Canon 7D ISO Handling

[scottthehat]

I don't see why there should be any special procedure for a Sigma lens on a Nikon body. Canon have issued their own guidance on how to test, which will probably apply equally well to other brands, but you might want to double check whether there is anything peculiar you need to be aware of.

Here's the Canon version (scroll down a little to get to the right section) - http://digitaljournalist.org/issue0812/tech-tips.html

I guess one piece of advice you might take from the Canon advice is to test at not less than 50X focal length. For a lens at 500mm that means testing at a distance of 25m or more, not indoors across the dining table. Another alternative bit of advice is to calibrate at a "typical shooting distance for your photography". If you do routinely shoot your 500mm lens at 2m or 3m then you might indeed want to run your calibration at that distance, but more realistically, if you are using your long lens for shooting at large distances that should be where you perform your calibration, for the most pertinent and reliable results. You should probably also check out the performance at, say 150mm and 300mm, or something like that, in case different amounts of adjustment are required. You might need to settle for an "average" adjustment that best suits all focal lengths reasonably well, if not perfectly.

that works out that a 500mm lens should be 82 feet away from subject when testing,

 

[tdodd]

that works out that a 500mm lens should be 82 feet away from subject when testing,

Well, that's "a" recommendation. Like I said, if you shoot at closer distances then feel free to test/calibrate at a distance better suited to your own typical shooting style. I certainly don't follow the 50X rule religiously. But I also make sure I don't test everything at MFD, or anything like it. I think that would be quite silly, and not representative of real world shooting.

e.g. I have a 100-400 zoom. It is used mostly for wildlife, especially birds, which are generally around 20m away from me or more. The minimum focus distance is 1.8m. If I want to calibrate the lens at a focus distance that has most significance and benefit for me then calibration at 20m seems like an excellent plan. In contrast, I would seldom use the lens at distance between 5m-10m so there would be less value in performing calibration at such distances. 20m is a good balance, overall, for shooting at between 10m and 100m, as far as DOF etc. is concerned.

At the other extreme I do often use the lens pretty close up, like between 2m-3m, for shooting small birds like finches, tits, robins etc.. It could be worth me also performing a calibration at that much closer distance. If the adjustment needed differs significantly from that required at 20m then I have a choice. Either I send the lens in for proper calibration, I remember to change my adjustment amount depending on how I intend to use the lens, or I live the the results being less good at one distance than the other.

It really is horses for courses. Do what makes sense for you.

 

[Canon Bob]

One thing worth noting when performing micro adjust calibration is that the AF sensors perform differently with differing light sources. Artificial light has a tendency to give a little back focus and the MA values will be different to a calibration done in natural light.

It's certainly evident on the 1D3 and 5D2 but less so on the 7D.....Canon are seemingly aware of the problem and likely make use of the in-built colour temperature sensing on the 7D to adjust the focus accordingly.

I use a Lens Align Pro for MA and found that my long lenses all required less +ve MA when calibrated outdoors on the 1D3 but the difference is minimal on the 7D (on the two lenses tried so far).

Bob

 

[nigpd]

I use a Lens Align Pro

$179

 

[George]

Blimey, Bob, thats an expensive solution, is it worth it?
I shoot newsprint stuck to a wall outside (when the snow melts....!!!)
George.

 

[Canon Bob]

Blimey, Bob, thats an expensive solution, is it worth it?
I shoot newsprint stuck to a wall outside (when the snow melts....!!!)
George.

Things are relative George......4 adjustable bodies and 20 lenses....works out about £1.50/lens to ensure good calibration....even a Scotsman wouldn't call that expensive, surely

Bob

 

[George]

Bob, that's a few wee drams with the haggis we're talking about....serious stuff!!!

 

[Canon Bob]

Bob, that's a few wee drams with the haggis we're talking about....serious stuff!!!

The Lens Align will improve focus whilst "a few wee drams" is very likely to have the opposite effect.

Bob

 

[CT]

The Lens Align will improve focus whilst "a few wee drams" is very likely to have the opposite effect.

