Mishkin
Senior Member
CIA won't tolerate 5-pixel blur. I bet their AF systems are optimized to provide maximum information, that is, deliver pixel-sharp pictures.
Mishkin
--
Mishkin
-
Welcome to the new forums! Please read this first. For known issues we are working to resolve, click here.
Focus Precision
- Thread starter Isaac Sibson
- Start date
Larry H. Smith
Veteran Member
I very much appreciate your detailed response.
Re. your comment about the 1Ds having a "different" system:
Speaking of using the 1Ds, with such lens-quality as the 135/2, 300/2.8, or the Canon macro 185(obviously a seperate, slower case), do you have an opinion as to which VF screen would BEST permit MF, while LEAST hindering AF performance, ...or is some compromise to one or the other inevitable, ...and to what extent, if so?
Thanks, (I don't imagine these to be easy questions to answer, although I hope they may be for you ;-) ...but this issue is too-little mentioned in the info I've seen.)
Larry
Re. your comment about the 1Ds having a "different" system:
Speaking of using the 1Ds, with such lens-quality as the 135/2, 300/2.8, or the Canon macro 185(obviously a seperate, slower case), do you have an opinion as to which VF screen would BEST permit MF, while LEAST hindering AF performance, ...or is some compromise to one or the other inevitable, ...and to what extent, if so?
Thanks, (I don't imagine these to be easy questions to answer, although I hope they may be for you ;-) ...but this issue is too-little mentioned in the info I've seen.)
Larry
Mishkin
Senior Member
In 10 years, when 3000x2000 and higher resolution monitors will be common, you will open your old 6MP photos, which you thought 5-pixel blur is acceptable, and cry.
--
Mishkin
Isaac Sibson
Senior Member
It's a very good question.
The focus precision is really a worst-case scenario result. Generally the camera will get rather closer than the precision limit.
So, if it falls within 50% or 33% of the DOF, you can regard it as having stuck to the accuracy but for a smaller CoC.
One assumes that in a static test such as those that phil is performing that the lens is carefully manually focused using a magnifier and then stopped down to F8-F11 or so, so that there is no question of minor variations in the focusing being an issue.
It's funny to think, is it not, that while people are complaining about poor performance that generally the camera does much better than the specification suggests?
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
The focus precision is really a worst-case scenario result. Generally the camera will get rather closer than the precision limit.
So, if it falls within 50% or 33% of the DOF, you can regard it as having stuck to the accuracy but for a smaller CoC.
One assumes that in a static test such as those that phil is performing that the lens is carefully manually focused using a magnifier and then stopped down to F8-F11 or so, so that there is no question of minor variations in the focusing being an issue.
It's funny to think, is it not, that while people are complaining about poor performance that generally the camera does much better than the specification suggests?
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
Mark Tuccillo
Leading Member
Having been caught behind Canon in autofocusing technology in the early 90's, Nikon's latest technology is now very good, and in most cases better then Canon's. They expecially excel in low light performance. I was told by a Nikon rep several years ago that focusing accuracy was the number one objective of any new development, and this seems to bear out by the complete absence of focusing complaints of the D100.
Isaac Sibson
Senior Member
The Ec-B split-circle screen is the best choice for critical manual focusing.
No focus screen actually has any impact upon autofocus performance because the AF system takes light that passes through the main mirror to a secondary mirror. The focus screen is not in this lightpath.
However, focus screens DO affect metering, and you'll find CF 0 on the EOS 3, 1v, 1d and 1ds is there to compensate for the difference in brightness of Laser Matte and New Laser Matte screens. The split-circle Ec-B screen also prevents accurate use of the central spot-metering and evaluative metering modes (although you can still use off-centre spot metering).
However, if you do not wish to have the split circle in the viewfinder in general use, my choice is the Ec-CIII, as the focus snap on it, particularly with a bright lens like those mentioned (and even with my 300 F4L IS + 2x TC) is perfectly usable for field manual focusing, and should work very nicely for macro work if combined with the magnifer.
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
No focus screen actually has any impact upon autofocus performance because the AF system takes light that passes through the main mirror to a secondary mirror. The focus screen is not in this lightpath.
However, focus screens DO affect metering, and you'll find CF 0 on the EOS 3, 1v, 1d and 1ds is there to compensate for the difference in brightness of Laser Matte and New Laser Matte screens. The split-circle Ec-B screen also prevents accurate use of the central spot-metering and evaluative metering modes (although you can still use off-centre spot metering).
However, if you do not wish to have the split circle in the viewfinder in general use, my choice is the Ec-CIII, as the focus snap on it, particularly with a bright lens like those mentioned (and even with my 300 F4L IS + 2x TC) is perfectly usable for field manual focusing, and should work very nicely for macro work if combined with the magnifer.
