Home · Register · Join Upload & Sell

Moderated by: Fred Miranda
Username  

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

FM Forums | Canon Forum | Join Upload & Sell

1
       2       end
  

Archive 2014 · AFMA Settings, Comparison of Bodies

  
 
Gunzorro
Offline
• • • • • •
Upload & Sell: On
p.1 #1 · p.1 #1 · AFMA Settings, Comparison of Bodies


I thought I'd share some freshly mined data after just updating my MFA records, listing all the MFA settings for three different bodies (only hold 20 entries each). There were a few interesting points to be learned from such a broad comparison.

By and large, the input data was the same between the three bodies. The only difference was a lack of listing for 100-400 + 1.4X on the 5D2, giving it 22 separate lens listings instead of 23 shared with the 1Ds3 and 1D3. The lens inventory encompasses a range of production from 1988 (35-105/3.5-4.5 and 135/2.8 SF) to present day (24/2.8 IS and 35/2 IS).

Each body has a distinct range or inclination of correction, reflected in the Average Setting (zero for each camera).

The "best" or tightest Standard Deviation was on my 1D3 which had its shutter and complete mirror box/AF sensor module replaced by Canon shortly after I purchased it used a couple years ago. The low SD is also mirrored by the lowest Extreme Spread -- only 6 adjustment points between all the lenses compared.

1Ds3 -- 23 lenses, Average Setting: +1.608, Standard Deviation: 2.759, Extreme Spread: 10 (-2 to +8)*
1D3 -- 23 lenses, Average Setting: +2.814, Standard Deviation: 1.677, ES: 6 (0 to 6)
5D2 -- 22 lenses, Average Setting: -2.591, Standard Deviation: 2.839, ES: 7 (-7 to 0)

These numbers show that each camera body has a distinct "personality" regarding AF, which is highly variable. It makes me wonder what other differences each model might have based on differences in it's processor/sensor. Certainly the 21MP images of the 1Ds3 are different from the 5D2. . .

Your comments, data, and examples are welcome!

-----------------
* one lens posted at +8, skewing the ES of the 1Ds3. Eliminating this one lens would produce an ES of 7, in line with the other two bodies.

Edited on Aug 28, 2014 at 12:09 AM · View previous versions



Aug 26, 2014 at 10:03 AM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #2 · p.1 #2 · AFMA Settings, Comparison of Bodies


...and I thought I was a number nerd.


Aug 26, 2014 at 10:27 AM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #3 · p.1 #3 · AFMA Settings, Comparison of Bodies


Here's what I get for my current AFMA cameras.







Aug 26, 2014 at 10:39 AM
Gunzorro
Offline
• • • • • •
Upload & Sell: On
p.1 #4 · p.1 #4 · AFMA Settings, Comparison of Bodies




Yes, you are a big numbers nerd. I'm one too (slightly slighter, though)

Thanks for sharing, Jim. This might be boring to some readers, but the more we compare these sorts of numbers, the better we get at unraveling the mystique associated with why some cameras seem better or worse on functioning and IQ -- the basis of our "likes".

I'm sure there are others out there who have a library of MFA and other sorts of settings.

Small sampling between the two of us, but offhand it looks like the new 1DX is a highly precise machine. That is one sample, so I'd love to get other members' confirmation. The 6D seems less so, looking at those two sets of numbers.



Aug 26, 2014 at 01:11 PM
kaycephoto
Offline
• • •
Upload & Sell: Off
p.1 #5 · p.1 #5 · AFMA Settings, Comparison of Bodies


Definitely good information to know.. I suspect the final conclusions will somewhat resemble the AF consistency findings of Roger Cicala's LensRentals testings though (1DX & 5D3 in the lead, 1D Mark IV next, followed by everything else), which also happens to correspond with my personal experiences.

Unfortunately I haven't kept records for various lenses on my 1-series bodies (while I had each of them)..



Aug 26, 2014 at 01:24 PM
boingyman
Offline
• • •
Upload & Sell: Off
p.1 #6 · p.1 #6 · AFMA Settings, Comparison of Bodies


IMO it unforunately varies from body to body (regardless of model) and lens to lens. Get a 100 different 5DIII and 85L's roate them to 100 different photographers to do MA of their choice and I wouldn't be surprised if you see ton of inconsistencies just from body to body variations, lens to lens variations and user error. There's also no gold standard to do MA so these numbers really don't apply to anyone except to you. With that being said final conclusions will be anecdotal at best.


