Home · Register · Search · View Winners · Software · Hosting · Software · Join Upload & Sell

Moderated by: Fred Miranda
Username   Password

  New fredmiranda.com Mobile Site
  New Feature: SMS Notification alert
  New Feature: Buy & Sell Watchlist
  

FM Forums | Canon Forum | Join Upload & Sell

1       2       3              15      
16
       17              23       24       end
  

Dot-Tune: Autofocus Fine Tuning in under 5 minutes
  
 
jmvdigital
Offline
• •
Upload & Sell: On
p.16 #1 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Snapsy, I don't doubt the repeatability without refocusing or recomposing on the target. The issue that brought me to discover the horizontal/vertical thing was that depending on how large your checkboard target and/or where exactly you put the focus point, the tune value can change significantly between tune attempts if you try recomposing a bit.

I don't think it's all that accurate to just point the camera at the target, and tune once or twice and call it good. If you move the center focus point around to different parts of the target, and if you have the camera refocus (even with CDF), the tune values change almost every time. I'm getting swings of up +/- 5 point in some cases.

And I think part of that variability has to do with the significant difference in how the AF system is responding to vertical or horizontal info. While I agree that you wouldn't WANT to tune using only horizontal or vertical bars, don't you think it's important to have that in mind when choosing your target? This goes not only for Dot Tune I would imagine, but for any MFA adjustment procedure.

As an aside, I have started testing using two targets with more "organic" and random shapes and contours on it, rather than perfectly straight lines or grids. I'll report back.



Apr 21, 2013 at 09:44 PM
snapsy
Offline
• • • •
Upload & Sell: On
p.16 #2 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


jv, I might have misunderstood a response from you a page ago but I thought you indicated a large AF tune differential with different targets using a locked focus position. If you're seeing a tune delta when changing both the target and focus then I'd suggest changing each individually to isolate the source of your delta. I don't doubt that different targets produce different tune values; I observed this myself while developing DotTune, which is why I suggested Atkin's target or something similar. Unless the goal is to experiment with how the AF system reacts to inferior targets you'll run yourself ragged trying to get an accurate tune.


Apr 21, 2013 at 10:44 PM
jmvdigital
Offline
• •
Upload & Sell: On
p.16 #3 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


My point is though, that if a person just sets up their target, focuses, and does a single round or two of tuning, the resulting tune value may not be all that accurate.

I can leave the camera position untouched, simply run the CDF to focus the lens again, and the tune value changes a couple of points. Or I can move the center AF point to a different part of the target and the tune value changes a couple of points. I also found that if your light levels change while doing the tune process, the end points of the tune range change. I had cloud cover come over and reduce my light level a couple of stops, and the tune range shrank several points. While the Dot Tune method does remove a bunch of the mechanical variability, you still have to account for the mechanical variability of the initial focus, and how the AF system responds to changing stimuli, even if you aren't moving the servos.

What's to say the single position on the target you chose, with the single focus setting, is the most accurate tune setting?

Personally, given my experience with this process over the last week, the only way to go is to run the tune process multiple times, refocusing, moving the view to slightly different part of the target, etc. and get the approximate range of tune values. Then take that set of a dozen tune values or whatever, and average it. For instance, using the same checkerboard target, consistent light, no movement of the tripod... only moving the camera's view of the target slightly and refocusing, I got a range of values of +2 to +9.5.

If I setup a target, and no matter where I focused on that target (within reason) it gave me a +/- 1 tune value every time, my argument would be moot, and a single tuning session would be all that's needed.



Apr 22, 2013 at 02:46 AM
gschlact
Offline
• • •
Upload & Sell: On
p.16 #4 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Please clarify. You mention the instances where you get different range values keeping everything the same but AF on different part of your test chat. However, you don't indicate whether the center point of the different ranges for this situation remains fairly constant on the MFA scale?

jmvdigital wrote:
My point is though, that if a person just sets up their target, focuses, and does a single round or two of tuning, the resulting tune value may not be all that accurate.

I can leave the camera position untouched, simply run the CDF to focus the lens again, and the tune value changes a couple of points. Or I can move the center AF point to a different part of the target and the tune value changes a couple of points. I also found that if your light levels change while doing the tune process, the end points of
...Show more



Apr 22, 2013 at 02:57 AM
jmvdigital
Offline
• •
Upload & Sell: On
p.16 #5 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


gschlact wrote:
Please clarify. You mention the instances where you get different range values keeping everything the same but AF on different part of your test chat. However, you don't indicate whether the center point of the different ranges for this situation remains fairly constant on the MFA scale?


Do you mean was I only using the center focus point? Yes, only the center point. I meant that I would loosen my ball head and move the camera ever so slightly so that the exact pattern underneath the AF sensor would change.



Apr 22, 2013 at 02:59 AM
snapsy
Offline
• • • •
Upload & Sell: On
p.16 #6 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


jmvdigital wrote:
My point is though, that if a person just sets up their target, focuses, and does a single round or two of tuning, the resulting tune value may not be all that accurate.

