20D Focus issues

[smudge]

I'm being told quite a lot, "oohh that looks soft". When I only had my 50mm1.8 I put it down to a dodgy lens. When I bought my 17-40L I still had comments even on a band shoot with studio lights and a tripod. ISO 100 f.8-11, 1/125s.

Has anyone else had anything like this?

It's bugging me because I'm extremely inexperienced and am not sure how crisp the images should be. They could just be as good as you get with this camera.

So what I would like to know is, can anyone suggest a reasonable test I can carry out and post the results for your consideration. I have various lights and tripod. 50mm1.8 and 17-40L
It would also be good if another 20D 17-40L owner or several could do the same test and we could compare results.

Any suggestions/ guinea pigs.

[Bob_A]

What parameters are you using for unsharp mask in Photoshop?

[tim]

I would strongly suggest you do this test (http://www.canon-dslr.com/Canon_Jan05/Canon_SLR_Focus_Test.htm), on a tripod, preferably with mirror lock up and a cable release, otherwise with self timer. Without a tripod the results won't be useful. Do that, the post the results for us to have a look it, it could be something needs calibrating.

Take a look around my site, all images from late 2004 onwards are taken with 300D, from Nov 2004 on they're with the 20D, mostly with the Tamron 28-75.

[smudge]

Bob, I've ben reading several threads regarding this topic and must admit am surprised at those suggesting that it depends on the level of sharpening applied, either in camera or in post.
If my 20D has a focus issue surely the way to test it is without any sharpening in camera or otherwise. The issue is focus not sharpening.

I think that by resorting to over sharpening the image, I would be simply masking the problem, if indeed I have one.

Tim thats the one, I'll carry out that test this weekend.

[Ron Wilson]

Take the camera in to Canon.

My Drebel is still there.
http://photography-on-the.net/forum/showthread.php?p=602922&posted=1#post602922
Same problem.

[griff2]

I briefly had doubts about the 50 f/1.8, however I conducted a simple rough and ready test with the 50, and 17-40 f/4L which proved that things were ok - for me anyway. Rather than using a tripod, I decided to use the 300D's own built-in flash (not the most poweful, but adequate at close-range). I took a picture of a beer can, and a shampoo bottle, the latter positioned a little further back. I focused on top of the beer can, and was more than happy with the results. Here it is with a little USM applied:
http://www.griff2.co.uk/pictures/example.jpg
Taken with the 17-40 f/4L, and after drinking the contents of the can:mrgreen:
As you can see, depth of field is everything, I focussed on the red band at the top: it and the lettering on it are perfectly in focus.

[Bob_A]

Bob, I've ben reading several threads regarding this topic and must admit am surprised at those suggesting that it depends on the level of sharpening applied, either in camera or in post.
If my 20D has a focus issue surely the way to test it is without any sharpening in camera or otherwise. The issue is focus not sharpening.

I think that by resorting to over sharpening the image, I would be simply masking the problem, if indeed I have one.

Tim thats the one, I'll carry out that test this weekend.

No kidding, if you have a focus issue applying usm won't help. It will nicely sharpen the area that is in focus though :lol:

There is a difference between images being in focus but soft and images where the camera is not focusing at the proper point. Lots complain about soft images assuming it is the camera though and don't understand that the 20D may produce soft images when viewed before post processing. You may be clearly seeing that the focus point is being missed though, indicating the camera is at fault. Tim's suggestion will help confirm this.

[KBarno]

smudge,

One of the things that will come with time and experience is the ability to understand what you are seeing through the view finder. What you need to start watching for is where the camera is trying to focus. Be aware of what the af is up to and where it is acutally focusing. When I was just starting out with a 10d I would get pictures that would be crisp and the next frame would be slightly out of focus. After I learned how to set the camera up to only manually use the center point for focusing, I was able to get the camera to focus on the point I was trying to focus on. After much trial and at times tribulations, I was able to watch the camera af points and understand where it was trying to focus. One tip if the camera fails to focus on the proper subject is to aim the camera of center and hold your button half way down to get the af to lock onto the proper point. Realign without releasing and complete the picture.

Keep shooting and the experience will happen.