Aug 26, 2014 at 03:53 PM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #7 · p.1 #7 · AFMA Settings, Comparison of Bodies


boingyman wrote:
... With that being said final conclusions will be anecdotal at best.


Not if many different copies of many different lenses and cameras are tested using the same methodology by the same people.

Roger... ?



Aug 26, 2014 at 03:58 PM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #8 · p.1 #8 · AFMA Settings, Comparison of Bodies


boingyman wrote:
... With that being said final conclusions will be anecdotal at best.


Also, if most AFMA testers are competent and use relevant methods, then I believe that statistically significant trends will emerge, as the number of test results grows.



Aug 26, 2014 at 04:04 PM
RCicala
Offline
• • • •
Upload & Sell: Off
p.1 #9 · p.1 #9 · AFMA Settings, Comparison of Bodies


jcolwell wrote:
Also, if most AFMA testers are competent and use relevant methods, then I believe that statistically significant trends will emerge, as the number of test results grows.


It certainly can be done, but I think it's a pretty huge task. We want to compare different camera types. But we have a big variation within each camera type. For example, let's just take a dozen copies of Camera A. There are at least 6 calibrations (mirror angle, submirror angle, AF sensor angle ((3 of these actually)), electronic for each AF sensor spot on the AF sensor, electronic for distance) within each copy, and if they're calibrated, then they vary. Then we've got lenses we use, and those have calibration points (AF motor speed, lens distance sensor, electronic recognition of infinity focus point).

The reason I only tested center point focus is that lateral focusing points are affected by lens variation. Few of us want to know this, but on optical bench testing about 25% of prime lenses and nearly half of zooms have a detectible tilt. It's not severe enough to affect our photographs, but it probably is severe enough to affect off-axis focus (since phase detection is comparing the light path from two sides of the lens).

All that being said, enough data points can overcome multiple variables. We gave up before doing off-axis testing, though, because we figured we needed at least 500 data points from different copies of cameras and lenses for each camera-lens-lighting configuration (so 500 off-axis data points for 5DIII with 24-70 f/2.8 at 24mm under tungsten light of XXX lumens, for example).

I'd suggest limiting to prime lenses (less variance) and a few bodies. Then if maybe a dozen people each did a series of 50 shots under reasonably controlled conditions, we'd have a reasonably accurate database. But I doubt it's going to happen. In the meantime I trust comments made by numerous people who have two or three cameras and notice a difference. I think those comments are pretty consistent.

For me, personally, the more I understand phase detection AF, the less I complain about how well it works and the more amazed I am that it works at all. It may be the most amazing engineering triumph in optics.



Aug 26, 2014 at 06:01 PM
philshoots
Offline
• • •
Upload & Sell: Off
p.1 #10 · p.1 #10 · AFMA Settings, Comparison of Bodies


Roger, I was unaware that today the sensor itself had three adjustments, thanks for that info. When you mention the 25% tilt can any of this group be the result of bent/ non plane parallel flanges? Or is this attributed to internal groups? What about body flanges, in the old days of film one could shim the flange to insure FFD was spot on and the flange true. Are DSLR flanges still shim dependant of has MFA done away with that element? I ask because I'm aware of the torque a big tele can put on the flange mounts. Many thanks for your time.
Phil



Aug 26, 2014 at 06:26 PM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #11 · p.1 #11 · AFMA Settings, Comparison of Bodies


jcolwell wrote:
Also, if most AFMA testers are competent and use relevant methods, then I believe that statistically significant trends will emerge, as the number of test results grows.

RCicala wrote:
...The reason I only tested center point focus is that lateral focusing points are affected by lens variation...


All of my AFMA tests are centre point only, because that's what I use most often, regardless of which camera I'm using.




Aug 26, 2014 at 06:38 PM
RCicala
Offline
• • • •
Upload & Sell: Off
p.1 #12 · p.1 #12 · AFMA Settings, Comparison of Bodies


philshoots wrote:
Roger, I was unaware that today the sensor itself had three adjustments, thanks for that info. When you mention the 25% tilt can any of this group be the result of bent/ non plane parallel flanges? Or is this attributed to internal groups? What about body flanges, in the old days of film one could shim the flange to insure FFD was spot on and the flange true. Are DSLR flanges still shim dependant of has MFA done away with that element? I ask because I'm aware of the torque a big tele can put on the flange mounts. Many
...Show more

Phil, all of those things can cause tilt - my most accurate measurements of tilt are on the optical bench, so body flange tilt and sensor-to-flange tilt don't count in my numbers, but lens mounting flange tilt would.

