p.3 #1 · Risk of focus shift issues in new 1.4/35?
Great shots Rödluvan, particularly #2 & #3. That's really the sweet spot of the 85 Planar. As for sharpness, it's difficult to say on the latter two as they have been very oversharpened in post but the first one is fairly weak.
Ok, on the original topic - focus shift. I have performed some tests now and confirmed that it was as I thought. My experiment was conducted like this: I placed an ISO focusing chart on a flat surface and pointed the camera at it at a 45 degree angle. I used the 85 Planar for this. First I set the aperture to f/1.4 and focused using the viewfinder. Finding the focus point is relatively easy. I took a shot and then decreased the aperture to f/2.8 and took another shot. Then I turned on live view and focused on the same spot, took a shot at f/1.4, stopped down and took a second shot. I did not refocus between the f/1.4 and f/2.8 shots in both cases.
As you can see, the viewfinder focus was dead on where the focus should be at f/1.4 while it was off at f/2.8. It could of course not be any other way as you are looking through the lens and it's the lens that sets the focal plane and that suffers from focus shift.
You can try this yourself. Take a lens that you know suffers from focus shift. Take a target such as an ISO chart or a news paper or anything where you will see a focus shift. Set the aperture to f/2.8 (electronic, not mechanical, you don't want to actually stop it down!). Focus on any point of your choice on the target, at an angle (45 degrees or so). Press the DOF preview button and see the focus point jump.
p.3 #2 · Risk of focus shift issues in new 1.4/35?
i thought the precision matte focus screens for use with fast lenses showed a minimum dof closer to like f/1.8 than f/2.8 and the stock screens showed something around f/3?
p.3 #3 · Risk of focus shift issues in new 1.4/35?
denoir wrote:
It could of course not be any other way as you are looking through the lens and it's the lens that sets the focal plane and that suffers from focus shift.
Either your screen is very matte, or it is calibrated very well for that particular lens (miscalibrated for everything else). The screen doesn't show f/1.4, that's for sure.
Do you have any other f/1.4 lens that doesn't suffer from focus shift? You'll probably fint that it will back focus in that case.
You can try this yourself.
I've done it with my 50/1.4 and it gives correct focus at ~f/2.5, just as it should in theory. It front focuses at f/1.4 and back focuses at f/4.
Regarding AF, the AF module doesn't "see" the f/1.4 aperture either. Most cameras use angles corresponding to f/5.6 (otherwise slow lenses wouldn't work), but some have multiple sensors that also covers angles corresponding to f/2.8.
A split prism works the same way. You can see at what aperture it blacks out, and it's that angle you focus with, thus giving correct focus at that aperture.
Edit: I don't know how Canon cameras work, but my D700 stops down the lens when using live view, if there's too much light (most of the time). It doesn't matter that I've chosen f/1.4, so in that case LV doesn't help.
p.3 #4 · Risk of focus shift issues in new 1.4/35?
sebboh wrote:
i thought the precision matte focus screens for use with fast lenses showed a minimum dof closer to like f/1.8 than f/2.8 and the stock screens showed something around f/3?
The Super-Precision mattes are optimized for f/2.8 and below. What equivalent DOF they show I'm not sure. AFAIK the stock screen shows DOF equivalent to about f/4 or even f/5.6.
p.3 #5 · Risk of focus shift issues in new 1.4/35?
Makten wrote:
Either your screen is very matte, or it is calibrated very well for that particular lens (miscalibrated for everything else). The screen doesn't show f/1.4, that's for sure.
Do you have any other f/1.4 lens that doesn't suffer from focus shift? You'll probably find that it will back focus in that case.
I'd estimate that the precision matte screen gives me somewhere between f/1.4 and f/2. Difficult to tell exactly but better than f/2.8.
As for my other lenses, they are dead on wide open except for the 135L which back focuses slightly but as it's an AF lens, I've applied a micro focus adjustment. I've checked all of them, including my 58mm f/1.2 wide open, with a precision laser rangefinder.
I've done it with my 50/1.4 and it gives correct focus at ~f/2.5, just as it should in theory. It front focuses at f/1.4 and back focuses at f/4.
Sorry, then your 50 Planar needs adjustments. It should not give correct focus at f/2.5 - that means that it is off.
