Canon 50mm f1.4 autofocus advice
Discussion
Evening all,
Hoping for a bit of advice really as i'm getting a bit frustrated with my new Canon 50mm f1.4 lens. I cannot seem to get the autofocus to work very well, as the focus always seems to be in the foreground of what i've tried to focus on.
I've changed the focus to be single point in the centre, however in the example below, even pointing at various bits of the car, when i've got home i've realised the focus on the car is slightly out.
Any tips/advice? I want to keep the aperture at f/1.4 to keep a shallow dof, but just can't seem to get the focusing right. I know a backup plan is to manually focus but i'm hoping it should be possible to get the autofocus to work better than it is currently?
Hoping for a bit of advice really as i'm getting a bit frustrated with my new Canon 50mm f1.4 lens. I cannot seem to get the autofocus to work very well, as the focus always seems to be in the foreground of what i've tried to focus on.
I've changed the focus to be single point in the centre, however in the example below, even pointing at various bits of the car, when i've got home i've realised the focus on the car is slightly out.
Any tips/advice? I want to keep the aperture at f/1.4 to keep a shallow dof, but just can't seem to get the focusing right. I know a backup plan is to manually focus but i'm hoping it should be possible to get the autofocus to work better than it is currently?
I'm guessing that you are about 10 mtrs from the car in which case the DoF would be about a half car length with apporx 1/3 of that ahead of the focus point and 2/3 behind. Roughly, as a rule of thumb.
So to get the whole car in focus would be trick but using a string horizontal line of, say, the wipers might get a decent result. Slightly better at f1.8 I would guess.
However looking at the largest image on Flickr suggest that the mosty in-focus section it actually in front of the car with the number plate just slipping out of the back of the acceptable focus zone. That's a little odd to say the least and even micro-adjust might be pushed to deal with that. Not worth thinking about anyway since I don't think the 550D has microadjust?
Maybe the first approach would be to compare the AF results with what you can get from Live View using Manual Focus on the same subject (10x enlargement should help) and then use the results of that comparison to decide what to to next.
So to get the whole car in focus would be trick but using a string horizontal line of, say, the wipers might get a decent result. Slightly better at f1.8 I would guess.
However looking at the largest image on Flickr suggest that the mosty in-focus section it actually in front of the car with the number plate just slipping out of the back of the acceptable focus zone. That's a little odd to say the least and even micro-adjust might be pushed to deal with that. Not worth thinking about anyway since I don't think the 550D has microadjust?
Maybe the first approach would be to compare the AF results with what you can get from Live View using Manual Focus on the same subject (10x enlargement should help) and then use the results of that comparison to decide what to to next.
Thanks guys. Looks like the first step is to put it back on the tripod and test vs live view whilst zoomed in. Will try a few other scenes/distances as well as the usual car stuff.
I've had the lens about 6 months, bought from Digiplus via Amazon so hopefully I can at least get a reasonable response if it does look like it's the lens itself.
I've had the lens about 6 months, bought from Digiplus via Amazon so hopefully I can at least get a reasonable response if it does look like it's the lens itself.
Kermit power said:
Is it just me, or would anyone else have expected the far end of the track to be much more out of focus than it is at f1.4?
That's exactly what I was thinking. DoF will be small, and the back of the track is a long way back, so has more clarity than I would expect. Oli.
zcacogp said:
Kermit power said:
Is it just me, or would anyone else have expected the far end of the track to be much more out of focus than it is at f1.4?
That's exactly what I was thinking. DoF will be small, and the back of the track is a long way back, so has more clarity than I would expect. Oli.
Had me wondering if the aperture blades were stuck a bit narrower than f1.4 or something.
Kermit power said:
zcacogp said:
Kermit power said:
Is it just me, or would anyone else have expected the far end of the track to be much more out of focus than it is at f1.4?
That's exactly what I was thinking. DoF will be small, and the back of the track is a long way back, so has more clarity than I would expect. Oli.
Had me wondering if the aperture blades were stuck a bit narrower than f1.4 or something.
Same lens at f/1.8 but a close up shot.
However on the car shots it does seem a fair bit sharper than i'd expect, even with the longer distance to the focal point....so maybe some testing vs other lens' is required.
I had also thought along the lines of Justin's comment that it might be the car itself (especially with the pearl paint).
I'll do some "proper" testing on the weekend with the car, and something non car related and compare to the live view. Obviously i'm aware that it's no good having the kit if you don't try to learn how to use it, but this one had me stumped when i've been trying to take a relatively simple shot (tripod.. focus on the subject, then snap!)