It's an interesting question about internal groups versus mounts. Both are possible. The question becomes when we optically adjust for a tilt are we compensating for a mount tilt? I'm not always sure of the answer. Many zooms, though, are tilted at only part of their range, so one would assume that's not a mount issue. But it still could be mechanical, if it's caused by problem with one of the helicoids being slightly out of round.

Now 1.5% of you see why I find this stuff so fascinating, and 98.5% are screaming 'Go take some pictures'



Aug 26, 2014 at 06:49 PM
Gunzorro
Offline
• • • • • •
Upload & Sell: On
p.1 #13 · p.1 #13 · AFMA Settings, Comparison of Bodies


Thanks Roger! Your experience is invaluable in determining what might be realistic to understand, and what we simply don't have the ability to quantify.

Yes, it is amazing that phase detect focus works! So much to potentially go wrong! It seems there should be a way to use improved camera computing power with contrast detect (edge contrast) to establish quick and exact focus across the frame -- obviously, the physics is much harder to discover than to imagine!

Your remark about center AF is interesting, and I wanted to mention that is all I use for establishing MFA. As you said, there can be physical alignment issues as well as optical (micro alignment of elements and internal mechanisms) aberrations that could skew critical focus alignment (similar to astigmatism?).

My point is pretty elementary: how do different models respond to establishing reliable MFA adjustments? It seems some might be better than others, but perhaps that is only naive thinking. It is possibly a coincidence that two of my bodies are now quite accurate, when before they would "drift" (sometimes focusing accurately, other times being OOF -- shot to shot). After repair, they were what I consider better than acceptable.

Anyhow, I appreciate knowing to what degree other photographers need to set their bodies' baseline average (C.Fn III, 7, 1: Adjust all by the same amount), based on their overall testing of a large number of samples (lens types).

boingyman -- I'm not talking about opinion for when something is in best focus. I expect competent users can establish when a photo is in the best possible focus for the lens being evaluated (with enough magnification of details). It involves experience and critical observation, not opinion, regardless of the method used to establish or refine the correct focus.



Aug 26, 2014 at 07:02 PM
gdanmitchell
Offline
• • • • • • •
Upload & Sell: Off
p.1 #14 · p.1 #14 · AFMA Settings, Comparison of Bodies


I liked the "the more amazed I am that it works at all" line. Looking at many complex systems, this thought sometimes comes to mind. (I once took a class from American composer Lou Harrison, who was also a builder of beautiful musical instruments. At one point he reviewed all of the things that have to happen "perfectly" in order for a musical performance to occur, and he came to much the same conclusion: it is a miracle that music happens at all.)

Excellence is a wonderful thing, and I'll bet we all strive mightily for it. Perfection, on the other hand, really doesn't exist.

Dan



Aug 26, 2014 at 07:54 PM
jcolwell
Offline
• • • • • • •
Upload & Sell: On
p.1 #15 · p.1 #15 · AFMA Settings, Comparison of Bodies


gdanmitchell wrote:
.... Perfection, on the other hand, really doesn't exist.


I agree. Anyway, if it did, it would soon get pretty boring.




Aug 26, 2014 at 08:13 PM
philshoots
Offline
• • •
Upload & Sell: Off
p.1 #16 · p.1 #16 · AFMA Settings, Comparison of Bodies


Thanks for your response and the pm Roger. I come at this from a different angle. I’ll explain, as a cinematographer from the film era the FFD [Flange to Focal Distance, measured in thousands/inch] was/is routinely checked with a depth gauge on film cameras before every shoot. If the flange needed adjustment it was re-shimmed to tolerance spec so the image collimated perfectly and very exactly on the film held in the gate. Back in the day the same went for your still camera flange it was shimmed to spec. All sensible Directors of Photography also set up at least one prep day before a shoot to project all the lenses booked for the job in a lens testing booth - most every cine rental facility worldwide has at least one, some have a dozen. I've always found this part of the process fascinating, I'm clearly in the 1.5% Roger!

