Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
07-29-2009, 10:33 PM   #76
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by foxglove Quote
Ah! These cameras only allow centre-weighted metering with manual lenses, then.
I think spot might work too - maybe only with "A" lenses? I forget. But multi-segment is definitely out - if you have that set normally, it automatically changes to center-weighted when you mount a manual lens. And spot is kind of useless a a general purpose metering mode - you pretty much can *never* just point, green button, and shoot with spot meter and expect anything reasonable.

(and you're welcome!)

07-29-2009, 10:36 PM   #77
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,937
QuoteOriginally posted by PentaxPoke Quote
I have never talked about AF select. That is no different than what I do which is to always "select" the center.
The difference is that by "always selecting the centre" you have to recompose if the subject is not the centre. By manually selecting on of the other AF points, you may not need to recompose. I thought this was what the "recomposition" discussion was about. Of course you are free to throw in a comparison with "AF Auto" as well, but I just wanted to clarify that this is off-topic as far as recomposition is concerned.

What do you think of my math/situations?
Note that I added a bit about using a focal length higher than 50mm.

I'm happy for others to recompose and surely many times the error introduced will be insignificant but it is not the case that this is a non-issue for all situations, is it?
07-29-2009, 10:52 PM   #78
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by falconeye Quote
As for the recompose DoF error: you can easily compute the effect.
Hmm. I have a degree in math, but I highly distrust calculation as a way of figuring this out. DOF is more subjective than the formulas generally make it out to be (in addition to distance, focal length, and apertue, it also depends on reproduction size, viewing distance, the observer's visual acuity, and his/her tolerance for slight blur).

So rather than a formula to tell me how big this effect *should* be, I've tried on several occasions to reproduce the effect myself. I can, to a small degree (that is, I can see the shift in the zone of acceptable focus), reproduce it even at f/2.8 with a 50mm lens. I can see it to a potentially problematic degree at f/1.7, but even then it depends on subject distance, how far you are recomposing, etc. Focus recompose might not be perfect, but it pretty clearly doesn't "suck".

BTW, one fairly obvious flaw in the article in question is the claim that good lenses have perfectly flat plans of focus. Good macro lenses strive for this, and good lenses of any kind do try to be flatter than the default circular arc you'd expect, but most fall somewhere in between.
07-29-2009, 10:55 PM   #79
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by Class A Quote
BTW, for shots involving movement, recomposition still sucks.
If you're not practiced at it. Similarly, if you're not practiced at selecting focus points, that's going to be pretty hopeless for tracking movement, too. I'm not convinced one will suck less than the other when you compare the abilities of people experienced with each. Personally, I tend to use center / focus-recompose to get me there for the first shot, then MF (all my AF lenses have quick shift) to track motion after that.

07-30-2009, 01:00 AM   #80
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,937
QuoteOriginally posted by Marc Sabatella Quote
If you're not practiced at it. Similarly, if you're not practiced at selecting focus points, that's going to be pretty hopeless for tracking movement, too.
The scenario I'm thinking of is the following:
You want to take a picture of an ice hockey player when he is charging the goal. The player should be in focus but in the composition you want the player to be off-centre. To do this, select the appropriate off-centre AF point. Leave it there; no more work to do. Keep focusing (e.g., with AF-C or by repeatedly activating AF-S) panning along with the player as he moves towards the goal. In the decisive moment ( ), say when the goal enters the frame, you press the shutter button. The result is an off-centre player in focus.

I find it hard to believe that you can do this with a recomposition technique, even if you are trained to do so.

QuoteOriginally posted by Marc Sabatella Quote
Personally, I tend to use center / focus-recompose to get me there for the first shot, then MF (all my AF lenses have quick shift) to track motion after that.
Good on you, if you can pull that off.
But then you are mainly using an MF technique for which AF points are not really required.
07-30-2009, 09:43 AM   #81
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
QuoteOriginally posted by Marc Sabatella Quote
Hmm. I have a degree in math, but I highly distrust calculation as a way of figuring this out. DOF is more subjective than the formulas

Anyway, no reason to distrust. Just keep in mind that the Circle of Confusion (CoC) as used for DoF calculations is 20Ám wheras a pixel is 5Ám only, i.e., the CoC is 4 pixels wide.

If you compute the DoSP (Depth of Sharp Pixels), you therefore get 0.29cm in front and behind the subject. So indeed, the shift by 1.2cm makes you significantly loose full sharpness at 100% crop pixel peeping. So, pixel peeping will indeed reveal the effect. This would even hold true (to a lesser extent) for d=3.2m distance or f/2.8.