[DavidEB]

Even with a flash to provide a very short effective exposure time, there's still a need for a tripod. With a hand-held photo you might move a little between the time you achieve focus lock and the time of the actual exposure, especially if you're shooting an empty can of beer. If the DOF is thin, this movement may throw off your test.

also, your focus test should be done in bright light, which means that at least some of the exposure comes from the ambient lighting in addition to the flash, and movement could blur that component of the photo.

tripod - it's not just a good idea, it's the law (or should be)

[ed2day]

The simplest, most straighforward way to check autofocus is to compare it to shots where you carefully focus manually. If you can get sharper shots manually focussing then the camera (or lens?) definitely needs calibration. Nothing wrong with doing rigorous contolled tests, just be careful--really careful--as it's easy to get mislead. If you want to do it right get a focus chart--I prefer the ones with a single black line on which you focus with scales on each side. Some prefer a focus target grid that sits at a 45 deg angle to the chart. Read up. Make sure the target is VERY well lit, and there's nothing in the AF sensor region to confuse it. Use a tripod, mirror lockup, remote shutter release. Take multiple shots (at least 3) as focus will vary. Make sure you are well beyond the minimum focus distance.

But I would listen to the suggestions about sharpening too. Don't look at it as a bandaid, but an essential part of the process. And the people telling you it looks soft are almost certainly comparing it to pictures they've seen where sharpening has been applied, whether in-camera or out.

[griff2]

DavEB wrote: Even with a flash to provide a very short effective exposure time, there's still a need for a tripod. With a hand-held photo you might move a little between the time you achieve focus lock and the time of the actual exposure, especially if you're shooting an empty can of beer. If the DOF is thin, this movement may throw off your test.

Well, a quick flick of the finger would solve that problem I think, and it certainly didn't get in the way of my quick test. However, you've raised an interesting point..

[smudge]

Thanks for all the tips. I'm not quite as inexperienced as some of the posts might suggest. My inexperience relates to different models. i.e I'm not sure how the 20d compares to others as I haven't owned any others.

If you read the link provided by Tim, that is the test I shall be carrying out this weekend(with tripod) I will post results for your opinions. If you have any.

[Bob_A]

Good luck with the test smudge. I hope it's something simple and not your gear ... even when the service is fast and efficient it's never fun waiting for stuff to get repaired.

Let us know how things work out!:D

[griff2]

I decided to do the test that Tim suggested on all my lenses. However, I also decided to make it more real-world, and decided to use flash hand-held, since thats how I use my camera most (hand-held that is).

The 70-200 f/4 L proved flawless, the 17-40 f/4 L was also ok, although there appeared to be a slight predisposition to forward focus, nothing to write home about though. The nifty fifty proved altogether more interesting. Here's the first shot at f/2.8:
http://www.griff2.co.uk/pictures/50_1.8_test1.jpg
It's pretty well spot on, however here's the very next shot, same aperture:
http://www.griff2.co.uk/pictures/50_1.8_test2.jpg
In all cases I aimed for the centre cross. The nifty is the only lens that exhibited this inconsistency. Oh well, it seems the adage "you get what you pay for" has never seemed more true. Time to start saving for the 50 f/1.4.

[mwj]

In all cases I aimed for the centre cross. The nifty is the only lens that exhibited this inconsistency. Oh well, it seems the adage "you get what you pay for" has never seemed more true. Time to start saving for the 50 f/1.4.

My 50/1.4 does the same thing in that focus test. In reality, it does seem to front focus generally, but often gets close enough to get a usable shot. I'm planning to take it in to be calibrated sometime soon.

mwj

[ed2day]

Keep in mind that when using flash, the focus is locked before flash occurs so you are focusing with whatever available light is in the room (unless flash assist is active). So if there's not much light you're testing things other than just the ability to focus normally. Actually the wider apertures like the 50mm should have an advantage.

[I Simonius]

I'm being told quite a lot, "oohh that looks soft". When I only had my 50mm1.8 I put it down to a dodgy lens. When I bought my 17-40L I still had comments even on a band shoot with studio lights and a tripod. ISO 100 f.8-11, 1/125s.

Has anyone else had anything like this?

It's bugging me because I'm extremely inexperienced and am not sure how crisp the images should be. They could just be as good as you get with this camera.

So what I would like to know is, can anyone suggest a reasonable test I can carry out and post the results for your consideration. I have various lights and tripod. 50mm1.8 and 17-40L
It would also be good if another 20D 17-40L owner or several could do the same test and we could compare results.

Any suggestions/ guinea pigs.


