Lots of shots out of focus at wide apertures

[mark u]

Please also see Canon's Chuck Westfall's comments here:

 

http://www.digitaljournalist.org/issue0804/tech-tips.html

 

You really do need the right screen to use manual focus with a fast lens.

[mark u]

Some more about focus screens - microscopic images of screen types here:

 

http://www.mir.com.my/rb/photography/companies/canon/fdresources/fdlenses/index1.htm

 

Optical principles that apply to focus screens here:

 

http://doug.kerr.home.att.net/pumpkin/Split_Prism.pdf

[ken_thalheimer]

G Dan Mitchell basically says it all very simply in his post near the top of the thread

[alan_myers]

Hi Peter,

 

I'd suggest forgetting about manual focus. Pretty much all the 1.6X and 1.5X cameras on the market are near impossible to manually focus with much accuracy.

 

If you want to use AF in a similar way, do a search on discussions of Custom Function 4 in all Canon cameras.

 

In addition, most of the time try just using the center focus point (This doesn't mean your subject will always be centered, though, so long as you are also using CFn 4/1 or CFn 4/3).

 

Now, someone suggested a special focus screen, and yes that would help you manual focus. However, I think you are better off learning to use AF well.

 

The other issue is depth of field. All these cameras also tend to exaggerate it in the viewfinder, as mentioned. It's only once you look at an f1.4 or even f2 shot later in on your computer screen that you realize how shallow DOF really was when you were taking the shot.

 

Here again, it's part of the learning curve. If you are coming from full frame/35mm film cameras, you need to get in the habit of stopping down about one extra f-stop, over what it appears to need. This will help with slight focus errors, too.

 

Looking at one of your more OOF images above (the corn flakes box), I also notice a lot of noise in the image. That's from shooting at ISO 1600 with an XT. A lot of noise tends to give an image an impression of less sharpness, too. (Color balance and exposure are way off, too... but that's another issue.)

 

The more in-focus image of the same cereal box appeared less than sharp, too... but more like I'd expect from the 50/1.4 used wide open. Stop down to f2 or a little more and it will sharpen up. It will certainly do so by f2.8.

[peter_viitasaari]

Roger Smith&Philip Kam - The focus test here http://www.photo.net/learn/focustest/ is what got me started on this in the first place. The AF shots seemed to be consistently a little fore-focused which is fine. As long as something is consistent I can deal with it. The manual focused shots however where all over the map...some as bad as 2-3 cm fore-focused to some as bad as 2-3cm back-focused and everything in between. This was at f/1.4 with the scale about 2 ft. from the film plane. I have taken an extensive number of shots using this focus test and can't find any consistency whatsoever in manual focus mode.

 

Mark U&Roger Smith - Thanks for the info on viewfinders and focusing screens.

 

Ken Thalheimer - I am not exactly sure what I am looking for in this test by G Dan Mitchell?

 

For some of you that are wondering why I am so caught up with manual focusing, it's quite simple. An out of focus picture cannot be fixed in the mix, making it the single most important element of taking a picture. With that being said and AF fails, I feel it is important to be confident in the manual focus system to be able to repeat the precise focusing possible with a manual focus film camera.

 

It's possible that I am just experiencing the reason why the Rebel XT is a $400 camera and the 1Ds III is a $7000+ camera.

 

It seems that if I run the camera at f/2.8 and above on AF and f/4.0 and above for MF the results are consistently nominal though definitely not the quality I am used to from film.

 

Thanks again for all the input.

 

Best Regards, Hedghog

[ned1]

Are they out of focus or do they look misty? Misty means that your particular lens doesn't do well wide open. 1.4 is skirting right at the edges of what even a very good lens can do.

 

It can vary from individual lens to lens.

 

If it isn't misty then it's possible that the camera's autofocus is slightly off. Test it by putting the camera on a tripod, taping some newsprint to the wall, putting the camera at an angle so that only one section is in focus, and see if anything has sharp focus.

[peter_viitasaari]

Alan:

 

Thanks for the response...even though some of the other posts aluded to it, your response really drove home the possible shortcomings in lower end DSLR cameras.

 

For the time being I will take your advice and stop down my lens.

 

If I chose to start relying on AF I will definitely keep CF4 in mind...I have some experience with this type of function from using point and shoot cameras.

