Take your photography to the next level and beyond...

  • NEWS
  • REVIEWS
  • INSPIRATION
  • COMMUNITY
  • COMPETITIONS

Why not join for free today?

Join for Free

Your total photography experience starts here


PRIZES GALORE! Enter The ePHOTOzine Exclusive Christmas Prize Draw; Over £10,000 Worth of Prizes! Plus A Gift For Everybody On Christmas Day!

Canon 20D soft photos


Steppenwolf 3 1.0k
19 Apr 2014 2:50PM

Quote:
Its unlikely to be the positioning of the unit - with CNC machining these days positioning will be dead accurate, and they are calibrated in position - unless you are most unlucky.



What happens is that the DSLR (in the case of non-professional cameras, anyway) is assembled on a mass production line to whatever tolerances the manufacturer thinks it can get away with - it's never "dead accurate". The camera is then tested at the end of the line to check how accurate the focus is (with some generic lens and standard focus range). If it's out of focus the discrepancy is calculated and this is fed into the firmware as a correction which the camera then applies after the camera has focused but before the photo is taken - this is the "calibration" that happens. If it's outside tolerances then it's rejected - in theory.

The trouble is that testing a camera with one lens and one object distance doesn't really work very well. It's better than nothing but it's never going to work properly. It also relies on the lens ROM providing accurate data on lens gearing (and direction) which is why some non-OEM lenses don't focus properly on some cameras.

It's a design flaw - which was forgivable when the DSLR was the only way you could look through the lens - but is now an anachronism. Time to move on.

Join ePHOTOzine for free and remove these adverts.

MGJ 6 168 5
19 Apr 2014 9:07PM
There is ALWAYS a tolerance. Greater in some areas than others.
AndyMoore 12 6 United Kingdom
19 Apr 2014 9:28PM
Well I went out and purchased an EOS 100D today as my budget was limited but the difference is astounding. Super sharp photos from my test shots, I'm really pleased with it so far. More testing will need to be done but so far it's been like night and day. I'll put up a shot from it once I've posted this.

Thanks everyone for your help and input into this thread.

Grin
19 Apr 2014 9:35PM

Quote:Try photographing a brick wall !!...


As has been said AF can be a bit hit and miss depending on how accurately the AF can assess the detail at the AF point.
Brick walls are usually geometric patterns.
AF cannot always accurately detect geometric patterns, especially if you do not centre the AF point with peect accuracy.
While a brick wall can be good 8 times out of 10 testing is best done with a subject good 10 times out of 10 Grin
19 Apr 2014 10:40PM

Quote:Well I went out and purchased an EOS 100D today as my budget was limited but the difference is astounding. Super sharp photos from my test shots, I'm really pleased with it so far. More testing will need to be done but so far it's been like night and day. I'll put up a shot from it once I've posted this.

Thanks everyone for your help and input into this thread.

Grin


Congratulations on your new camera! I was eyeing it for a while myself, but having already two perfectly fine DSLRs on my hands by now I am reluctant to buy anything new. These days I pay more attention to image composition and development. Cameras simply don't die on me - I love them, and it seems to be mutualSmileSmileWink
MGJ 6 168 5
20 Apr 2014 10:10AM
Again - I'm astonished why people get these autofocus problems. Maybe I have just been lucky with my 4 SLRs, and they have all been good'uns, but I doubt it. Yes, you can get a problem in low light and at wide apertures, but by and large the potential for error would seem to be over hyped in normal conditions. Yes, its a machine so it can fail or be used outside its operating envelope, and the system - any system, has limitations.

(Like aeroplanes fly 9/10, and passengers trust them to do so, but once in a while someone puts one in an unhappy condition, and all goes pear shaped))

So. almost invariably its going to work fine - unless there is something wrong with the set up, when there is likely to be a consistent error. Defective units generally have inconsistent errors.

Hell it doesn't have to be that accurate. When you are measuring phase shifts with respect to the wavelength of light, (and you can do that quite easily and cheaply nowadays as long as you can "see" what you are looking at) a hundredth of a millimetre focus error is a hell of a long way. (The amateur in a home workshop can machine to a thousandth of a milimetre for those who want to talk tolerances, and the pro using laser interference measuring can get closer than that. Digital readouts on machine tools are bought off the shelf accurate to a micron or .001mm - 1/1000, and specialist ones a lot better than that. )

So let us keep it all in perspective.
20 Apr 2014 11:55AM

Quote:Again - I'm astonished why people get these autofocus problems. Maybe I have just been lucky with my 4 SLRs, and they have all been good'uns, but I doubt it....


I had a small talk on this above, but let me explain a bit deeper. I start with an example. I have one of those "weird" lenses which front-focuses on one of my DSLRs, EOS550D, but works just fine on the other, EOS10D. Interestingly enough, it is EF lens by Sigma ( 18-125mm). Despite having EF mount it was designed specifically for APS-C sensor, not full frame like in original EF mount equipped cameras. Yes, EF-S cameras work with EF mount lens, but then my first (and only known to me) EF mount APS-C camera EOS10D focuses it perfectly while EF-S EOS550D needs some degree of care with it, but focuses pin sharp with my other Canon lenses.