But this is a concern only if you crop. And if you do, why did you recompose in the first place

QuoteOriginally posted by Class A Quote
When I derive the formula (with simple trigonometry) I get
x = d (1/(cos 9░) - 1)
Class A,
I stand corrected What great forum members we have
To my defense, what I wrote
QuoteQuote:
This moves the subject away from focal plane by x = ...
was correct. But light won't take the shortest path to the focal plane, but the extrapolation of a line through lens and subject.

For normal shifts, the difference is marginal (as you pointed out).

But for a full 90░ turn, my formula yields x=d (the focal plane is still there at some finite distance) where your formula correctly yields infinite (because the subject actually moved out of image for any rectilinear lens).
QuoteOriginally posted by Class A Quote
Also note that if you are using a focal length higher than 50mm, the DOF will become shorter [...] the recomposition error one may be able to introduce also reduces
Yes, smaller FoV and smaller DoF counteract. If I have a spare minute, I'll check if both effects don't actually cancel out for say, a given magnification.
QuoteOriginally posted by Class A Quote
Come on, what are you portraying here? A bus?
I revisited my claim that 3.2m at 50mm APS.C portraits a 50cm field. It was 2AM at night when I wrote it ... and I was wrong again

The correct distance for a 50cm FoV is 1.04m.

I've now rewritten this and my original post to correct for my mistakes. As it turns out, recompose may indeed induce a visible error in portraits shot at f/1.4.

Last edited by falconeye; 07-30-2009 at 10:59 AM.
07-30-2009, 09:49 AM   #82
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by Class A Quote
The scenario I'm thinking of is the following:
You want to take a picture of an ice hockey player when he is charging the goal. The player should be in focus but in the composition you want the player to be off-centre. To do this, select the appropriate off-centre AF point. Leave it there; no more work to do. Keep focusing (e.g., with AF-C or by repeatedly activating AF-S) panning along with the player as he moves towards the goal. In the decisive moment ( ), say when the goal enters the frame, you press the shutter button. The result is an off-centre player in focus.
Good point. Not something I would normally think about doing, but indeed, if that the effect you want - a perfectly timed action shot with an off-center subject - focus-recompose is not the best method. Like I said, I'd tend to do this with one focus-compose to get me started, then quick shift after that. I'm pretty good at that. Also, if the goal (sorry) is to shoot the player with the puck at the goal, simple pre-focus works extremely well, too - just focus a bit in front of the goal with AF disabled (I have the OK button on my K200D set to do this) then shoot when ready. Yes, I know, it's not precisely the same thing, but my point is, there are lots of alternatives that do *well enough* that I have no incentive to change my methods.

I have played with selectable focus points, but it usually just took far too long to get the focus point where I wanted - and chances are none of the points were positioned perfectly, so some amount of recompose was still needed. And ten half the time I'd forget to center or otherwise set the focus point appropriately before the next shot. I'm sure the latter mistakes would disappear with sufficient practice, but it just struck me as a complicated solution to what was, for me, a non-problem, so I went back to center only. The fact that my MF lenses only support the center point for confirmation gives me added incentive to just deal with that.

But I can see that for someone who shoots a lot of sports or other action, and virtually always uses AF lenses so they aren't concerned with needing different techniques for different lenses, or uses AF lenses without quick shift that make my technique much more difficult, I could see why getting accustomed to the selectable focus points would be worth the effort.
07-30-2009, 10:04 AM   #83
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by falconeye Quote

Anyway, no reason to distrust. Just keep in mind that the Circle of Confusion (CoC) as used for DoF calculations is 20Ám wheras a pixel is 5Ám only, i.e., the CoC is 4 pixels wide.
You're using numbers to convince me I should trust numbers? :-)

Seriously, I don't see how that resolves anything for me. My issue is that I know from experience that something that looks in focus when viewing an image at one size might look to be just outside the focus zone when viewing the image bigger. So there is not and cannot ever be a single number that tells me what DOF *is*. That's why I don't really like DOF calculations in general, although I recognize they can tell me some thing in broad general terms.

Now, I suppose if you calculate DOF for the worst case scenario - pixel peeping at 100% - and can then show that the subject will always remain "comfortably" within the DOF at a given focal length, aperture, and distance, then I guess I'd believe that there is *nothing* to worry about with focus recompose. But since my own empirical tests have shown this is *not* the case - I *can* make an object noticeably less sharp through the use of focus-recompose under the right conditions - I know better than to even begin thinking through the calculations, as they cannot possibly convince me I haven't seen what I know I have seen.

