Possible Back focus?

I've moved to using the 24-105 as my primary lens for portrait work. I've noticed with well lit subjects eye lashes and other details are not as criticaly focused as I expect.

Now before we go crazy with comparisons to other apetures and lenses and bodies, I've done this test on several lenses along with an XTI, 30D and 20D. The 20D is the only body that looks criticaly focused while all other bodies I have seem to back focus slightly.

A friend of mine has a 1D-N and it locks perfectly on every lens we tried. The math seems different for AF compared to lower end bodies. Is this by design? Softer skin? I expect canon to let me deal with this in PS vs the body/lens.

Does this image look slightly out of focus? Has anyone gone through this and sent your body/lens in for calibration? Did canon improve AF for your lens?

125/f4/ISO100

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

Did you take this image at a 90 degree angle? It looks like the back 2mm line is sharper than the front 2mm line, but if this wasnt taken at a 90 degree angle it might explain that.

xStan wrote in post #4609393
Did you take this image at a 90 degree angle? It looks like the back 2mm line is sharper than the front 2mm line, but if this wasnt taken at a 90 degree angle it might explain that.

Requires a 45 degree angle for the shot.

http://www.focustestch​art.com/chart.html

I think xStan means you have to shoot perpendicular to the chart, rather than referring to the vertical angle of your camera. Judging from the vertical lines in your sample it doesn't look like you shot the chart straight on. This may skew your results somewhat.

Yeah it looks the tiniest bit back-focused. Can you post one from the 1D IIn for comparison's sake?

looks a little back focused to me.
i WISH my 70-200 only backfocused that much

https://photography-on-the.net/forum/showthre​ad.php?t=429874

tagvestibule wrote in post #4609678
I think xStan means you have to shoot perpendicular to the chart, rather than referring to the vertical angle of your camera. Judging from the vertical lines in your sample it doesn't look like you shot the chart straight on. This may skew your results somewhat.

It may well be ok, because the lines will converge towards the background, and the top part of the chart will be further away than the foreground.

However, the horizontals should be just that, parallel to the top and bottom of the frame.

Kind regards, Wim

tagvestibule wrote in post #4609678
I think xStan means you have to shoot perpendicular to the chart, rather than referring to the vertical angle of your camera. Judging from the vertical lines in your sample it doesn't look like you shot the chart straight on. This may skew your results somewhat.

The article indicates this test should be at a 45 degree angle and has an entire section dedicated to explaining why.

The same test on the 1DN proved to have a much sharper focus .. Bottom line. I bought a 5D and shoot L glass .. I expect the image to be as sharp as my eyes are when looking at the paper.

I'm sending it in. Will let you know what Canon says about it.

MIME changed to 'text/html' | Content warning: script

Looking at this pic, the set-up seems ok, but for the left hand side of the test chart, which seems a little elevated relative to the right hand side.

Other than that, DoF distribution for a relatively short distance looks perfectly ok.

Don't confuse relative sharpness with DoF. A 1D will look sharper due to lower pixel density. Furthermore, even if the 20D seems critically sharper, it looks within range to me. DoF stretches from about 4 mm before to about 8 mm after, which is near perfect. If anything, it should be slightly less perfect.

My advice first of all, is to apply a little USM in PhotoShop to sharpen it up a little.

Secondly, try other apertures as well, to see if sharpness increases. If it does a lot, the lens may need calibration, but not of the AF-kind. The 24-105L can be a little soft at 105 mm full aperture, but should not be excessively so. Therefore test at F/5.6 and F/8 as well.

HTH, kimd regards, Wim

Yeah I'd send it in if you're not happy with your 5D and L glass.

They're not referring to THAT angle, they're referring to the fact that the VERTICAL lines in the test chart image you posted are not VERTICAL. That has nothing to do with the second picture you posted showing what a 45 degree angle looks like. What they're trying to tell you is that you need to twist the paper a smidgen

tdaugharty wrote in post #4610386
The article indicates this test should be at a 45 degree angle and has an entire section dedicated to explaining why.

The same test on the 1DN proved to have a much sharper focus .. Bottom line. I bought a 5D and shoot L glass .. I expect the image to be as sharp as my eyes are when looking at the paper.

I'm sending it in. Will let you know what Canon says about it.

HTTP response: NOT FOUND | MIME changed to 'text/html' | Content warning: script

scot079 wrote in post #4610468
Yeah I'd send it in if you're not happy with your 5D and L glass.

They're not referring to THAT angle, they're referring to the fact that the VERTICAL lines in the test chart image you posted are not VERTICAL. That has nothing to do with the second picture you posted showing what a 45 degree angle looks like. What they're trying to tell you is that you need to twist the paper a smidgen

No, the vertical lines (bottom to top) will have a vanishing point away from the camera (top of image), and will therefore converge towards the top, even when aligned properly. Note that this is a crop, not the full image.

The horizontals (left to right) need to be parallel to the top and bottom of the viewfinder/image, that's all, and they appear to be.

Kind regards, Wim

Understand Wim, OP is confused about the angle that earlier posters are speaking of. Thanks!

wimg wrote in post #4610537
No, the vertical lines (bottom to top) will have a vanishing point away from the camera (top of image), and will therefore converge towards the top, even when aligned properly. Note that this is a crop, not the full image.

The horizontals (left to right) need to be parallel to the top and bottom of the viewfinder/image, that's all, and they appear to be.

Kind regards, Wim

scot079 wrote in post #4610468
Yeah I'd send it in if you're not happy with your 5D and L glass.

They're not referring to THAT angle, they're referring to the fact that the VERTICAL lines in the test chart image you posted are not VERTICAL. That has nothing to do with the second picture you posted showing what a 45 degree angle looks like. What they're trying to tell you is that you need to twist the paper a smidgen

Exactly - I meant this angle (in green):

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