I can leave the camera position untouched, simply run the CDF to focus the lens again, and the tune value changes a couple of points. Or I can move the center AF point to a different part of the target and the tune value changes a couple of points. I also found that if your light levels change while doing the tune process, the end points of
...Show more

I hear what you're saying, which again is why I advocate in the technique the use of a target (and lighting) that is known to produce reliable focus in most situations. The fact that a different target or lighting can produce different AF tune values isn't significant in my mind unless those differences are unique to a given shooting scenario and whose incorporation into tuning is essential for getting acceptable AF results in those situations.



Apr 22, 2013 at 03:12 AM
Lars Johnsson
Offline
• • • • • • •
Upload & Sell: Off
p.16 #7 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


snapsy wrote:
I don't doubt that different targets produce different tune values; I observed this myself while developing DotTune, which is why I suggested Atkin's target or something similar. Unless the goal is to experiment with how the AF system reacts to inferior targets you'll run yourself ragged trying to get an accurate tune.


+1



Apr 22, 2013 at 04:57 AM
jmvdigital
Offline
• •
Upload & Sell: On
p.16 #8 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


I feel like you guys are missing my point. Take the same target, same lighting, same distance. Set your camera up like you were doing it for the first time (focus on target, camera moves slightly in relation to exact spot on the target), but do it three or four times... I guarantee you'll get different tune values almost every time.

You will only get consistent tunes values (-4, -4, -4) if you don't touch the camera at all. The second you move it a tiny bit or refocus, you now have a new tune value. How do you know if the single tune value you achieved on your only focus attempt at the dead center of the target is best value when it changes a little bit every time you refocus or point the camera at a different view of the SAME target.

I still say that a range of tune values should be taken and averaged.



Apr 22, 2013 at 02:44 PM
snapsy
Offline
• • • •
Upload & Sell: On
p.16 #9 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


jmvdigital wrote:
I feel like you guys are missing my point. Take the same target, same lighting, same distance. Set your camera up like you were doing it for the first time (focus on target, camera moves slightly in relation to exact spot on the target), but do it three or four times... I guarantee you'll get different tune values almost every time.

You will only get consistent tunes values (-4, -4, -4) if you don't touch the camera at all. The second you move it a tiny bit or refocus, you now have a new tune value. How do you know
...Show more

I don't see this run-to-run DotTune variation on my 5DM3 or D800, either originally and in a re-test I ran this morning to confirm, refocusing each run.



Apr 22, 2013 at 04:26 PM
Ralph Wagner
Offline
• • •
Upload & Sell: Off
p.16 #10 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Worked like a charm on all my lenses and the 7D, they are all spot on now. Thanks for a great contribution to the photo community.


Apr 23, 2013 at 09:16 PM
 

Search in Used Dept. 



badlydrawnboy
Offline
• • •
Upload & Sell: On
p.16 #11 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


I wanted to share my experience and ask a couple of questions.

I used DotTune successfully in the past to calibrate my Canon 24-70 and 50L. I just got a Sigma 85/1.4. I decided to compare Focal's new TurboCal feature with DotTune. TurboCal returned a value of -3, and DotTune gave me 1.5. I tried both in practice, and -3 was spot on.

Has anyone with FoCal compared the accuracy of TurboCal with the DotTune method?

With my Canon lenses, it was easy to tell when the AF confirmation stopped. With this Sigma, it was less clear. For example, sometimes on the first half-press, there would be a short delay. But then it would lock and stay solid green for 5 seconds, and continue to focus and lock very quickly with subsequent half-presses. Since I did the TurboCal procedure first, I went through the DotTune process a couple of times to make sure I wasn't misinterpreting the results. Each time I did it, I ended up with the same +1 or +2 result.

Any idea why there would be a difference, and why the TurboCal would be correct? My next step will be to compare the TurboCal values with DotTune for my 24-70 and 50L.



May 03, 2013 at 11:00 PM
mitesh
Offline
• • • •
Upload & Sell: On
p.16 #12 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Yes, I also have tried this with my 24-70, 70-200 and 600. I got some very different results, and from inspecting the photos, it looks like the TurboCal was more correct. I am not saying that DotTune is not accurate- it was more than likely user error. I also let Focal run its automatic calibration, and those results were very close or identical to TurboCal.

I'm going to retry DotTune again to see if I can isolate my errors.



May 03, 2013 at 11:07 PM
badlydrawnboy
Offline
• • •
Upload & Sell: On
p.16 #13 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


mitesh wrote:
Yes, I also have tried this with my 24-70, 70-200 and 600. I got some very different results, and from inspecting the photos, it looks like the TurboCal was more correct. I am not saying that DotTune is not accurate- it was more than likely user error. I also let Focal run its automatic calibration, and those results were very close or identical to TurboCal.

I'm going to retry DotTune again to see if I can isolate my errors.


Probably user error in my case too, but strange that I had no problem with my Canon lenses. I'll report back after testing the 50L and 24-70 with TurboCal.