QuoteQuote:
If you compute the DoSP (Depth of Sharp Pixels), you therefore get 2.8cm in front and behind the subject. So indeed, the shift by 4cm makes you loose full sharpness at 100% crop pixel peeping. So, pixel peeping will indeed reveal the effect.
There you go!

QuoteQuote:
But this is a concern only if you crop. And if you do, why did you recompose in the first place
Cute. Of course, for the record, recomposing could be to include something in the frame that wouldn't have been there otherwise, or you might not want to sacrifice the resolution you lose from cropping. And in an event, pixel peeping isn't only relevant for crops. But again, I agree that we are talking fairly subtle shifts that aren't really worth worrying about most of the time. From my own empirical tests, I have a sense of when it might become and issue and I know to use MF to fine tune things in those cases. So I remain a happy focus-recomposer.

07-30-2009, 11:11 AM   #84
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
I have rewritten both my recent posts.

My error was that 3.2 m distance was wrong for a portrait scenario. At the correct distance of 1m, the analysis turns out to be somewhat less comforting

QuoteOriginally posted by Marc Sabatella Quote
So there is not and cannot ever be a single number that tells me what DOF *is*.
[...]
There you go!
[...]
So I remain a happy focus-recomposer.
I think that we generally agree. And my revised numbers tell the same story as your eyes
[...]
As for single number DoF... This is an endless debate I don't want to restart here (and why I used the term DoSP to differentiate). I see what you are saying. But as a matter of fact, DoF is well defined for a given image (impression), i.e., a full scale view onto it. If you crop, or magnify on a computer screen, or approach closer to a printed image, or print larger w/o moving away, the DoF will change by definition. Because the image (impression) changes!

To be on a safe pixel peepers side with a 14.6 MP sensor (and a good lens), the DoSP is only about 1/4 of the DoF. But: the normal viewer won't notice the difference.

Personally (not used in my above analysis), I use a CoC of 12Ám. This is more pessimistic than the Zeiss formula (or 720p HD) (20Ám) but still not pixel-peeper maniac. It happens to be the fullHD resolution too

Last edited by falconeye; 07-30-2009 at 11:21 AM.
07-30-2009, 02:34 PM   #85
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,937
QuoteOriginally posted by falconeye Quote

I stand corrected What great forum members we have
Thanks, and no worries.

Rest assured I double checked when I obtained something that was in disagreement with what you had written. I know that you can be trusted in these matters (when you don't do back on the envelope calculations in the middle of the night ).

QuoteOriginally posted by falconeye Quote
Yes, smaller FoV and smaller DoF counteract. If I have a spare minute, I'll check if both effects don't actually cancel out for say, a given magnification.
I was tempted to do that myself but then I could not be bothered because it is not critical for me and no one else (with the exception of Marc) seems to be interested in this analysis anyhow.

QuoteOriginally posted by falconeye Quote
As it turns out, recompose may indeed induce a visible error in portraits shot at f/1.4.
Which confirms Marc's results and please note that your calculations are based on 100% accurate and 100% consistent AF. In reality, errors will be worse.

QuoteOriginally posted by Marc Sabatella Quote
...my point is, there are lots of alternatives that do *well enough* that I have no incentive to change my methods.
Point fully taken. I can perfectly see why your methods fit your equipment and shooting style very well. I don't want to convert anyone. I just wanted to make sure that the manual AF selection method receives the credit it deserves.

QuoteOriginally posted by Marc Sabatella Quote
But since my own empirical tests have shown this is *not* the case - I *can* make an object noticeably less sharp through the use of focus-recompose under the right conditions - I know better than to even begin thinking through the calculations, as they cannot possibly convince me I haven't seen what I know I have seen.
As we have seen it is partially a matter of doing the calculations correctly and also a matter of acknowledging where the idealisations made in modelling the problem may lead to overly conservative results.

This is why I mentioned the in reality less than perfect AF and the fact that typically one will not perform a perfect rotation but also introduce spatial translation.
07-30-2009, 05:15 PM   #86
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
QuoteOriginally posted by Class A Quote
Which confirms Marc's results and please note that your calculations are based on 100% accurate and 100% consistent AF. In reality, errors will be worse.
I suppose I should come clean about my own haphazard testing methods. Here's the way I'm most likely to test: sit on couch with my feet on coffee table, point camera at big toe, focus, snap, then recompose with toe in the corner, then shoot again without refocusing. Then I compare where on my toe the focus zone is. I can see it move from the toenail to the first joint even at f/2.8 on my DA40.

