Wednesday, July 2, 2014

Persistence Pays Off


Once in a while, when you are ready to resign things to being just the way they are and give up, the planets align, the clouds part, and the gods smile down, making everything right again.  Well, perhaps I’m exaggerating a bit.  However, my initial testing of the new firmware Version 1.01 for the Canon mount Sigma 120-300 f2.8 “Sports” lens is complete and things are looking good.

To briefly recap, my issue with the lens and original Version 1.00 firmware was basically two-fold.  First, despite all sixteen adjustable focus points available in the Sigma Optimization Pro software being perfectly calibrated using the USB dock, on the 1D Mark III and Mark IV the lens would badly front-focus in between the 200mm and 300mm settings.  Second, the lens required completely different Optimization Pro focus parameter settings for the 1D Mark III and 1D Mark IV compared to Canon’s other bodies like the 7D, 5D Mark II and 50D.  This meant my lens was either “set up” to work with my 1D bodies, or my other bodies, but never ALL my camera bodies at once.  I began my testing with a very methodical approach.  I needed to find out first if the new firmware fixed the issue of the different parameter requirements for the different bodies.  I also wanted to get a baseline to see how different my new Optimization Pro setting would be from my old ones, and I wanted to do it for all my different camera models. 

One of the programs I discovered in my search for the “perfect” focus calibration tool (which btw, doesn’t exist on its own in my opinion) was Reikan FoCal Pro.  This software works by tethering the camera to the computer with a USB cable, and pointing the lens at Reikan’s special focus calibration target.  The software takes control of the camera, firing off a series of “in-focus” frames of the target while adjusting the camera’s internal auto-focus manual adjustment (AFMA) setting for each frame.  The software compares the “sharpness” of the target image from one frame to another, finding the AFMA value that produces the image with the best focus (ingenious).  It works very well, but can be time consuming as each “test” can take five minutes or longer depending on the accuracy and consistency of the camera/lens combination.  Considering the sixteen focus settings for the 120-300 in the Optimization Pro software, and the five different camera bodies I wanted to test the lens on, this adds up to a lot of time. 

Since it is not physically possible to print a Reikan target large enough to test a telephoto lens at infinity (or even reasonably close to it), I stuck with testing for the three closer distance settings.  These are 10 meters, plus the lens’ minimum focus distance which I labeled as “Min”, plus the distance in between the two, which is marked only by a line on the lens distance scale (no numbers) which I labeled as “Mid”.  I ran a full battery of tests at the three distances, with the lens set to 120mm, 150mm, 200mm, 235mm and 300mm, using the 1D Mark IV, 1D mark III, 5D Mark II, 7D and 50D (3 x 5 x 5 = 75).  At around 5 minutes each, that’s over 6 hours of just watching the software and camera do its thing (in case you were wondering why it has taken me so long to get this post up).

Below is a screenshot of the Excel document I created containing all the results.

At each focal length and distance I had Excel average the determined AFMA setting for the cameras together, thus generating a full set of values representing the average AFMA settings for all five models (note that on my 7D at the minimum focus distance for 300mm, 235mm, 200mm and 150mm, the software found the required AFMA value to be more than +20, which is the camera’s limit, so I had no choice but to just ignore those values).  Armed with these numbers, I proceeded to plug them into the corresponding values in the Optimization Pro software (screenshot below).  My experience with the older firmware was that the adjustment parameters for the infinity settings were close, if not identical, to the parameters for 10 meters, so I just copied them over.  Also, there is no adjustment setting for 235mm, I added that value because in between 200mm and 300mm was where the issue was on the 1D cameras, and 235mm is physically right in between the motion of the zoom ring for 200mm and 300mm.  I wanted to confirm that the new firmware was addressing this.

Being labeled with the “S” designation, the new Sigma 120-300 f2.8 is the first lens in the “Sports” category of Sigma’s “Global Vision” series of lenses.  This is a true sports lens, and that is what I use it for 90% of the time.  Unfortunately the end of June is not the best time of year to do practical testing on a sports lens, since not a lot of different sports are being played.  Baseball is not ideal because there is not a lot of focus tracking being done from far to near distances like football or soccer.  Luckily, there is an Ultimate Disk league that plays regularly at a park near my home, and is just what I needed to test the lens after plugging in the new values. 

Below are samples of frames from each camera, all taken at f2.8 and various distances and focal lengths.  The focus for all the frames is spot-on.