In the past I had inconsistent results with the full calibration methods of FoCal, but TurboCal seemed to nail it right away.



May 04, 2013 at 01:13 AM
badlydrawnboy
Offline
• • •
Upload & Sell: On
p.16 #14 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Okay, I did some more testing. Here's a list of how DotTune compared to TurboCal:

Sigma 85/1.4
DotTune: +2
TurboCal: -3
Real-world testing: TurboCal wins

Canon 50L
DotTune: -3
TurboCal: +4
Real-world testing: TurboCal wins

Canon 24-70 II L
DotTune: +1 @24mm and +2 @70mm
TurboCal: +2 @24mm and 0 @70mm
Real-world testing: too close to tell a difference on the wide end, but TurboCal seems a bit better on the long end.

I'm guessing this is definitely user error, because I don't always feel I can discern exactly when the autofocus confirmation stops with DotTune when the results are inconsistent. But the difference between DotTune andTurboCal on the 50L is really significant, and I don't think I could have misjudged it that much. Any ideas on what's happening here?



May 06, 2013 at 12:51 PM
jmvdigital
Offline
• •
Upload & Sell: On
p.16 #15 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


badlydrawnboy wrote:
Okay, I did some more testing. Here's a list of how DotTune compared to TurboCal:

Sigma 85/1.4
DotTune: +2
TurboCal: -3
Real-world testing: TurboCal wins

Canon 50L
DotTune: -3
TurboCal: +4
Real-world testing: TurboCal wins

Canon 24-70 II L
DotTune: +1 @24mm and +2 @70mm
TurboCal: +2 @24mm and 0 @70mm
Real-world testing: too close to tell a difference on the wide end, but TurboCal seems a bit better on the long end.

I'm guessing this is definitely user error, because I don't always feel I can discern exactly when the autofocus confirmation stops with DotTune when the results are inconsistent. But the difference
...Show more

Are you testing TurboCal and DotTune on the same Reikan targets in the same lighting?

In my experience, DotTune behaved differently for me depending on the target. I believe Reikan uses dense checkerboard-like target. FWIW, I gave up the checkboard (or any horizontal/vertical artificial "target") in favor of a black/white subject with more a more random pattern and organic shapes. I would also run DotTune (or TurboCal for that matter) more than once, readjusting your camera slightly at the target and refocusing between runs. I know snapsy disagrees, but I have always gotten run-to-run differences with DotTune. Even just forcing the camera to reacquire focus in LiveView can sometimes tweak the DotTune value several points, simply because of the physical focusing process margin for error. I ended up running 5-12 runs of DotTune per lens and averaging the result. This seemed to give me very accurate and repeatable results so far.

Here is a low-res of the target I now use for DotTune, in case anyone is interested. I usually work my way around the target, running DotTune with the center AF point over various parts of the design. (See my previous posts about the 5D3 and how it responds differently depending on whether the AF sensor is picking up vertical vs. horizontal info).






May 06, 2013 at 02:43 PM
badlydrawnboy
Offline
• • •
Upload & Sell: On
p.16 #16 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


jmvdigital wrote:
Are you testing TurboCal and DotTune on the same Reikan targets in the same lighting?

In my experience, DotTune behaved differently for me depending on the target. I believe Reikan uses dense checkerboard-like target. FWIW, I gave up the checkboard (or any horizontal/vertical artificial "target") in favor of a black/white subject with more a more random pattern and organic shapes. I would also run DotTune (or TurboCal for that matter) more than once, readjusting your camera slightly at the target and refocusing between runs. I know snapsy disagrees, but I have always gotten run-to-run differences with DotTune. Even just forcing
...Show more

That's helpful, thanks. I'm using the FoCal target for both TurboCal and DotTune. Perhaps I'll try again with a target more like the one you suggest (is that a copyright-protected image, or can I get a high-res copy somewhere?)

I know I would often get different results with FoCal in the past when I ran it multiple times, which has always been one of my frustrations with it. TurboCal seems more consistent, though.

I'll do some more testing and report back.



May 06, 2013 at 02:57 PM
caMARYnon
Offline

Upload & Sell: Off
p.16 #17 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Thank you very much snapsy, I used your method to all my lenses and now the autofocus is much more accurate.


May 29, 2013 at 06:39 AM
wardy
Offline

Upload & Sell: Off
p.16 #18 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


Tag


May 29, 2013 at 05:52 PM
Richard Nye
Offline
• • • •
Upload & Sell: On
p.16 #19 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


TAG


May 29, 2013 at 07:11 PM
outlawyer
Offline
• • •
Upload & Sell: On
p.16 #20 · Dot-Tune: Autofocus Fine Tuning in under 5 minutes


tag


Jun 03, 2013 at 07:20 PM
1       2       3              15      
16
       17              23       24       end




FM Forums | Canon Forum | Join Upload & Sell

1       2       3              15      
16
       17              23       24       end
    
 

You are not logged in. Login or Register

Username   Password    Retrive password