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
Mishkin
Senior Member
FrancisCF
Leading Member
You could be right.
Just one point I've notice, the only sharpest or in focus plane is only one, the pin point will be larger and larger in the front and back, but do notice that the relation is not linear, that is, the sharpness doesn't decrease so fast at the beginning but rather fast later on. Do you notice it or just my own perception?
Go ahead, never look back
Just one point I've notice, the only sharpest or in focus plane is only one, the pin point will be larger and larger in the front and back, but do notice that the relation is not linear, that is, the sharpness doesn't decrease so fast at the beginning but rather fast later on. Do you notice it or just my own perception?
--
Go ahead, never look back
Mishkin
Senior Member
Since you seem to have extensive knowledge of EOS, what are these sensors on 10D? There are no such sensors on D60.
Mishkin
--
Mishkin
Larry H. Smith
Veteran Member
Isaac Sibson wrote:
I'll give these two screens a try when the time comes (waiting on the 1Ds to come down some ;-):
Ec-B Ec-CIII
Very helpful!
Larry
I'll give these two screens a try when the time comes (waiting on the 1Ds to come down some ;-):
Ec-B Ec-CIII
Very helpful!
Larry
dciobota
Leading Member
...Canon cameras usually outperform their specs (90%) of the time. The other 10% is what I believe is causing all the furor lately. Funny to me that most D60 users never noticed this "feature", they probably attributed it to the quirky AF and gave it another go. Maybe this is why D30 users raved about the sharpness of their pictures, since the COC is closer to the D30 pixel size. And you know, it would've been even funnier if way back when film users started to examine each grain of film under a loupe and started complaining about the grains being out of focus! ;-)
I own a D30 btw, and it's happened to me as well. I had the camera checked and the AF sensors adjusted, and it got much better. Did it solve the problem completely? Nope, but when it happens I just reframe and reshoot. This is where digital shines, take ten pics, choose the best one and chunk the rest... ok, not great for action pics, but still.
I don't know how Phil and others shoot their tests, but I hope they don't manual focus, or at least they should definitely use a magnifier and have better eyes than me. This has been and will be my biggest gripe about these sub-frame slr's, the viewfinder is about useless for judging focus. I'll hold off buying a newer camera until they have that fixed somehow.
Thanks for the info, and I'll throw this message in the wind for Canon: if you need a good image processing software engineer to help improve the AF, I know I can do it. ;-) Now back to focusing on the 10D out-of-focus "feature". ;-)
Daniel
I own a D30 btw, and it's happened to me as well. I had the camera checked and the AF sensors adjusted, and it got much better. Did it solve the problem completely? Nope, but when it happens I just reframe and reshoot. This is where digital shines, take ten pics, choose the best one and chunk the rest... ok, not great for action pics, but still.
I don't know how Phil and others shoot their tests, but I hope they don't manual focus, or at least they should definitely use a magnifier and have better eyes than me. This has been and will be my biggest gripe about these sub-frame slr's, the viewfinder is about useless for judging focus. I'll hold off buying a newer camera until they have that fixed somehow.
Thanks for the info, and I'll throw this message in the wind for Canon: if you need a good image processing software engineer to help improve the AF, I know I can do it. ;-) Now back to focusing on the 10D out-of-focus "feature". ;-)
Daniel
Larry H. Smith
Veteran Member
Mishkin,
Using your own conclusion (with which I don't disagree) one may still, with adequate VF visibility, see when the star is "a point", and thus tell whether or not you have hit, or missed, ...and therefore will know whether-or-not to "try again".
AF gives you, instead, an indication that things are "good enough", (as determined by the mfgs. settings), ...and you don't have a clue as to WHERE in the spec range the actual focus IS.
Incidentally, I am working on an degree-indexed disc ( about 5" dia., from a drafting instrument, ) on the focus knob of a TV 76 scope, using parallax-focussing via separate eyepiece/flip-mirror, for digi-scoping with a G-1. I forget the exact specs at the moment, but with the focusser gearing, and the disc (with vernier), my discrimination is pretty fine. Coming from the same turn-direction each time equalizes any "lash" in the gearing., for repeating focus settings on a known target (tree-limb, for example, when shooting birds).
Larry
Using your own conclusion (with which I don't disagree) one may still, with adequate VF visibility, see when the star is "a point", and thus tell whether or not you have hit, or missed, ...and therefore will know whether-or-not to "try again".
AF gives you, instead, an indication that things are "good enough", (as determined by the mfgs. settings), ...and you don't have a clue as to WHERE in the spec range the actual focus IS.
