Digital Rebel front focusing

[msvadi]

I tested AF accuracy of my Digital Rebel with 135mm f/2.8 lens. I'm not sure if I did the test right, I used a home made target. Anyway, it appears that there is front focusing.

This is the link to the cropped frame:
http://pantheon.yale.edu/~vm49/fftest/IMG_2108_ss.jpg

I focused the camera on the center line, but one can see a bias in front of it. I would really appreiate your opinions and suggestions

[Haifidelity]

What warranted you to try a ruler test? I'm not trying to sound rude, but where you getting consistent out of focus shots?

-hza

[Tom W]

I tested AF accuracy of my Digital Rebel with 135mm f/2.8 lens. I'm not sure if I did the test right, I used a home made target. Anyway, it appears that there is front focusing.

This is the link to the cropped frame:
http://pantheon.yale.edu/~vm49/fftest/IMG_2108_ss.jpg

I focused the camera on the center line, but one can see a bias in front of it. I would really appreiate your opinions and suggestions

I'd try a similar test, but put some kind of easily-discernable object on the same focus plane as the center line and focus on that instead of the chart. Its difficult to get a lock on that kind of target, given that the focus sensor on the center focus point of the Rebel (and 10D) is a cross pattern and is larger than the actual square in the viewfinder. If the edge of the cross catches something of contrast to focus on, it will lock on that target.

[msvadi]

several reasons. it's not that my pictures were consistently out of focus. but there were quite a lot of pictures that looked too "soft". I thought that the camera shake is too blame. Another thing is that dpreview.com forums are full of reports of front/back focusing with 300D/10D

What warranted you to try a ruler test? I'm not trying to sound rude, but where you getting consistent out of focus shots?

-hza

[robertwgross]

I'd try a similar test, but put some kind of easily-discernable object on the same focus plane as the center line and focus on that instead of the chart. Its difficult to get a lock on that kind of target, given that the focus sensor on the center focus point of the Rebel (and 10D) is a cross pattern and is larger than the actual square in the viewfinder. If the edge of the cross catches something of contrast to focus on, it will lock on that target.

I agree. Use the same ruler, but stick a bright red arrow point at the zero focus point. Just see if that makes any difference.

It can't hurt.

---Bob Gross---

[msvadi]

Tom, Robert : I tried to make a target chart similar to one described here: http://www.photo.net/learn/focustest/ I'm going to do that again as you suggested.

[robertwgross]

First of all, there can be a certain amount of "sloppiness" of focus in just about any autofocus camera. It might be a tiny amount, or it might be a lot. The trick is to study the conditions that lead to a lot of problem, and then be alert when you are into those conditions. For example, low contrast subjects tend to make the worst problem. Low lighting makes it bad also.

Second, when the 10D was new (about one year ago), there was a problem like this in a certain small percentage of the bodies. But then it was corrected, and lot of owners sent in their cameras and lenses to be checked or recalibrated in this respect. But then after a while, that problem died out.

I haven't heard anything about the same problem in a Digital Rebel. Keep in mind that a problem could be in the lens, or it could be in the body, or it could be a fit of the two. Then also it might be user error, but I think you are close to ruling that out.

One last thing. Do you have any filter, like a UV, in front? If so, try removing that. It can't hurt.

---Bob Gross---

[Tom W]

Just for the heck of it, I downloaded Bob Atkins' test and printed it out. I taped it to a box, put it at a 45 degree angle and took a few shots. My 17-40 was spot-on at f/4.0, so I put the 50 on (reputedly not as accurate at focusing as the true USM of the 17-40). I took 4 shots, one at f/3.2 (program mode with flash chose this for me), one at f/2.8, one at f/2.0, and one at f/1.4. All were spot on except the f/2.0 which was forward about 3 bars. I chalk this up to the possibility that I moved since these were all handheld.

I'm certainly satisfied with my results.

[droosan]

I think my 10D front-focuses a few mm with my 28-105 lens. I am pretty sure my old film Rebel did the same, though my EOS-3 is spot-on. I haven't really worried about it too much, since for me focusing is only critical in portraits and I use an EF100/2 (or an EF50/1.4 for couples) for those.

[msvadi]

Thank you guys for your suggestions. Now I wish I have not started at all with the testing. I wasted a half of the day on that just to arrive to conclusion that there is a good chance that it's me and not the camera.

I think that the focus area is actually bigger than the squares we see in the viewfinder. My first target chart was too small, lines were to close to each other and that may have confused the AF.

I made another chart, took another hundred of shots with diferent lenses. The variety of the results was from horrible front focus to almost okay with only minor bias toward the front. Funny, but the pictures taken with a tripod were worse than the handheld shots.

I'll continue tomorrow with better light and another target chart (I cannot stop now).

Never test your camera for anything unless you see persistent problems on the regular shots.

[Tom W]

Well, the thread got me testing as well. I downloaded Bob Atkins' test pattern and fiddled around with it a bit. I got hold of a really useful prop and placed it on the plane with the big black line on Bob's chart, perpendicular to the line from lens to target. The golf tee is the focus point. The chart sits at a 45 degree angle and the focus plane intersects with the focus point that Bob had intended to be used.

To make a long story short, I was wonderfully surprised by the 50 when I took this shot at f/1.4 (I didn't expect it to be this good, frankly):

http://images.fotopic.net/?id=4420551

Click on the image for the 1200X800 size.

This is probably the sharpest f/1.4 picture I've ever taken. Too bad the subject is about useless.....

[Tom W]

I think that the focus area is actually bigger than the squares we see in the viewfinder. My first target chart was too small, lines were to close to each other and that may have confused the AF.