What I would do is first of all if you don't fancy doing official tests is take any shot outside in good light so you get a shutter speed of at least 250 at about f5.6 to f8
adjust ISO if needed

- then , in photoshop apply some sharpening - the 20Ds shots DO look soft before sharpening!

[griff2]

ed2day wrote: Keep in mind that when using flash, the focus is locked before flash occurs so you are focusing with whatever available light is in the room (unless flash assist is active). So if there's not much light you're testing things other than just the ability to focus normally.
I decided to test the 50 outside today (Saturday 18th) it was quite bright (for the UK) and @ f/1.8 I got shutter speed in the area of 1/2000 to 1/4000. The lens performed as it did with the flash, sometimes spot on, other times forward focusing to varying degrees. The interesting thing is, the target was always perfectly in focus before I took the picture - occasionally I adjusted the lens in manual to make sure. So, it seems the loss of focus occurs at the instant when the picture is taken; food for thought..

[BearLeeAlive]

I have had fairly soft looking pictures with my 20D which I have had for a month now. At first I only had the EFS 17-85 but have since got the EF 70-200 F4. I set up the focus test and found that in most cases, with both lenses, that it had a back focus centered anywhere from 2- 10 cm behind the target. The actual center of focus was not consistent but the fact it was behind the target was.

A couple of questions;
1. Should I use auto or manual focus. I worry that using my slightly less than perfect eyesight might be a problem in manual focus.
2. With these lenses what settings should I use to make the tests and what distance should I be away from target?
3. I imagine I should do a few different aperatrue settings for the same focal length as it seems with others this can make the difference.

We have a Canon Service Centre here in town. I guess my best bet would be to take the camera in to them. If I do determine the camera needs calibrating, should I also take the lenses in too?

[spun_duck]

I have a theory.....In the last week I have bought and returned three 20D's, due to soft focus issues. And frankly USM N PS is BS. If my camera aint tack sharp, then its not my camera - no excuses. However I may have stumbled onto the solution to the soft image problem.

If your having soft image issues with a 20D please help me with my theory by answering the following two questions?

1) What type of flash card are you using? (brand and speed)

2) What firmware version is your 20D using?

Thanks to all who answer. either we will be amazed or I will look like an
idiot....again.

As for my theory, we'll see how people answer the questions before I state it. I would hate to muddy the water of a highly scientific experiment.:D

[lancea]

I had similar feelings to you for a time after I got my 17-40 lens. It really is a different world with the 20D. I wondered why my little C-50 Zoom took sharper shots. I didn't have the luxury of being able to swap cameras or lenses, which forced me to read lots and experiment till I was happy. I haven't done any formal focus tests.

I strongly recommend you read this document http://photoworkshop.com/canon/EOS_Digital.pdf, especially the bits about sharpness and what settings Canon applies in-camera depending on the market the camera is aimed for. Would you have imagined that they intentionally make the top-of-the-range cameras give the softest photos "out of the camera"? It's true.

Of course it is just possible that you've been very unlucky with your bodies and lenses, but I will be very surprised (and impressed!) if your test results show that a particular brand of flash card is implicated.

[I Simonius]

I have a theory.....In the last week I have bought and returned three 20D's, due to soft focus issues. And frankly USM N PS is BS. If my camera aint tack sharp, then its not my camera - no excuses. However I may have stumbled onto the solution to the soft image problem.

If your having soft image issues with a 20D please help me with my theory by answering the following two questions?

1) What type of flash card are you using? (brand and speed)

2) What firmware version is your 20D using?

Thanks to all who answer. either we will be amazed or I will look like an
idiot....again.

As for my theory, we'll see how people answer the questions before I state it. I would hate to muddy the water of a highly scientific experiment.:D

I use Sandisk extreme 3 1gb cards mostly
Firmware is 1.1.0

[BearLeeAlive]

I did not use a flash in these tests and so far have only used the camera flash for pics.

My firmware is Ver. 2.0.0 which was updated just 2 weeks ago.