Look this is very elementary physics. The lens and only the lens decides the focus plane. I don't know in how many different ways I can say this. If you are looking at the projected image through a viewfinder, with live view or not at all does not change the focal plane.
Come on Martin, think about it a minute and you'll realize that what you are saying is physically impossible. The focus shift is a lens attribute. How on earth would the focusing screen know where and how the focus would shift without the lens being stopped down?
Makten wrote:
Edit: I don't know how Canon cameras work, but my D700 stops down the lens when using live view, if there's too much light (most of the time). It doesn't matter that I've chosen f/1.4, so in that case LV doesn't help.
Live view on Canon keeps the aperture you set it to regardless of brightness. By default this aperture is wide open, but you can set it to automatically stop down or simply press the DOF-preview button.
Mar 12, 2011 at 12:55 PM
Steve Spencer Offline • • • • • • • Upload & Sell: On
p.3 #6 · Risk of focus shift issues in new 1.4/35?
I agree with Luka, the precision matte screen on my Canon cameras gives somewhere between f/1.4 and f/2.
p.3 #7 · Risk of focus shift issues in new 1.4/35?
denoir wrote:
How on earth would the focusing screen know where and how the focus would shift without the lens being stopped down?
The size of the microlenses and the steepness of the "parabola" of the focusing screen (I guess it acts as a lens somehow) determine the maximum angle of incident rays that are displayed on the screen. Stock screens just don't display very oblique rays (of wide blur discs) in the place where they would be on the sensor, leading to an image with deeper DOF than the exposure would have.
Unfortunately I have no experience with lenses faster than f/1.8 and my 5D's VF is out of calibration (causing frontfocus).
p.3 #9 · Risk of focus shift issues in new 1.4/35?
AhamB wrote:
The size of the microlenses and the steepness of the "parabola" of the focusing screen (I guess it acts as a lens somehow) determine the maximum angle of incident rays that are displayed on the screen. Stock screens just don't display very oblique rays (of wide blur discs) in the place where they would be on the sensor, leading to an image with deeper DOF than the exposure would have.
Unfortunately I have no experience with lenses faster than f/1.8 and my 5D's VF is out of calibration (causing frontfocus).
p.3 #10 · Risk of focus shift issues in new 1.4/35?
Makten wrote:
This can't be correct.
I figure if you focus, using live view, thus knowing what you are doing, with the lens wide open you should get an image that is as sharp as expected where you focused it. Then if you without touching the focusing ring take a series of images, each stopped down one more stop than the image before, you'll be able to follow how the focus plane shifts.
If you get a focus shift the way you describe using the OVF all it means is that your focus screen is in need of adjustment.
There is no theory saying the correct focus is achieved anywhere else but wide open. Wide open is the preferred way to focus; to get a bright image and because that's how the AF function does it. That's what's count. Then witha lens having no focus shift everything works fine. With focus shift we may run into certain problems (at some distances and some aperture values - like with the Canon EF50/1.2L).
p.3 #12 · Risk of focus shift issues in new 1.4/35?
I think it's the other way around - the AF sensor can see no more than f/5.6. Canon cameras can't for instance use AF on f/8 lenses and and the AF is barely usable for f/5.6 lenses. AF focusing f/1.2 lenses with perfect accuracy (such as the 50L) is on the other hand no problem as long as you stay wide open. Stop down a bit on the 50L and AF starts to miss due to focus shift (as it focuses wide open and stops down just before taking the shot).
p.3 #13 · Risk of focus shift issues in new 1.4/35?
it strikes me that it could be possible that you see a shifted image through the viewfinder if you can only see a f/2.8 dof. let me know if there is an error in my thinking here: think about spherical aberration from the perspective of stopping a lens down from f/2 to f/1.4. at f/2 the rays converge to focus at specific point. when you open the lens up to f/1.4 those rays don't change their point of convergence instead a group of additional light rays are also allowed through that converge on a different point. spherical aberration means that the rays in the f/1.4 ring are in focus at a different point than the rays in the f/2 ring (and so on). what causes the classic glow is that not all the light going through the lens is focused at the same distance. if none of the additional light that comes through lens from f/2 to f/1.4 is actually visible in the viewfinder due to the reflective properties of the focus screen doesn't that mean we are seeing only the same light rays we see at f/2 and thus we can't see the correct focus at f/1.4?
