Ad
Ad
Ad
Pages: [1] 2 »   Bottom of Page
Print
Author Topic: Reikan FoCal focus micro-adjustement automation software  (Read 16892 times)
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« on: January 10, 2012, 04:11:40 PM »
ReplyReply

http://www.reikan.co.uk/focal/

I just purchased this utility - I think that MA is a great concept, but I just hate tedious repetitive tasks. And to trust something like this, I would like to repeat it at different distances, different focal lengths and perhaps for a few different (large) apertures. Life is too short for doing that manually...

Only working for the Canon 5Dmk2 and 7D for now. I purchased the "pro" version.

So far I have only done an initial test of my 17-55 f/2.8 IS USM EF-S at a fixed distance of approximately 3.5 meters and at 17mm, 30mm and 55mm. Seems that you need lots of light and very stable light conditions to get good results.

So far the application have been quite helpful in informing me why a certain operation cannot be carried out (e.g. that the lense is set to manual focus, that the camera must be in single-shot, centre focus point Av mode ect). It has crashed once, and a couple of times I have had to quit or close more windows than seemed necessary in order to do what I set out to.

By choosing "fully automatic" and "Analysis information", you can track the progress as it goes along (see attachements). I found that interesting and a nice early feedback if something was dubious. So I seem to need [-2 +4 -2] at [17mm 30mm 55mm]. 0 Might not be so bad then...

I have repeated this test several times, and the tendency seems to be there every time.

In every case there was a visible improvement from the "before" to the "after" shots (final attachement).

-h
« Last Edit: January 10, 2012, 04:43:25 PM by hjulenissen » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #1 on: January 10, 2012, 04:46:28 PM »
ReplyReply

I just tested a 100mm f/2.8 Macro USM lense and it seems to be quite erratic. Response is all over the place. I did not try the af range limiter. FoCal properly informed me that the fit was "Not Acceptable" and no correction was loaded into my camera.
Logged
snoleoprd
Sr. Member
****
Offline Offline

Posts: 443



WWW
« Reply #2 on: January 11, 2012, 12:49:39 PM »
ReplyReply

Thanks for posting, have you tried any other lenses?

Alan
Logged

Alan Smallbone
Orange County, CA
BartvanderWolf
Sr. Member
****
Offline Offline

Posts: 3793


« Reply #3 on: January 12, 2012, 07:15:34 AM »
ReplyReply

I just tested a 100mm f/2.8 Macro USM lense and it seems to be quite erratic. Response is all over the place. I did not try the af range limiter. FoCal properly informed me that the fit was "Not Acceptable" and no correction was loaded into my camera.

Hi,

I know that my copy of that lens showed quite a bit of hysteresis, so it matterd whether the AF comes from the minimum focus distance direction of from the infinity position direction. Reversing the direction by 1 smallest focus step could throw focus off significantly.

I'm not sure if that plays a role here, as it seems from the description that focus is approached from -20 to +20 AFMA settings, which should move from closer to farther AF positions, but if focus overshoots the optimal position then such a reversal of direction on the next AF could play a role.

Cheers,
Bart
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #4 on: January 13, 2012, 02:54:39 PM »
ReplyReply

Now doing my 85mm f/1.8 and 50mm f/1.8 II.

LenseFocal LengthSuggested MAFit QualityAge/condition
17-55 f/2.8 EF-S IS USM17-2Excellent2y/Excellent
17-55 f/2.8 EF-S IS USM304Good2y/Excellent
17-55 f/2.8 EF-S IS USM55-2Excellent2y/Excellent
100mm f/2.8 EF USM Macro100-1Not Acceptable??/Good
85 f/1.8 EF USM85-12Excellent??/Good
50mm f/1.8 II EF50-12Good5y/Good
70-200 f/4.0L EF IS USM700ExcellentNew/Excellent
70-200 f/4.0L EF IS USM1200ExcellentNew/Excellent
70-200 f/4.0L EF IS USM2001/2GoodNew/Excellent

Did the 85mm and 50mm now with very different results. My batteries are flat, so that might have affected results (should not, but who knows). Will redo those and also my telezoom when battery is charged.

Enclosing screenshots of new calibrations.

The 85mm seems to be very consistent (large aperture, hits same focus every time, seems to have a significant predictable offset -> perfect candidate for MFA).

The 50mm is less consistent, but I am surprised at how consistent it was (a lot more bell-shaped than the considerably more expensive, although more specialized 100mm macro). When using this lense tethered, I have been shocked by how large the minimum step of the AF motor was, thinking that indicated coarse AF performance.