The background is not sharp but neither it is enlarged, as it would be with a telephoto for example, so what you see is just out of focus small leaves (in this case) rather than out of focus small leaves that have been enlarged and probably quite changed from their original appearance. No matter how blurred thay actually are our brains willl still interret them as small leaves on a tree or two.
The paint and the curves of the car will not exactly help the focus system. The ideal test, so far as the car is concerned, is probably to focus on the number plate. For best performance the focus points like good light, strongly contrasting lines in the target area and, for preference, horizontal lines (in landscape mode.) I assume the 550D central focus point is a cross type so vertical or horizontal lines should be OK for that point but the rest of the focus points could struggle.
The plate of the wiper arms - if the spot target can be made to rely just on them and not the sweep of the bonnet too - may be good options. For the 1/3rd to 2/3rds split the wipers are a bit far back really but there doesn't seem to be much else readily availble. Don't think the front wheel would work.
You could cheat and take an assistant. Get them to be the focus point then move out of the picture before taking the shots. Or use another 'ideal' object at the perfect focus point and remove it once you have focus or clone it out afterwards.
However the example seems to have focus but some way in front of the car. That's odd and to discover whether such a phenomenon is consistent really doesn't require a highly specific focus test - in the sample image the discrepancy looks like metres rather than centimetres!
Canon's DPP software can show you where the focus point(s) for the shot is(are) in the frame. Might be interesting to see that although I'm not sure how it deals with a 'focus and re-compose' scenario - intentional or otherwise.
Does the lens allow manual over ride of AF when in AF mode? Id so it there any chance that the way you are holding the lens is interfering with AF operations or overriding them? It's the sort of thing that can easily happen without noticing.
All very intriguing - as odd lens problems often are.
The paint and the curves of the car will not exactly help the focus system. The ideal test, so far as the car is concerned, is probably to focus on the number plate. For best performance the focus points like good light, strongly contrasting lines in the target area and, for preference, horizontal lines (in landscape mode.) I assume the 550D central focus point is a cross type so vertical or horizontal lines should be OK for that point but the rest of the focus points could struggle.
The plate of the wiper arms - if the spot target can be made to rely just on them and not the sweep of the bonnet too - may be good options. For the 1/3rd to 2/3rds split the wipers are a bit far back really but there doesn't seem to be much else readily availble. Don't think the front wheel would work.
You could cheat and take an assistant. Get them to be the focus point then move out of the picture before taking the shots. Or use another 'ideal' object at the perfect focus point and remove it once you have focus or clone it out afterwards.
However the example seems to have focus but some way in front of the car. That's odd and to discover whether such a phenomenon is consistent really doesn't require a highly specific focus test - in the sample image the discrepancy looks like metres rather than centimetres!
Canon's DPP software can show you where the focus point(s) for the shot is(are) in the frame. Might be interesting to see that although I'm not sure how it deals with a 'focus and re-compose' scenario - intentional or otherwise.
Does the lens allow manual over ride of AF when in AF mode? Id so it there any chance that the way you are holding the lens is interfering with AF operations or overriding them? It's the sort of thing that can easily happen without noticing.
All very intriguing - as odd lens problems often are.
Also remember the focus 'point' typically is a vertical line that extends a fair bit beyond the square you see in the viewfinder and often will grab contrast at one end or the other , and WILL focus on the closest item under the whole focus point.
If you have a camera with 'spot' focus point mode that is much more precise.
If you have a camera with 'spot' focus point mode that is much more precise.
A-Dep mode? is that still on canons? useless thing, was better in the film days where it would work out from two focus points.
DOF preview works OK if your stopping down but remember the typical viewfinder will show f4 or greater DOF at a maximum, useless for judging f1.4!
Live view is very useful for this for static subjects and IMO the way to go, especially with magic lantern installed and focus peeking on.
DOF preview works OK if your stopping down but remember the typical viewfinder will show f4 or greater DOF at a maximum, useless for judging f1.4!
Live view is very useful for this for static subjects and IMO the way to go, especially with magic lantern installed and focus peeking on.
RobDickinson said:
DOF preview works OK if your stopping down but remember the typical viewfinder will show f4 or greater DOF at a maximum, useless for judging f1.4!
I never knew that! A quick test on my cameras confirms it to be the case, which explains why I sometimes get shots which I'm expecting to have the DOF I'm after which end up having a shallower one. 
Why on earth doesn't the viewfinder accurately represent DOF?
uzzled:Gassing Station | Photography & Video | Top of Page | What's New | My Stuff