 

Good call on the ISO1600 noise...I didn't want the flash bothing my testing so I changed my film to an ISO1600 (which I find rather convenient with digital cameras).

 

Thanks again for everyones help.

 

Best Regards, Hedghog

[peter_viitasaari]

More data on manually focused shots.

 

I had a chance to borrow a colleagues Rebel XT and I put the 50 f/1.4 on it which yielded the exact same inconsistent focusing results when manually focusing.

 

Out of 10 photos 2 really bad, 5 so-so and 3 really good.

 

I decided to try my colleagues Rebel XT with the kit lens and shot at 18 f/3.5 which again yielded inconsistent results when manual focusing.

 

Out of 10 photos 0 really bad, 4 so-so and 6 good.

 

So, I am going to conclude that one should not get into the habit of running a Rebel XT at anything less than around f/4.0 and I would suggest f/5.6 just to be on the safe side.

 

One could probably run with slightly larger apertures if one was using AF however, this habit may provide inconsistent results if one needed to switch to manual focusing to expedite a shot.

 

I would like to see a new specification for digital cameras called "Focus Latitude" to indicate how well a digital camera should manually focus...but thats just me.

 

Best Regards, Hedghog

[aaron_lam]

I apologize if I missed it but these tests are not accurate unless you have the camera on a tripod or a table or something to keep the distance to the subject still. You can't depend on hand holding a "test" shot. The shutter speed has nothing to do with it... shutter speed is used for freezing action.

 

It is very possible to change the distance to subject from the time you AF confirm and shutter release... even at 1/500. Ideally, you should use a remote or timer and MLU to remove as many variables as possible.

[PuppyDigs]

An XT's tiny postage screen is pretty much possible to focus manually in a critical manner. You FM has a huge VF and a

ultra high contrast screen with prism split. Big diff. If you really need to use MF consider an aftermarket FS designed for

MF (high contrast screen with prism split). If you're shooting off a tripod you might also consider an eyepiece magnifier.

 

But, yeah, I have a FM3A and am amazed how easy MF is compared to stock DSLRs. Of course the FM won't AF worth a

damn...

[zml]

Ah, the good old times of film...NOT! I did run some focus test with 50/1.4, 35/1.4 and 24/1.4 on two different (years

apart) EOS-1n bodies witht he same results like on DSLRs, i.e. consistent front focus fully open. Except EOS-1n

doesn't offer

micro adjustment. On 1Ds3/1D3, both 24 and 35 are correctable (+9 and +11 respectively) but the 50/1.4 is not

correctable and front-focuses even at the +20 MA setting. Does/did the same on a 30D, 40D, 5D and 1D2n/1Ds2. I

know that, and focus at close distances an inch or so behind the intended plane of focus wide open. Here's an

example: Canon

40D, EF 50/1.4, 1/25s, f/1.4, ISO 3200 handheld. Full frame resized for web. I have focused on the eye/viewfinder and

the

focus plane is on the guy's temple.<br>

<img src="http://d6d2h4gfvy8t8.cloudfront.net/7452272-md.jpg" alt="Canon EF 50/1.4 f/1.4"/>

<br>As you can see the lens is plenty sharp even fully open. You may want to send your sample to Canon for

callibration (mine is going next time I'm sending equipment for service.)

[carlwakefield]

Peter I have not a lot of experiance but from the experiance I have I think, I can only say I have a 50 1.4 and with my 30D 5D and EOS 1v it works great. I found my father complained about a lot of out of focus shots with his 350D and it was that many of his close up shots were just not good in focus. After he upgraded to a 40D he does not have this problem. I have think the autofocus on a 350D is not as good. I have a friend with a 400d and it is much better but still not as good as the 30D or 40D. I think this is one area where you get more for you money when you pay for the more more exspensive cameras. Regards Carl

[glenn_wasserman]

I haven't seen the one answer that I think is the REAL problem.

 

Peter - in all of Canon's AF bodies, the AF sensors are much larger than the markings in the viewfinder. That means

you could be missing focus and not seeing it (due to the view through the viewfinder being a lot like looking through a

drinking straw).

 

If you keep this in mind when you shoot at wide apertures, i.e. make sure you've got the center of the focus point on

the most contrasty thing, and look for it - you may get better results.