I was hoping to repeat the same test at large distances. FoCal supposedly can do this thing against any target that is flat and contains sufficient high-frequency contrast. 3.5 meter and 50 meter measurements should do for general purpose photography, right? If results are similar@3.5m and 50m, then I suppose that they are global. If they are very different, then I'll either ignore them, average them, or do more detailed testing.

I am very confident that I would never do this much testing if I had to do it manually (30-50 shots at 1/3 focal lengths for primes/zoom lenses, hopefully 2 distances, repeated at least 2 times), and judging from the progress of the analysis, errors/noise are sufficiently large that I could have ended up doing things worse had I settled for smaller sample sets.

-h
« Last Edit: January 14, 2012, 04:03:41 PM by hjulenissen » Logged
futura
Newbie
*
Offline Offline

Posts: 42


WWW
« Reply #5 on: January 14, 2012, 04:06:00 AM »
ReplyReply

Thanks for the post hjulenissen, I am interested in purchasing a license and its good to find a good writeup. I do have one question though, how accurate have the micro adjustments been in post real-world shooting?
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #6 on: January 14, 2012, 02:05:37 PM »
ReplyReply

...I do have one question though, how accurate have the micro adjustments been in post real-world shooting?
A good question. How can one be certain?

I am doing some rounds with testing, and will double-check against liveview for stationary targets afterwards.

Edit:
Updated the table with 70-200 results.

Edit:
This application is fiddly. I have restarted my camera and unplugged the USB cable quite a few times, seen a hand-full of application crashes, and often see that the application unable to connect to the camera (or able to connect ut not to run liveview/AF) for no apparent reason, only to work again after some time and possibly a restart. Be warned.

-h
« Last Edit: January 14, 2012, 03:52:05 PM by hjulenissen » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #7 on: January 14, 2012, 04:17:07 PM »
ReplyReply

Redoing the 85mm f/1.8 test, I get consistent results.

Have also been using the "semi automated" test on this lense, and observed some peculiar effects. Seems like some "hysteresis" or mechanical hang-up that cause a complex relationship between MFA, sharpness and history.

-h
Logged
MarkL
Sr. Member
****
Offline Offline

Posts: 342


« Reply #8 on: January 15, 2012, 06:53:45 AM »
ReplyReply

Interesting. SLR manufactures should really could implement this into camera, all it would have to do is compare a live view focused image with a normal af image.
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #9 on: January 15, 2012, 03:14:35 PM »
ReplyReply

Yes, I think that this feature could easily be built into live-view equpped DSLRs. Even more useful now that the 1DX is reported to have separate MFA for the wide and narrow end of zooms. The next will be close focused and infinity?

Anyone know what is causing this error? Ideally, a PDAF sensor should return a "peak" when in focus, what would cause it to consistently miss? A miss-alignement of the contrast sensors vs the alignement of the image sensor? Is there reason to believe that the different AF sensors could have different biases?

-h
Logged
futura
Newbie
*
Offline Offline

Posts: 42


WWW
« Reply #10 on: January 18, 2012, 11:22:55 AM »
ReplyReply

A good question. How can one be certain?

A good question. How can one be certain?

I think one the reasons I want to test ma is because I've identified which of my lenses are poor performing and would like to know if a ma update with fix the issue. So I guess the best way to see if FoCal is doing a good job is to pick a lens that you know is not focussing correctly (taking technique and other variables out of the equation).

One thing that held me back from picking up a copy is the recent disclaimer about mirror lockup not working when used with the 1dsmk3 (something about the SDK used by FoCal not working too well with older cameras). In your opinion is this as much of a deal breaker as I think it is?

Thanks!
Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #11 on: January 18, 2012, 12:34:07 PM »
ReplyReply

One thing that held me back from picking up a copy is the recent disclaimer about mirror lockup not working when used with the 1dsmk3 (something about the SDK used by FoCal not working too well with older cameras). In your opinion is this as much of a deal breaker as I think it is?
I don't know anything about the support for 1dsmk3.

I do believe that the "fully automatic" mode was important for me. Given the "noisy" estimates and the time-consuming process, I need to base my conclusions on a bell-shaped curve that I can believe in, otherwise I might succumb to making erroneous conclusions based on what seemed like patterns but was really only coincidence.