Bob

Yes Bob but would you care? hic!

 

[George]

I'm hoping Canon's next invenshun is leg stabilisashun

 

[Canon Bob]

Yes Bob but would you care? hic!

Believe it or not, I'm teetotal so the haggis holds more appeal for me

I'm hoping Canon's next invenshun is leg stabilisashun

I reckon a Wimberley II mounted on a Zimmer frame is as close as you'll get.

Bob

 

[artyman]

I had a go with my Sigma 150-500 a couple of days ago, at around 18ft, I had it all set up on a tripod, focused, then liveview to flip the mirror and 10 sec delay to remove vibration. The results were mixed, so I tried maxing out at each end of the adjustment and looked to see which was worse. I settled on a -5 setting, later I took some bird shots and they seem to be a bit crisper, now whether this is just fanciful I'm not sure. Hopefully results over the coming days will confirm if it is an improvement or not.

One thing I'm pretty sure about though is that any slight misfocus is probably no more than 10mm either way, possibly significant on a small bird, but I'm wondering if this is getting to be obsessive, a bit like pixel peeping and noise, are we chasing our tails looking for image nirvana rather than just shootings pics.

 

[tdodd]

I've seen it written that with lenses slower than f/2.8 Canon AF is only specced to be accurate to somewhere within the DOF, and not to the exact mm. Lenses of f/2.8 and faster should be focused to an accuracy of within 1/3 DOF, but only when coupled with high precision AF points, which often may only be the centre point.

Assuming the above to be true, I don't know whether that margin of error is intended to cater for calibration errors only, with a high degree of repeatability from one shot to the next, or whether it is designed to include variable focuing precision from one shot to the next. I suspect that it is mostly the former. However, Chuck Westfall does recommend taking at least three shots at each microadjustment value, in order to help eliminate a rogue/spurious focus which might throw you, so I guess some variation in shot to shot accuracy is expected. It is also recommended that you manually defocus the lens in between each AF test shot. Do not AF once and then repeatedly keep hitting the AF button. Focus once, defocus, focus again etc..

With all that said, what is considered to be "within the DOF" or "within 1/3 DOF" when viewing a 10" print from 12" away is not going to be good enough when pixel peeping at 100%, when the DOF will be very much reduced. In other words, for "normal" photography, where you make prints, or fit to screen, rather than pixel peeping, the AF system should always deliver an acceptably sharp subject. When you pixel peep that may no longer be true.

So, I think you can go mad trying to perfect focus to the mm, when really it is not possible in the real world, especially with a long, slow, consumer zoom. If you can get a 500mm zoom lens accurate to within 1cm at 18' then I think that is the best you can hope for. For "normal" prints, that will be ample accuracy. If you need even higher accuracy then maybe Live View and Live AF will improve results, if your subject remains still enough.

 

[Trev4]

Thanks to everyone for an interesting thread and useful info. I also comment that your second slide show is very good Tim, I enjoyed both shows and it gives a quality of picture to aim for .

After reading this, I am going to re-test my 100-400. The last time I did it, the setting required was -6, but I am still not totally satisfied with the results.
I used a chart that is downloadable and may be of use to you. chart.
I am a little confused by a contradictory suggestion however. On your set up for microadjustment (Tim) you show the tape measure on a slope to indicate the DOF and focus point, this is a similar principle to the Lens Align Pro, and the chart I have liked to above. I accept that you have included a box that is square to aim at though. In the 'Digital journalist' link, it suggests that the target should be set up square to the lens, and indicates that a sloping target should not be used.

"Do not attempt to autofocus on an angled chart, because doing so will degrade the consistency of the camera's focusing measurement. Keep in mind that the camera's AF sensor is comprised of multiple pairs of linear pixel arrays. If you attempt to autofocus on a single line in an angled focusing chart, only a few pixels from each active pixel array will "see" the target. Ideally, the contrast in the reference target should cover the entire area of the camera's center focusing point, and the reference target should be perfectly parallel to the camera's focal plane."

I wonder what your thoughts are on this, do you think a square target the same as Tims would overcome the difficulties sugested in this article.
Cheers, Trev

 

[tdodd]

My target is designed to exactly replicate everything that the LensAlign system achieves, but at zero cost. So long as the focus target - the box in my case - is set fairly square on (IMHO it does not have to be mm perfect, but the closer the better) and is large enough to cover an area at least 3X the size of the chosen focus point in height and width then there should be no problem.

I am aware of the chart you linked to. It probably will be fine. I understand the principles which suggest it should be OK, but when it is so easy to set up a target that is even better suited to the task then why not do that? In my opinion the box and rule is a pretty rock solid solution. There is no possibility of confusion for the focus point, which is the complaint levelled at angled targets, albeit not necessarily with good foundation.

As for getting good alignment of the box, just look through the lens at the box. If you can see any side of the box other than the side facing you then your alignment is off. Even so, unless you are a real klutz your error will probably only be a mm or two. When your DOF is measured in cm and visibly extends over several cm then I really don't think an error of a mm or two is material. Maybe with an 85/1.2 at MFD the error could be an issue. With a 400mm f/5.6 lens at 20m, not a chance.

 

[Trev4]

but when it is so easy to set up a target that is even better suited to the task then why not do that? In my opinion the box and rule is a pretty rock solid solution.

Thanks Tim, thats interesting and makes sense. I'll have a play with your set up, it'll be interesting to see what the difference is between this test and my original one.

 

[tdodd]

If you look at the LensAlign Pro Plus, you will note the inclusion of a 48" rule/scale. That could actually be useful for a 400mm f/5.6 lens calibrated at a greater distance. A single sheet of A4 might make such a calibration quite tough to execute.

By the way, I have dabbled with other variants of calibration targets over the last 15 months, including this one....

It offers an unambiguous "square on" focus target, with alignment marks to make sure it is squared up correctly, and the scale to judge the focus position. The problem is that it is really a bit small for use at 50X focal length as the focus target does not exceed the size of the AF point indicator square sufficiently at such lengths. It good for MFD calibrations though, even those are generally not recommended.

I now like the "box and rule" solution best. I'm sure LensAlign is a great product, but I do not like the price. That's why I pursue the DIY equivalent.

p.s. Here's the setup I used to calibrate my 100-400, out in the back garden. I simply sat the box on a garden table and laid the rule out on the table too. Easy peasy. I know you can see the top of the box, which indicates the box is tilted too much towards me, but I would estimate the resulting error to be about 1cm, and with a DOF of around 20cm it was an error that (a) did not matter; (b) I could make allowance for if I really cared. Note that the active focus point is highlighted and the box is so large that there is no possibility of the camera getting confused and focusing on the wrong thing.

 

[markta]

Very interesting thread..

I'm trying to persuade SWMBO that a 7D is a worthwhile improvement on a 40D..

Trying to come up with a suitable excuse/justification.

Any ideas chaps/esses?

 

[Trev4]

I now like the "box and rule" solution best. I'm sure LensAlign is a great product, but I do not like the price. That's why I pursue the DIY equivalent.

I adapted your tape measure and box idea Tim, its certainly easy enough. I found also that by adjusting the angle of the tape in relation to the box you could get a better idea of the DoF, but a too acute angle made fine adjustments more difficult to decide on, so using much the same as you I checked my 100-400 which is the lens I was unsure of (I also checked my other lenses for reference but they seem fine as they are).
Originally, using the chart I mentioned a few weeks ago, I had to dial in -6, this wasn’t far out as it turned out, but using your method I finally settled at -7. I took many test shots at both settings (in fact I started back at 0 and worked from there) to confirm these adjustments, and feel quietly confident that this is the right setting, time will tell now, just need some decent weather and co-operative birds
Thanks for all the useful info you have given here

 

[tdodd]

"Sigma's woeful reputation..."

Hmmm, I've had no problems.

http://www.talkphotography.co.uk/forums/showthread.php?t=202002