Yes, the focus area is larger than the square. In fact, the center focus area is a cross (+) shape and the legs extend about 1/2 the width of the box beyond the edges of the box. So it is very possible for the focus sensor to capture focus on something that isn't within the box. The other focus points are not the same size either. The ones to each side are a vertical bar (I)-shaped, while the top and bottom points have a horizontal bar (--). Each extends beyond the borders of the indicating box on the focus indicators in the viewfinder.

Somewhere, there's a picture of this, but I don't remember if it was Bob Atkins or Chuck Westfall that put it together.

[Volatile]

OK, I did the test also. I tried my 50 1.8 and my 70-200 f4L. I found the focus varied from right on to 1-2 cm of back focus. I tried different distances from camera to subject, and switched the angle from right-side-back to left-side-back. Always came in from 0-2 cm back, on both lenses.

Another observation: I'm indoors under incandescent lighting, and could not achieve AF (Av mode) unless I put my 380EX on for the AF assist beam. Good to know that at least that works.

I have no idea if these results would warrant servicing of the camera...

One other thing... Check this shot I got at 200mm (ISO 100, f4, 1/3 sec exposure):

http://members.cox.net/erin555/sIMG_2390.jpg

Aside from the target being on a slight angle and underexposed, do you see the vertical blur on the point of sharpest focus? I used my tripod and the ten second timer, could this be camera shake from the shutter itself?

Thanks...

[msvadi]

Tom,

I have to say that this is a fantastic picture ;) Seriously, great results.

Today I spent another couple of hours on testing. (My wife looks at me like if I were going mad. I wonder why... It cannot be the target charts hanging on every wall in our aprtment ;) )

The results varied from bad front focus to almost okay. My mistakes, poor light conditions, distance from the target, properties of the target - all can affect the outcome.

Anyway, here's the result with 50mm 1.8 at 1.8:

http://pantheon.yale.edu/~vm49/fftest/IMG_2204.jpg

And another one:

http://pantheon.yale.edu/~vm49/fftest/IMG_2220.jpg

[Tom W]

...
Aside from the target being on a slight angle and underexposed, do you see the vertical blur on the point of sharpest focus? I used my tripod and the ten second timer, could this be camera shake from the shutter itself?

Thanks...

That could well be from mirror slap putting a tiny shake on the camera. You can use mirror lockup if you have the 10D. Another possibility is that the floor of your house may have shaken a bit (assuming you aren't on a concrete slab). Its surprising how much it moves when people just walk on it.

[Tom W]

Tom,

I have to say that this is a fantastic picture ;) Seriously, great results.

Today I spent another couple of hours on testing. (My wife looks at me like if I were going mad. I wonder why... It cannot be the target charts hanging on every wall in our aprtment ;) )

The results varied from bad front focus to almost okay. User erros (me ;) ), poor light conditions, distance from the target, target properties all can affect the outcome.

Anyway, here's the result with 50mm 1.8 at 1.8:

http://pantheon.yale.edu/~vm49/fftest/IMG_2204.jpg

And another one:

http://pantheon.yale.edu/~vm49/fftest/IMG_2220.jpg

Thanks! I appreciate it, and the lens does as well. Frankly, I was a little surprised at how well it did, considering how I struggle with it sometimes in that very shallow DOF. ;)

The red stripe really pulled that focus into the middle! One of the problems with using the bare chart is that you have quite a few hash-marks that the camera can potentially lock focus on. Here's a link to a shot of the focus sensor arrangement overlaid with the focus indicator boxes:

http://www.pbase.com/image/18920257

The reason I used an external focus point was to avoid the camera locking on to the wrong thing (and this is especially true in low light). That center cross is quite big, and for some reason, the 50's like to grab focus on the wrong subject sometimes. They're accurate, but incorrect if you know what I mean.

Check out Pekka's test setup here:

http://photography-on-the.net/forum/showthread.php?t=10526

[msvadi]

Thanks Tom, very usefull info

[drisley]

...
Aside from the target being on a slight angle and underexposed, do you see the vertical blur on the point of sharpest focus? I used my tripod and the ten second timer, could this be camera shake from the shutter itself?

Thanks...

That could well be from mirror slap putting a tiny shake on the camera. You can use mirror lockup if you have the 10D. Another possibility is that the floor of your house may have shaken a bit (assuming you aren't on a concrete slab). Its surprising how much it moves when people just walk on it.

If you have 300D, you cant lockup the mirror unfortunately. However, here is one tip that will help. If you are taking a picture that is in the timing range in which mirror slap affects the picture the most, turn the camera vertically and take the picture. This will reduce mirror slap almost completely.

[Volatile]

Thanks for the advice. I've been thinking about it more, and the possibility of vibrations coming up from the floor seems most likely. BTW, it's a 300D. No one was walking around, but I did have the height extension cranked way up and I did notice that the camera wobbled quite a bit. It steadied out by the time the timer was up, but that's only what I could perceive through the viewfinder. Perhaps it was moving still... Or else it was this "mirror slap" that you speak of.

Thanks again...

Bill

[Tom W]

Thanks for the advice. I've been thinking about it more, and the possibility of vibrations coming up from the floor seems most likely. BTW, it's a 300D. No one was walking around, but I did have the height extension cranked way up and I did notice that the camera wobbled quite a bit. It steadied out by the time the timer was up, but that's only what I could perceive through the viewfinder. Perhaps it was moving still... Or else it was this "mirror slap" that you speak of.

Thanks again...

Bill

Having the extension cranked up can amplify any small vibration. It doesn't take much movement considering that your target is several feet away.

[Volatile]

OK, re-confirmed it tonight. Everything the same, tripod head extension fully retracted. Much better and mucho gracias.

http://members.cox.net/erin555/sIMG_2393.jpg