-h
Logged
snoleoprd
Sr. Member
****
Offline Offline

Posts: 443



WWW
« Reply #12 on: March 01, 2012, 11:47:23 AM »
ReplyReply

Hi all,

I have been running the software now and I have done most of my autofocus lenses, I still have the 500mm f4L to do as well as the 100-400mm at 400mm when I get some time to setup for a long distance. I used approx 50x the focal length for all the testing and this was done using the Pro version of the software. One thing that is critical to getting good calibration, or at least for me, is to have a well lit target. I ended up setting up with led video lights to illuminate the target, this helped a lot, and also it really needs a place that is out of the wind, etc. running was pretty straightforward and I am overall happy with the results. Here are some of the numbers I got, all lenses are Canon EF lenses unless noted, the quality is the quality of the curve fit.

                                                      1DMarkIV                              5DMarkII
Lens                                                 AFMA    quality                AFMA    quality
50mm f1.4                                         6              ex                    5               ex
100mm f2.8L macro                            2               ex                   2                ex
180mm f3.5L macro                            3               ex                   1               ex
85mm f1.8                                         -8               acc                13               gd
100-400mmf4.5-5.6L @100mm            0               ex                   0               gd
70-200mmf4L  @70mm                       5               ex                  -4               ex
70-200mm f4L @200mm                     4               ex                  -5               ex
17-40mm f4L  @17mm                       -4               ex                 -7               gd
17-40mm f4L @28mm                        -4               ex                 -7               gd
17-40mm f4L @35mm                         0               gd                 -1               gd
17-40mm f4L @40mm                        -3               ex                  3               gd
24-105mm f4L @24mm                      -7               gd                  6               ex
24-105mm f4L @105mm                     3               gd                   5               ex
Tokina 16-28mm f2.8 @16mm             3               ex                   1               ex
Tokina 16-28mm f2.8 @28mm            11              ex                   7               ex
300mm f2.8 L IS                                3               ex                    7               ex
35mm f2                                          -4                pr                    0               pr
50mm f1.8 II                                      7               gd                   11               ex

At least I am getting repeatable results, I have a lensalign that I never got real repeatable results with so I never trusted it, but overall I am very happy with this software solution. I am sure that my problems with the LensAlign was strictly operator error but I was never happy with my results from it.

I am also surprised that how much the MA was off on some of the lenses and it kind of makes sense from what I have seen in the past. It was not surprising that I got poor results with my Canon EF35mm f2, that is a very cheap lens with a similar build to the EF50mm f1.8. My EF 50mm f1.8 was the one that I dropped and exploded into little pieces and Canon rebuilt it and I have hardly used it since but it is essentially a new lens and it worked quite well.

Be happy to answer any questions and I do recommend this software if you want to calibrate your lens and autofocus.

Alan
Logged

Alan Smallbone
Orange County, CA
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #13 on: March 02, 2012, 05:17:59 AM »
ReplyReply

It seems that the 5Dmk3 will have separate MA for near and close focus (as well as small and large focal lengths?).

That could make utilities like FoCal a lot more useful.

-h
Logged
albedo13
Newbie
*
Offline Offline

Posts: 8


« Reply #14 on: April 30, 2012, 11:37:52 AM »
ReplyReply

I have been running the software now and I have done most of my autofocus lenses, I still have the 500mm f4L to do as well as the 100-400mm at 400mm when I get some time to setup for a long distance.

Hi Alan-

Did you ever get results for your 500mm and 100-400mm?  I have both these lenses and have been thinking about getting this software, and would be curious to hear how it worked on these, and what results you found.  Thanks!

Jim
Logged
snoleoprd
Sr. Member
****
Offline Offline

Posts: 443



WWW
« Reply #15 on: April 30, 2012, 09:10:51 PM »
ReplyReply

Hi Alan-

Did you ever get results for your 500mm and 100-400mm?  I have both these lenses and have been thinking about getting this software, and would be curious to hear how it worked on these, and what results you found.  Thanks!

Jim

Jim,

I finished the 100-400mm and at 400mm it was very close to what it was at 100mm, since the camera can only have one MA value, I left it as it had been previous set. I think it was within 2 steps. I have not done the 500mm yet, mostly because of the length needed and every time I get around to setting up outside to do it, it is windy and the target will move slightly. I need to find a better place or try to do it indoors in a building. I have been very pleased with the software. For me it was a lot easier and more repeatable than the Lensalign and similar products. He has made some improvements to the software since I last posted about it, the features are quite nice. Overall I think it has improved my focus accuracy enough to justify the cost quite easily. Also it is pretty easy to setup and run a bunch of lenses in a afternoon, as long as the target is well lit. I end up using some led lights for video to illuminate the target and they work well to give enough light for good accuracy. Remember to use the larger target for the longer focal length lenses. I ended up printing the targets on some Epson cold press paper and mounting the targets to some black gatorboard, works well.