[cc10d]

Thanks lancea for the Chuck Westphall article, http://photoworkshop.com/canon/EOS_Digital.pdf,. this really explains a lot of stuff, although written for the 1 series cameras, most of it can be directlly applied to other cameras. I think this should explain most of the soft focus concerns. Depending on our preferences the camera can be set to provide what we want. if we feel as spun_duck, wanting it right out of the camera we can use parameter 1 or even use a custom parameter with more sharpness,contrast, saturation, etc. for pictures set up our way directly from the camera. If we want to post work the pictures for maximum user control, we can shoot RAW. What a camera, a set up for nearly anyone. I especially thought that the chart on pg. 26 was very informative for comparing how different models are set up. And the chart about the relative magnifications we look at on pg. 28 showing a bit of prespective on the levels of enlargment we are using when looking at our pictures, is also something we may overlook in our evaluations. (just how much enlargement is reasonable anyway. reduce the magnification to the levels that we will print at and many of the imperfections become invisible. we seldom expected poster sized prints to be tack sharp from the film based pictures without further processing.) All this, It took me a while to figure this out and am now convinced of the superiority of this equipment, enjoying the pictures it allows me to create, and here a lot of it is in one article. Thanks for all of us.

[smudge]

HHHHmmmmm. Haven't got round to doing the test, as my printer has blown a gasket. But I am greatly comforted by the fact that I appear to not be alone where it comes to focus/sharpness issues.

This part of the above article amazed me "...Avoid recomposing after focus lock during portraiture or close-up photography: This technique can cause focusing errors when shooting subjects within 15 feet of the camera, especially when using large apertures to reduce depth of field. Instead, select an off-center focusing pointor focus manually."

I always focus using the centre AF point, as I assumed it was the most sensitive and am sure I read somewhere that it is the only point that acts as a cross hair. Obviously you then don't want your centre of attention at the centre of the photo, so holding down the shutter you recompose...according to canon this is not correct procedure.

Your thoughts...

I am using SANDISK ULTRA 1GB Firmware 1.1

[griff2]

If your using AI servo then theoretically things should be ok if you re-compose, but in reality, with a narrow depth of field, when you re-compose, focus may snap to an area you don't want or expect. For the 300D I could use AI servo with the hack (for RAW), but I'm not bothered. Instead I'd rather use a specific focus point, which gives me the best composition, whilst retaining the site of interest I want in focus.


I've got my own theories why certain camera lens combinations give erroneous focusing results, and I'm thinking this is purely a mechanical problem..

[smudge]

If your recomposing whilst holding down the shutter(focus lock) focus should not snap anywhere else at all, but rather is locked at the point where you half pressed the shutter.

[lancea]

Thanks lancea for the Chuck Westphall article, http://photoworkshop.com/canon/EOS_Digital.pdf
Thank you. It certainly helped me more than any other single thing. It has been suggested it is made a sticky, but I'm darned if I know how that gets done. Probably only a moderator can do it.

[griff2]

Smudge wrote: If your recomposing whilst holding down the shutter(focus lock) focus should not snap anywhere else at all, but rather is locked at the point where you half pressed the shutter.

In any mode except AI Servo this is correct, but in AI servo, the lens will re-focus as you move, with your finger half depressed on the shutter release.

[smudge]

Now that is interesting...I'll have to check all this new info out. On the downside I've just spent about 8hrs trying to get some decent colour out of my new printer, which I had to buy to carry out the test, which I may now not need to do...My brain aches...just when I figure something out...I haven't.

[spun_duck]

Well. It looks like I was wrong:cry: ... no real connection between soft focus and media cards or firmware. Which ticks me off...I shot with my 20D and it was very soft using a canon 70-200L 2.8 IS USM (60th at F/8)I then put the lens on a friend's 20D shot the same thing on a tripod, same parameters, and it is dead sharp. There are three differences between her 20D and mine.....she was using a slower media card......her firmware is 1.1.....and her camera was made in a different facility. according to the SN#. So I thought maybe one of those may factor into the softness. I am at a loss and about to buy a Fuji S3 if my 4th 20D I ordered is as dissatisfying as the last three.

[Bob_A]

This part of the above article amazed me "...Avoid recomposing after focus lock during portraiture or close-up photography: This technique can cause focusing errors when shooting subjects within 15 feet of the camera, especially when using large apertures to reduce depth of field. Instead, select an off-center focusing pointor focus manually."

I always focus using the centre AF point, as I assumed it was the most sensitive and am sure I read somewhere that it is the only point that acts as a cross hair. Obviously you then don't want your centre of attention at the centre of the photo, so holding down the shutter you recompose...according to canon this is not correct procedure.