You are correct, there are some tolerances. I also strongly suspect that they were different for "semi-pro" EOS10D from "advanced beginner" EOS550D. The biggest trouble with the lens comes at maximum zoom and wide open aperture anyway.And I am doing digital photography long enough to see mirrorlless cameras fail under the same sort of circumstances. The photographer needs to know limits and particulars of the cameras he uses, and adjust to them. A perfect "do it all for me" camera never existed, and not very likely to be ever made, especially in "serious" amateur/professional category which DSLRs undoubtedly belong to.

Cheers!
MGJ 6 168 5
20 Apr 2014 2:15PM
Michael -you are quite right. The system is made up of several building blocks, each of which has its own limits. You do get additions of tolerances too - sometimes they work for you and sometimes against.

At the end of the day, certain components need placing and calibrating accurately, whatever system, and all have their limitations.

I know its a DSLR, but my 70D focuses fine, in poor light too. (so did they all) But the Sigma 18-35 Art isn't quite happy wide open in poor light. But that lens is fine at 2.8 as are other lenses. Or use liveview, if one likes it, which I don't.
Steppenwolf 3 1.0k
20 Apr 2014 2:28PM

Quote:Again - I'm astonished why people get these autofocus problems.


That doesn't surprise me as you don't seem to understand what's involved in getting an accurately focused image when using a remote AF sensor and a moving mirror. This is what David Kilpatrick said on Photoclubalpha:

"DSLR AF involves about eight separate collimation adjustments. The focusing screen (1) must be perfectly located relative to the adjustable sensor plane (2), and the primary 45 mirror must then be perfectly aligned (3). Then there is a secondary mirror behind the main one (4), a return mirror (5) and then the AF module itself which sits on a carriage controlled by three screws. Add to this similar multiple screws to adjust the sensor mounting, and the setup of the lens and the lens mount relative to the whole assembly (7 and still counting). Its a highly complex system." Et cetera. He could have added that, in the case of non-OEM lenses, the ROM has to be perfectly compatible with the firmware of the camera - or the final correction will be wrong.

Contrast this with the mirrorless camera's system where the focus is measured at the image sensor - that's why it works. It used to be the case that mirrorless cameras had to rely on CDAF - which is slow. However, now that PDAF can be put in the sensor there's absolutely no need for a remote AF sensor - except, of course, if you insist on an optical view finder. It's called "design".

I don't know if you've been lucky with your 4 DSLRs or whether it's just that the lens focal length/aperture you tend to use gives you a big enough depth of field for the problem not to be apparent. I've had 3 DSLRs and they've all been inaccurate - one really badly so. I tend to use long big aperture lenses so if it's not right it's glaringly obvious. It's very easy to check the focus anyway - google it.

When I bought my current camera (an SLT), I made sure that it had micro-tuning (by lens) - so that if it didn't focus there was at least something that I could do about it. Predictably - isn't it always like this - it focused perfectly.

It's very easy to dismiss those who have focus problems as idiots who don't understand how to use their equipment. No doubt many are - but there is a design problem with focus on DSLRs. The accuracy required nowadays for correct focus (with modern high resolution sensors) is much greater than was the case 50 years ago with film - where the lack of planarity of the film was the main problem.
Steppenwolf 3 1.0k
20 Apr 2014 2:41PM

Quote:I have one of those "weird" lenses which front-focuses on one of my DSLRs, EOS550D, but works just fine on the other, EOS10D. Interestingly enough, it is EF lens by Sigma ( 18-125mm).


There are two reasons (that I know of) why a lens can "throw" a camera's focus.

Firstly it can be because the lens has differential focus. This means that the outer part of the lens focuses to a different point than the inner part of the lens - this tends to afflict big aperture wide angle lenses. Since the open-aperture DSLR focus preferentially uses the rays from the outer part of the lens (because PDAF relies on angle) it may focus wrongly if you're using a small aperture - or just look blurred at all apertures.

Secondly, if you're using a non OEM lens (e.g. Sigma) it may be that your camera doesn't recognise the lens (highly likely) - or that Sigma have "spoofed" the lens identity to match an OEM lens that is not exactly the same. Provided that your camera is reasonably accurately assembled this may not matter (much) but it will matter more if your camera is on the edges of the tolerances.

You see, none of this matter to the mirrorless cameras. No need for "corrections" or micro-tuning" etc.
MGJ 6 168 5
20 Apr 2014 5:24PM
Steppenwolf -I understand very well.

I'm a control engineer, by training.
Niknut e2
4 733 64 United Kingdom
20 Apr 2014 5:46PM
GrinGrinGrinGrin !!......thought you might go for the 100D !!......damn good choice, with top quality IQ !!....enjoy !!Smile

Sign In

You must be a member to leave a comment.

ePHOTOzine, the web's friendliest photography community.

Join For Free

Upload photos, chat with photographers, win prizes and much more.