Alan

edit: typos
« Last Edit: April 30, 2012, 09:12:48 PM by snoleoprd » Logged

Alan Smallbone
Orange County, CA
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #16 on: July 07, 2012, 06:31:07 AM »
ReplyReply

I have just tested the new "aperture sharpness test" found in recent versions of the Reikan software (1.4.4 Pro). A nice extra to have some crude, quantitative idea about lense performance vs aperture. Using an aluminium stand, target at 5.5 meters, Canon 85mm f/1.8 and Canon 7D.

Too bad that the y-axis "Quality of focus" is not described anywhere. If could be a linear or non-linear function of any "sharpness" criterion. But I think it is safe to hope that it is monotonically increasing for "better focus". Not sure why the curve seems to flatten between f/3.2 and f/10. The exposure time creeps from 1s (f/22) to 1/8 s (f/6.3), might be that minor focus misalignement along with camera shake make for an upper limit to achieved sharpenss?



« Last Edit: July 07, 2012, 06:39:47 AM by hjulenissen » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #17 on: July 07, 2012, 06:44:01 AM »
ReplyReply

Another new test option is the "consistency test". You choose how to defocus the lense in-between trials (Maximum defocus - far, Maxiumum defocus - near and some other options, a range of AFMA values to sweep across). I found this reassuring to find if there is a offset difference when the focus motor goes one way vs the other.

« Last Edit: July 07, 2012, 06:50:06 AM by hjulenissen » Logged
hjulenissen
Sr. Member
****
Offline Offline

Posts: 1683


« Reply #18 on: July 07, 2012, 07:14:46 AM »
ReplyReply

Yet another option is to run the test at an aperture different from the max aperture of the lense. I assume that this option still use max aperture (f/1.8 in my case) when doing actual PDAF, but that the image is taken at another aperture (for sharpness analysis). I believe that many lenses tend to change the focus slightly when changing aperture, and that Canon are not able to compensate for this in their AF. This undesired behaviour should be more pronounced at moderate to large aperture (where DOF is still small) - when DOF gets large, minor focus errors should tend to be irrelevant (as long as they are absolute distance errors, and do not scale with DOF). If there is a consistent bias, and you are mostly using your f/1.8 lense at f/5.6, you might want to do the test most suitable for your real-life needs.

Summary:
aperture | suggested AFMA
f/1.8:    |  0
f/2.2:    |  2
f/3.2:    | -3
f/5.6:    | -8
f/7.1:    | -6

I have to admit that I am confused by the results. Seems that for every new dimension I add to the testing matrix, a new and totally different MA value appears to be "best". If Canon wants to deliver "pixel-peeping" sharp AF at large apertures in the real world, it would seem that they would either:
1. Compensate for various errors in-camera/lense
2. Offer the customer a larger set of correction values

Option 3 is, of course, to ship the camera + lenses to Canon for adjustement (hoping that it is returned in a better state than when it was shipped). Since my camera is >1 year, the cost of doing that for my 4 lenses is about 50% of the value of the camera itself. It might be more economical to live with what it does today, and swap the camera for a new one (free adjustement of camera + lenses)....

-h
« Last Edit: July 07, 2012, 07:26:58 AM by hjulenissen » Logged
Ellis Vener
Sr. Member
****
Offline Offline

Posts: 1827



WWW
« Reply #19 on: July 07, 2012, 09:09:00 AM »
ReplyReply

It seems that the 5Dmk3 will have separate MA for near and close focus (as well as small and large focal lengths?).

That could make utilities like FoCal a lot more useful.

-h

The Canon 5D mark III and 1D X bodies share the same AF systems (phase based in reflex mode and contrast based in Live view mode) . I've sent the Niko D4 and D800 bodies  I was loaned for reviews back to Nikon but I'm pretty sure they also share AF systems as well. In both cases the RGB inputs in the metering circuits interact with the auto-focus systems. That may (or may not) effect the ability to use the contrast based AF to "tune" the phase based AF system.
Logged

Ellis Vener
http://www.ellisvener.com
Creating photographs for advertising, corporate and industrial clients since 1984.
Pages: [1] 2 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad