Focus Test Question

Something I've been wondering for a while, when people do focus test with a chart setting on a 45° angle, shouldn't it always actually front focus? If the chart is at an angle and the center focus point is on a particular line or point, won't the bottom of the actual focus area be where it focuses? How big is the focus area, larger than the square in the viewfinder isn't it? Since the chart is on angle, and the focus area is not a precise point...............? Make any sense?

Just another reason not to do these tests. Real life trials are the best way and I've yet to get a bad one.

How many focus points do you have active? If you have them all on then, yes, it will focus at the bottom of the image. But, any pix of rulers and focus charts are done with only one active focus point - the center one. My suggestion: bypass the charts and take real pix of the kinds of things you really do.

I agree, I'm not doing the test, just wonder about it every time I read of someone complaining after doing one. Using just one point I think technically it will still front focus a small amount. The one focus point is not a pin point, it is larger than the square you see in the viewfinder, so it should focus at the closest point to the camera, which would not be dead center of the one focus point.

Eagle wrote in post #5454717
I agree, I'm not doing the test, just wonder about it every time I read of someone complaining after doing one. Using just one point I think technically it will still front focus a small amount. The one focus point is not a pin point, it is larger than the square you see in the viewfinder, so it should focus at the closest point to the camera, which would not be dead center of the one focus point.

Well, the test chart has a bar with white area around it, which is the only area with any contrast right in the middle where the center point focuses, so getting correct focus should not be a problem.

However, I do agree with trying out a lens reall life first, and only if you find a problem, start testing in detail with focus charts, brick walls, rows of books or batteries, etc., to determine what exactly the problem is. And you may find that the problem proves to be a user problem as well, not necessarily a lens problem .

Kind regards, Wim

wimg wrote in post #5454739
Well, the test chart has a bar with white area around it, which is the only area with any contrast right in the middle where the center point focuses, so getting correct focus should not be a problem.

True, I was not thinking of the contrast that was needed for focus.

The bigger problem with the ol focus chart is that you are filling the frame with an 8x10 sheet of paper. That is much closer focus than most lenses (other than macros) will be used 99% of the time.

I can also tell you for sure that AF accuracy issues do not manifest at all distances equally. Shooting AF charts very close to the MFD is not a good test unless that is how you plan to use the lens most of the time.

When I posted this suggested method on another thread, I based it on this information http://www.bobatkins.c​om …hnical/focus_te​sting.html

If there is a better technique, please share. But this made sense to me.

A lot of the people that say they did a focus test and it is front/back focusing it is off by less than 2mm (.078"). Unless your shooting 1:1 macro does it really matter? I don't think it would ever show up in real life. Damn maybe that's why that jet going 600 mph, going from my left to right, from 1/2 mile out to directly above, with a heavy overcast sky, was OOF................

Frankly, I've wondered about all this front- or rear-focusing a bit, and I could see where it would be easy to get front or rear focus in a test based on one simple thing - depth of field. I mean, given that the autofocus area is actually bigger than the autofocus point shown in the viewfinder, and people are using a slanted chart, how accurate can auto-focus be, and is "failure" almost insured given the test conditions of focusing close on some chart and utilizing a wide aperture in an attempt to really find the plane of focus?

Looking at the three lenses I own, here's the calculated depth of field (from an online program) for a Canon20D at each lenses closest focus:

Tokina 12-24 at f4 @ 11.8 inches = 3.67" (12mm) to 0.86 inches (24mm)

Tokina 28-80 at f2.8 @ 19.2" = 1.21" (28mm) to 0.13" (80mm)

Tokina 80-200 at f2.8 @ 70.8" = 2.04" (80mm) to 0.3" (200mm)


Can autofocus be accurate enough to focus dead-on when depth of feld is only 3 tenths of an inch or less (or even less than an inch)? I wonder ...

Exactly what I was saying or trying to.

Love your sig AutoXer

Thanks Eagle - I'm pretty sure it was some of my own dumb questions that had me add that to the sig!

It was your response to the question that caused me to go run the DOF calculations and show how little there was at the close focus of many lenses. I don't know if Canon ever published a spec of how accurate its autofocus is, but if its best capability is only within a couple of inches then it would be easy for it to miss focus.

Basically, the focus test chart is a crude instrument. The instructions clearly state that as long as the center bar is in focus, your lens/camera should be considered to be focusing properly. Unless you're using a pro level camera, Canon only guarantees that the target you focused on will be somewhere within the DOF so slight variations in focusing should be ignored as long as the center focusing area is sharp.

Won't argue the fact that the focus chart is a crude instrument, but again I ask, what else do we have?

Below is from a series with various lenses and two bodies,
top image is a 70-200 f/2.8 on a Ds MkII and bottom is same lens mounted on a 5D.
Tripod, MLU, time delay, f/2.8, and everything. apples/apples

In my mind the Ds image is good, while the 5D image is front focusing.

What say you'll?

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

chauncey wrote in post #5457451
but again I ask, what else do we have?

Line up some batteries, beer cans, people, anything, spaced a little bit apart sideways and front to back. Real world stuff. How many things do you take pictures of 2mm(.078") apart sitting on a 45° angle?

chauncey wrote in post #5457451
, while the 5D image is front focusing.

What say you'll?

How much of a crop are those, please post the full size pic. Is it front focusing or is the bottom of the focus area where it is in focus, closest to the camera? Does the 5D have a bigger focus area around the single focus point box than ths Ds?