At this time one can evaluate many of the possible optical flaws of each lens copy as well as learn the optimized shooting stops. By optimized I'm referring to the stops that yield the highest resolution in terms of line pairs per mm. This can be clearly seen during the process and it's still surprises me that within a stop or two some lenses can gain or loose a shocking amount of resolution. Some, like me, also take this opportunity to project their still lenses - long still teles are commonly used in cine production. Over the years it has become clear that every lens is an individual with variations even in top optics from the same maker/model and I'm speaking of Zeiss PL mount motion picture primes like Arri Super Speeds, Master, Ultra Primes or Cookes. All lenses with fairly limited production, fastidious quality control and sizeable double-digit price tags. Zooms are a bit different because a zoom is basically a compromise lens optically, so the majority commonly excel at some point in their range and fall short in others. The trick was learning where not to shoot your particular copy. The comparatively large production runs of still lenses both prime and zoom allow for more QC variations so the numbers you're seeing with tilt issues, off centering or with bumps in zooms isn't surprising Roger. Its accepted and expected but it always surprises me to see people dangling big optics not properly supported off the comparatively thin body/lens flanges on still cameras. They are easier to bend than you may realize and many never give it a second thought. Incidentally PL refers to Arri’s massive stainless steel Positive Lock lens flange system, the industry standard mount and if they can be bent without doubt that flimsy flange on your still camera will bend too and it doesn't have to be by much, we are talking thousands, before the image is affected.

On the camera side when the bulk of cinema origination went HD/digital the FFD became a hot issue with DPs literally. Instead of film in the gate the lens now collimated the image on an electronic sensor. The new HD cameras heated up quite a bit during the shooting day – expanding enough to change the FFD significantly. Still shooters think MFA. Remember we are talking a FFD measured in thousands/microns tolerance to hit the sensor spot on. Considering that filming often dictates shooting wide open for artistic reasons suddenly focus became a big issue. So much so that Zeiss now offers a field collimator so the FFD can be checked and adjusted during the shooting day. Skilled pro motion picture crews now see this as common practice.

I don't know about you but I've had days sitting on the shutter button blazing away on my 1Dx/D4 or when shooting motion with my 5D2/3 when the body got noticeably hot in hand. Has the FFD/MFA changed during shooting? Absolutely. Are most photographers aware or care, I doubt it. Does it affect their images? Probably not, the stop carries them. And there is always 10x live view to check critical. But without question the MFA will change with heat during use or environmental factors so it’s a moving target. On top if that Roger has told me with cost cutting some digital bodies lack any adjustment at all.

So there's little worry about attaining perfection. But being aware the many factors that affect your MFA numbers and ultimately your IQ makes the game that much more interesting. And for this 1.5%er it's jolly good fun too.

Phil



Aug 27, 2014 at 03:58 AM
RCicala
Offline
• • • •
Upload & Sell: Off
p.1 #17 · p.1 #17 · AFMA Settings, Comparison of Bodies


Phil, you've hit the nail right on the head.


Aug 27, 2014 at 05:58 AM
philshoots
Offline
• • •
Upload & Sell: Off
p.1 #18 · p.1 #18 · AFMA Settings, Comparison of Bodies


Cheers Roger.

I like your hat.

best wishes,
Phil



Aug 27, 2014 at 06:35 AM
Gunzorro
Offline
• • • • • •
Upload & Sell: On
p.1 #19 · p.1 #19 · AFMA Settings, Comparison of Bodies


Awesome, Phil!

We still shooters frequently lose sight of what science the cinematographer has to know and constantly monitor to ensure repeatable results (as close to perfection as we can get!). It's daunting to realize what a big subject this is.

To continue what you've said, especially for shooting stills wide open, I can see that the MFA should be checked and adjusted with extreme changes of temperature. Let's say I MFA'd a lens in California in "winter" at 40 degrees F, and use it in summer sun at 100F -- there is going to be expansion (or contraction, if I MFA'd the other direction, in the heat). As you say, a lot of shooting "the f-stop will cover it", but that is an aesthetic limiter.

Thanks so much for your views.



Aug 27, 2014 at 09:42 AM
scalesusa
Offline
• • • •
Upload & Sell: Off
p.1 #20 · p.1 #20 · AFMA Settings, Comparison of Bodies


RCicala wrote:
For me, personally, the more I understand phase detection AF, the less I complain about how well it works and the more amazed I am that it works at all. It may be the most amazing engineering triumph in optics.



Well said. I'm really hoping that some of the variations can be replaced by on-sensor autofocus in the next generation of bodies (not likely) and all the mirror / sub mirror / AF sensor / exposure sensor can go away. Obtaining perfect alignment in a chain of moving parts like the mirror / submirror just will not happen.

Of course, variables like mount to sensor distance and alignment still exist, and shimming the sensor isn't really easy to do, a lot of fooling around to get that deep into a camera. Still, once the sensor is properly aligned, then it should not change, and the lens mount is more accessible to shim. (with the right tools)



Aug 27, 2014 at 01:11 PM
1
       2       end




FM Forums | Canon Forum | Join Upload & Sell

1
       2       end
    
 

You are not logged in. Login or Register

Username       Or Reset password



This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.