1D Mark IV 300mm
1D Mark III 220mm
1D Mark III 300mm
7D 300mm
7D 161mm
50D 300mm
50D 269mm
The frame below is some proof that the new Version 1.01 firmware has fixed my first issue with the lens front-focusing in between 200mm and 300mm on the 1D cameras.  This frame is shot on a 1D Mark IV with the lens at 252mm, right in between, and is tack sharp.  This would not have happened easily with the old firmware.
1D Mark IV 252mm
Also of particular note is the frame below, shot on a 1D Mark IV at 120mm, but here the full early evening sun is blaring just out of the left side of the frame.  Another improvement listed with the new firmware is faster focus speed and accuracy in the lens’ “Speed-priority” focus mode (which I use almost exclusively).  I shot a heck of a lot with the 120-300 f2.8 Sports and the Version 1.00 firmware, and don’t think this image would have been in focus with the old firmware.  Of course I have no direct comparison of the same image made with the old firmware, but my gut is telling me from previous experience that it is so.
1D Mark IV 120mm
Since I purchased the Sigma 120-300 f2.8 Sports over a year ago I have felt very confused and conflicted over it.  It is in a class by itself for focal range and maximum aperture, and has outstanding image quality, making it the most ideally suited lens I have ever owned for shooting sports, both indoors and out.  For the longest time I felt like I was the only one experiencing, and making noise, about the issues the lens had on the 1D Mark III and Mark IV, despite my results being consistent and repeatable on several different lenses and cameras.  In my previous “My Trip to Sigma” post I posed the rhetorical question of whether the long trip to Ronkonkoma Long Island was worth it (BTW, I have actually made two trips out there in total).  I concluded then that it basically would depend on whether Sigma was listening to my problem, and willing to do something about it.  I think I can say with some real confidence now that the answer to that question is a resounding “YES!” 

My test images above prove that a single set of focus parameters will now work for all my camera bodies, and that on my 1D bodies I can get consistent accurate focus in between 200mm and 300mm.  I will likely end up “tweaking” the focus parameter settings a bit as I go, but I think these are pretty close (I will start splitting hairs from this point on).  I did not shoot tests using my Canon 1.4 x III converter because I wanted to concentrate on the 1D issue specifically.  Subsequent testing so far has shown no issues or problems using the converter as well.


In my haste to get this post finished I neglected some important information.  I made no mention of the camera’s built-in AFMA adjustment option.  In his comment below Fnurck asked “I wonder how its possible using the average adjustment and still all image are in focus with the different cameras without change the settings in between”. 

Doing all the testing and averaging with the Reikan software allowed me to come up with a set of values that compensate for the front or back focus “bias” the lens has for each of the parameters in the Optimization Pro software.  As important as the numbers are themselves, more important is their relation to one another.  For example, suppose you add +5 focus units to each of the software settings, then dial a -5 setting into the camera’s AFMA setting.  In theory you should end up with the equivalent overall adjustment. 

After programming the averaged values into the lens with Optimization Pro, I then used each camera body’s internal AFMA adjustment to “calibrate” the body to the lens (for this I used a LensAlign MkII with the optional long ruler).  So in short, the first round of hours and hours of testing and averaging with the Reikan software is to “set up” the lens to compensate for its own front/back focus biases at each focal length and distance value.  The second round (which I neglected to include originally) is to “set up” each camera with the pre-programmed lens to compensate for the combination of the two together.   That is how that single set of averaged values can work well on all the different cameras. 

I also wanted to comment on the difference between the Optimization Pro settings I found necessary between the Version 1.00 and 1.01 firmwares.  If you look at the screen shots from my first “Initial Rant” post, you will see that for the 1D Mark III and 1D Mark IV the Optimization Pro settings make some pretty wild swings from deep on the “plus” side to deep in the “minus“ side in the adjacent values going from 200mm to 300mm.  The settings I found for the 5D Mark II, 7D and 50D cameras even required +20 settings in some fields and -18 settings in others (though for these cameras the progression from “plus” to “minus“ was spread over several sets of fields).  In general, the required settings I found with the Version 1.01 firmware had a little less overall “spread”, and also follow a relatively smooth progression from “plus” to “minus“ without any wild swings in adjacent fields.

For the in-camera AFMA values, the additional good news is that with all my different camera bodies, none require a setting greater than +/- 5 focus units with the pre-programmed lens.  This tells me Sigma did a good job in not only addressing the issue specific to the 1D cameras, they also did a good job of “tightening up” programming.  Bravo!
So unless I come across any unusual findings, or get reports of such from other users, I think this just may be the final post on this blog. I would like to offer my sincere thanks to the folks at Sigma U.S.A and Sigma Japan for listening, and releasing the Version 1.01 firmware update.  I will continue to be a big fan of Sigma lenses, and am anxious to see what they have coming down the pipeline. 

As a final note, I will be writing a post in the future on my methods for doing focus calibration on Sigma lenses with the USB Dock and Optimization Pro software, and other brand lenses.  That post will be on another upcoming site which I will provide a link for here.  Thanks to everyone who has read this blog, commented and contributed!