Incidentally, I am working on an degree-indexed disc ( about 5" dia., from a drafting instrument, ) on the focus knob of a TV 76 scope, using parallax-focussing via separate eyepiece/flip-mirror, for digi-scoping with a G-1. I forget the exact specs at the moment, but with the focusser gearing, and the disc (with vernier), my discrimination is pretty fine. Coming from the same turn-direction each time equalizes any "lash" in the gearing., for repeating focus settings on a known target (tree-limb, for example, when shooting birds).
Larry
http://forums.dpreview.com/forums/read.asp?forum=1019&message=4663636
E
Erik Magnuson
Guest
Remember that this number is more likely a statistical performance specification and not a precise measurement for any given camera. Most cameras will do better. If too many cameras do worse, than it's there is a problem in design/production to be fixed.
Anyone have enough experience with similarly complex systems to suggest whether the distribution would be more uniform vs. normal?
--
Erik
Isaac Sibson
Senior Member
I couldn't say for sure. They don't feature on any EOS body that I've owned or anyone else in my family has owned, but as Adam-T has observed before, they are present on the EOS 30. Potentially they could be something to do with the 7-point AF system (in which case they should also appear on the EOS 300 and 300v), but I do not know anything more of them.
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
Isaac Sibson
Senior Member
Just note that Ec-CIII is the standard screen that the 1DS comes with. Should you decide that you want a brighter screen then Ec-N has the same markings.
--
Mostly Full Frame user!
EOS Tree + Nikon Coolscan III
Deef Hurty.
Larry H. Smith
Veteran Member
Mishkin
Senior Member
You're right, Larry, if we were able to see at pixel level (in angular equivalent) in the VF, even the play in the focus ring won't prevent us from trying again and finally achieving precise focus.
However, by my estimate, 1 circular minute (which is average resolution power of human eye) corresponds roughly to 2 pixels in 10D's VF. Pad it by a factor of 2 or 3 (since 1' is the extinction resolution), and we get perhaps about 5 pixels that we can see in the VF. In other words, when we manually focus 10D/D60 and think that what we see is tack-sharp, it could be off by whopping 5 pixels or even more on the sensor.
Angle Finder C with its 2.5x helps (as reported by others, I don't own one yet), but it's a kludgy accessory in everyday situations.
I wish DSLRs had good means of manual focusing like telescopes, microscopes, etc. do.
Mishkin
However, by my estimate, 1 circular minute (which is average resolution power of human eye) corresponds roughly to 2 pixels in 10D's VF. Pad it by a factor of 2 or 3 (since 1' is the extinction resolution), and we get perhaps about 5 pixels that we can see in the VF. In other words, when we manually focus 10D/D60 and think that what we see is tack-sharp, it could be off by whopping 5 pixels or even more on the sensor.
Angle Finder C with its 2.5x helps (as reported by others, I don't own one yet), but it's a kludgy accessory in everyday situations.
I wish DSLRs had good means of manual focusing like telescopes, microscopes, etc. do.
--
Mishkin
Hi, you can use this:
http://www.rainpalm.com/Images/DOF%20Calc.xls
You can set the COC to anything you like, but I set it to match the D30.
Cheers
Ian S
http://www.rainpalm.com
http://www.mekongpicturehouse.com
http://www.rainpalm.com/Images/DOF%20Calc.xls
You can set the COC to anything you like, but I set it to match the D30.
Cheers
--
Ian S
http://www.rainpalm.com
http://www.mekongpicturehouse.com
FrancisCF
Leading Member
I'm sure this is no relation to the 7-point AF sensor according to their position. For AF, some light pass through the main mirror and is reflected by a small second mirror behind it, the light is reflected to the AF sensor module in the bottom of the mirror box. The AF light should be well focus at the sensor as well as the film plane (when the mirror flips up) if correct focus is achieved. Due to the light path to the film plane and the AF module plane is different, the mirror mechanism is complicated, I'm no winder there could be any small misalignment occurs in a few percent of camera in the production line.
May be some one could shine some light on this question!
Could it be likely the TTL flash sensor? Just a guess.
Go ahead, never look back
May be some one could shine some light on this question!
Could it be likely the TTL flash sensor? Just a guess.
--
Go ahead, never look back
Mishkin
Senior Member
I guess it's due to these sensors that E-TTL on 10D magically began to work, compared to severe problems on D60. My guess is that 10D uses both E-TTL (preflash metering through the lens) AND regular TTL (metering overall reflected light off the film... err... sensor plane during the exposure).
Mishkin
--
Mishkin
Keyboard shortcuts
- f
- Forum