muchness with regard to accuracy, or rather with precision and repeatability (an out-of calibration fast lens will focus faster, but will still give you the same unsharp image repeatedly with more precision than any other lens which is in calibration <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />). You need more accuracy wide open with an F/1.4 lens than with an F/2.8 lens anyway.



However, at F/1.4 a slight body movement between AF and actual shutter button press generally is of much more influence than the accuracy of the AF system.



BTW, focusing accuracy also varies with exisiting light, as the faster AF points don't work down to the lower light limits where the slower AF points still do, in theory anyway. In practice I have never really noticed a big difference, and I generally do a lot of my shooting in low light.



With regard to faster focusing, that really depends on the implementation of the AF system. Generally speaking, ring USM lenses focus faster than non-ring USM lenses. However, if a ring USM has to move a lot of glass for AF, it will likely become slower again (f.e., 85L and 85L II). However, because of the way a ring USM lens works in relation to its power of moving teh focusing elements and groups in a lens forwards and backward for focusing, it is much more accurate, generally speaking, than a more traditional approach, because for the simpler AF stepper motors in relation to the force and power used the latter generally can't be as precise.



However, take two lenses of the same design and concept, one with ring USM an done with a more traditional AF motor, and the ring USM equipped lens will focus faster and more accurately.



Regarding your 60D: you'd have to use the centre AF point. That is the F/2.8+ AF point.



HTH, kind regards, Wim

[genotypewriter]

[quote name='wim' timestamp='1294903493' post='5464']

1) Canon lenses are calibrated for AF at 50 X FL.[/quote]

Must make a fun visit to their repair center workshop when you send in a 800mm for calibration <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />



GTW

[wim]

[quote name='genotypewriter' timestamp='1295045178' post='5513']

Must make a fun visit to their repair center workshop when you send in a 800mm for calibration <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />



GTW

[/quote]

<img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/tongue.gif' class='bbc_emoticon' alt='' />



The put it on a computerised optical bench, as you well know <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/biggrin.gif' class='bbc_emoticon' alt='' />.



Warm regards, Wim