Obviously, the error induced by this sort of test has as much to do with the fact that in recomposing this way, there is no guarantee I'm keeping distance constant. But I figure, that's part of the error one deals with in the real world. Which is another reason why I am not sure DOF calculations tell the whole story. I *have* tested on tripod using a pan head and was able to confirm the same sort of effect, but again, the bottom line for me was actually to confirm that the effect is small and not worth worrying about except when it is, and in those cases I do it manually.
07-31-2009, 01:41 AM   #87
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
QuoteOriginally posted by Marc Sabatella Quote
I can see it move from the toenail to the first joint even at f/2.8 on my DA40.
Hmmh.
The focus zone should have moved to behind the toenail, though.
Did you make a reference shot before recomposition?
Non flat focal plane, moved more than turned, ... ?
07-31-2009, 09:33 AM   #88
Pentaxian
Marc Sabatella's Avatar

Join Date: Sep 2006
Location: Denver, CO
Photos: Gallery
Posts: 10,686
Yes, I shot before and after recomposition. And yes, of course you are right about what *should* happen to the focus zone, assuming a perfectly flat plane of focus. And you are equally right (or so I assume) about the explanation: to the extent the plane of focus is actually curved, then pivoting at the torso to recompose moved the camera further from my foot. Which is kind of my point - in the real world, there are enough variables to make me distrust a purely calculated answer. It's tough to predict how these factors (curvature of focus plane, center of rotation, interpretation of DOF) will interact for any given shot.

Of course, shooting at a distance of just a few feet exacerbates some of these problems, while diminishing others. DOF is smaller at a few feet than at 5 meters or at infinity, so there is less room for error. On the other hand, the actual distance by which a non-spherical plane of focus moves behind the subject for a given rotation is greater at larger focus distances. The extent to which these factors cancel each other out will depend on the actual shape of the focus plane. But then, the difference between pivoting at the torso versus rotating the camera on its own axis differs according to distances. How will this all play out in practice? Takes me a lot less time to do some tests than to work through the math - especially given that in order to get a useful result, I'd need to know the exact shape of the focus plane.

So of course, I've run the same sort of tests shooting objects at other distances. With similar results, actually - I can see the focus zone move in many cases, the specifics depending on lens, aperture, and distance. But it's pretty rare that it moves enough to make me care, and when it does, it's easy enough to fix manually.
07-31-2009, 01:58 PM   #89
Pentaxian
falconeye's Avatar

Join Date: Jan 2008
Location: Munich, Alps, Germany
Photos: Gallery
Posts: 6,862
QuoteOriginally posted by Marc Sabatella Quote
Which is kind of my point
Marc, thanks for the interesting reply. Really adds to the topic.

Guess, we can now all consider ourselves the recompose-focus experts
07-31-2009, 05:06 PM   #90
Pentaxian
Class A's Avatar

Join Date: Aug 2008
Location: Wellington, New Zealand
Posts: 8,937
QuoteOriginally posted by Marc Sabatella Quote
...in the real world, there are enough variables to make me distrust a purely calculated answer.
Just very briefly: I don't think anyone meant the calculations to exactly predict the focus shift you'll get when hand-holding and recomposing. They were useful though, to get a feel for the magnitude of error introduced by camera rotation only.

Your validation by testing does not provide a reliable answer either because I highly doubt that your testing methods lead to repeatable results in a precise manner (you called them "haphazard testing methods" yourself ).

I find it impressive that you can easily fix recomposition focus errors by using manual focus. With a stock focussing screen, which "lies" at such critical apertures, I personally don't stand a chance of accomplishing this.

Anyhow, we are all surviving our focusing challenges, one way or the other, which is the main thing.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
camera, dslr, ev, f/4, focus, image, iso, length, lens, pattern, photography, sec, tamron
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
Sharper images--auto or manual focus Dimeman Photographic Technique 10 02-05-2010 11:01 AM
OK, Which is Sharper ? wll Pentax SLR Lens Discussion 10 09-06-2009 06:27 PM
Tweaking the K20D for Sharper Images nanhi Pentax DSLR Discussion 58 06-19-2009 12:18 AM
Shouldn't this be sharper? FHPhotographer Pentax SLR Lens Discussion 17 01-17-2009 07:20 PM
What is sharper karq Pentax SLR Lens Discussion 6 10-25-2008 02:17 AM



All times are GMT -7. The time now is 12:25 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top