it seems to me the differences luka and makten are seeing may come down to differences in focus screens. makten - what kind of screen do you have in your d700?
p.3 #14 · Risk of focus shift issues in new 1.4/35?
you guys realize different cameras use different f-opening for focusing on live view? at least 5d2 used f1.4 and rebels use f2.8 with f1.4 lens wide open. (if you are not using the dof button)
p.3 #15 · Risk of focus shift issues in new 1.4/35?
Hmm. Sebboh, you are right about spherical aberrations - that the light rays don't converge in the same spot. I'm not exactly sure though how a focusing screen works and how it would eliminate rays from the f/1.4 ring. It could perhaps be by not letting through light rays that come in at a steep angle, but that doesn't really add up. The precision matte screens are typically darker than the stock high-DOF screens.
Anyway, you could be right - I'm not ruling it out. Your theory does sound plausible.
As for auto focus, I just did a test. I used my Canon 50/1.4 which does suffer from focus shift, but it's usually not noticeable as it's all kept within DOF. I focused at a ~45 degree angle on the center spot of an ISO chart (lens wide open), stopped it down to f/2, focused again, and to f/2.8 and focused again. I took a shot each time:
As you can see the AF definitely sees f/1.4 or the first shot would have been off while the f/2.8 shot would have been in focus. In this case it focused at f/1.4 in all three images and the focus was off in the f/2 and f/2.8 due to focus shift.
p.3 #16 · Risk of focus shift issues in new 1.4/35?
denoir wrote:
Hmm. Sebboh, you are right about spherical aberrations - that the light rays don't converge in the same spot. I'm not exactly sure though how a focusing screen works and how it would eliminate rays from the f/1.4 ring. It could perhaps be by not letting through light rays that come in at a steep angle, but that doesn't really add up. The precision matte screens are typically darker than the stock high-DOF screens.
yeah, that's the big thing i was wondering about as well - i really have no idea how the optics of focus screens work, particularly how they limit visible dof.
p.3 #17 · Risk of focus shift issues in new 1.4/35?
denoir wrote:
I think it's the other way around - the AF sensor can see no more than f/5.6. Canon cameras can't for instance use AF on f/8 lenses and and the AF is barely usable for f/5.6 lenses. AF focusing f/1.2 lenses with perfect accuracy (such as the 50L) is on the other hand no problem as long as you stay wide open. Stop down a bit on the 50L and AF starts to miss due to focus shift (as it focuses wide open and stops down just before taking the shot).
What you're saying is true, but I remember having read on this forum, and I am not able to find the thread or the context, that AF sensors receive the equivalent light of an f/5.6 aperture because of the angle at which the light hits them. This does not contradict your statement, both can be simultaneously true.
Lars Johnsson Offline • • • • • • • Upload & Sell: Off
p.3 #19 · Risk of focus shift issues in new 1.4/35?
denoir wrote:
I think it's the other way around - the AF sensor can see no more than f/5.6. Canon cameras can't for instance use AF on f/8 lenses and and the AF is barely usable for f/5.6 lenses. AF focusing f/1.2 lenses with perfect accuracy (such as the 50L) is on the other hand no problem as long as you stay wide open. Stop down a bit on the 50L and AF starts to miss due to focus shift (as it focuses wide open and stops down just before taking the shot).
The 50/1,2 lens like you say has no focus shift when shooting wide open. When stopped down a little bit a little bit it has focus shift up to about f/2,8 or f/4. Smaller f/stop and you can't see it. And it only has focus shift at very close distance. At normal or longer distance there isn't any shift that you can notice.
And the comment about Canon bodies and AF sensors is not really the truth. I have 3 different Canon bodies and they can all AF very good at f/8.
This is how the focus shift looks on my 50/1,2. It's not a big deal and in real life shooting I have never noticed it. But like the others wrote. Floating elements is the cure for focus shift. But it often make the bokeh less smooth/nice also
wide open no shift http://www.pbase.com/larsjohnsson/image/103117897/original.jpg