Read this article called "Why Focus-Recompose Sucks"

http://visual-vacations.com/Photography/focus-recompose_sucks.htm

[Bob_A]

Another way you can get out of focus shots is by swaying a bit forward or backward after gaining focus but before releasing the shutter. This is another reason for using a tripod.

[smudge]

Yeah the swaying thing was the first thing I put a stop to. The recomposing could be the problem...I'll know for sure by tomorrow. Thanks for the links and articles :)

[griff2]

Read this article called "Why Focus-Recompose Sucks"

http://visual-vacations.com/Photogr...mpose_sucks.htm

What this article is saying is that the focus plane follows the length of an an arc which is valid. However, unless there is a significant distance between the area of interest and the the focus point the difference will be minimal. E.g a head and shoulders shot, with the centre focus point being an eye, re-composed, would not move significantly enough along the arc to cause any focus problems. Having said that, I wouldn't use the centre point for such a shot, but rather the point nearest the subject's eye.

[I Simonius]

Well. It looks like I was wrong:cry: ... no real connection between soft focus and media cards or firmware. Which ticks me off...I shot with my 20D and it was very soft using a canon 70-200L 2.8 IS USM (60th at F/8)I then put the lens on a friend's 20D shot the same thing on a tripod, same parameters, and it is dead sharp. There are three differences between her 20D and mine.....she was using a slower media card......her firmware is 1.1.....and her camera was made in a different facility. according to the SN#. So I thought maybe one of those may factor into the softness. I am at a loss and about to buy a Fuji S3 if my 4th 20D I ordered is as dissatisfying as the last three.

60tn second is too slow to relaibly handhold, yuou weill get away with it with IS but it's still Slooooooooow

[jfrancho]

Well. It looks like I was wrong:cry: ... no real connection between soft focus and media cards or firmware. Which ticks me off...I shot with my 20D and it was very soft using a canon 70-200L 2.8 IS USM (60th at F/8)I then put the lens on a friend's 20D shot the same thing on a tripod, same parameters, and it is dead sharp. There are three differences between her 20D and mine.....she was using a slower media card......her firmware is 1.1.....and her camera was made in a different facility. according to the SN#. So I thought maybe one of those may factor into the softness. I am at a loss and about to buy a Fuji S3 if my 4th 20D I ordered is as dissatisfying as the last three.I'm getting the feeling you may never find a satisfying camera.

[jarod]

I found I've had a FAR higher percentage of sharp images when I use manual focus. Especially when shooting people.

[shim]

sorry to hear about your problems. I am on my third 20D. This one I sent to Canon, it helped. I'm going back to my Rebel and Elan.

[rmpcanon]

Aarrgh! I can't wait.

Obviously, you feel that there are qualitative differences in either the type of flash card or the firmware version used -- please elaborate; don't keep us waiting.

[transcend]

Well. It looks like I was wrong:cry: ... no real connection between soft focus and media cards or firmware. Which ticks me off...I shot with my 20D and it was very soft using a canon 70-200L 2.8 IS USM (60th at F/8)I then put the lens on a friend's 20D shot the same thing on a tripod, same parameters, and it is dead sharp. There are three differences between her 20D and mine.....she was using a slower media card......her firmware is 1.1.....and her camera was made in a different facility. according to the SN#. So I thought maybe one of those may factor into the softness. I am at a loss and about to buy a Fuji S3 if my 4th 20D I ordered is as dissatisfying as the last three.


There is absolutely no way that media itself will alter the image. That is like saying that your thesis content changed when you moved it from one hard drive to another.

Digital pro bodies are supposed to be generally softer so that the photographer can adjust this in post processing. If you want the photos tack sharp, shoot in jpeg mode and up the sharpness and contrast settings.

If the camera is truly out of focus (front/back focus) then I agree, you have other problems.

[djtowle]

I found I've had a FAR higher percentage of sharp images when I use manual focus. Especially when shooting people.

I never had so many out of focus pictures until I started using an auto focus camera, LOL.

[rmpcanon]

If your having soft image issues with a 20D please help me with my theory by answering the following two questions?

1) What type of flash card are you using? (brand and speed)
Five days have passed; no responses. Can't wait any more -- what is your theory re: the type of flash card?

[booggerg]

Why hasn't any one provided an example of such soft focus??

and NO the media card has NOTHING to do with your lens performance...

[Pekka]

Read http://photography-on-the.net/forum/showthread.php?t=10526&highlight=focus+issue+demystified and replace every occurrence of "10D" with "20D".