Manual Focus Ain't So Easy


Manually focusing fast lenses with a DSLR might surprise you!


February 17, 2007

Introduction
You just bought your shiny new DSLR and are beginning to run into the issue of affording new lenses for it. And heck, maybe its not even price. Maybe they just don't make what you need or maybe you don't want to get some huge zoom to do something as simple as shooting at 85mm and f/2.0 or faster. Or maybe you just have some old lenses hanging around that you'd like to try. Well those things describe my situation and the situations of a lot of other people too. And the solution seems simple. Get a lens adapter and buy some good old lenses on ebay.

Sure. That's it. You can buy adapters on ebay for just a few dollars ($20-$40 U.S.) too. DSLRs focus and meter through the lens (TTL), so the problem is solved! Right?

Well, unfortunately life isn't so simple. The focus screens typically found in modern DSLRs may not be quite up to this task depending on what lenses you intend to use.

When Focus Screens Lie
Modern focus screens seldom have focus aids. Most just have some markings to show AF and/or metering areas. But other than that, they tend to be plain "ground glass" screens. The split prism focus aid and the microprism collar that were common when I got my first SLR went away long ago. The invention of phase detect autofocus swept most of that away. And guess what, the focus screens in modern DSLRs really aren't ground glass either. Ther aren't ground, nor are they glass. They're typically plastic and their surfaces are made up of an a bunch tiny microlenses. Now I'm not an anti-plastic snob. Plastic can be a super material. But those pesky microlenses definitely drop a fly in our legacy, manual focus, lens ointment.

So what's the big deal? Why should you care about a bunch of little microlenses. You can still focus on that screen. Heck - we've all done it. You can sit there and see the image go in and out of focus. So what's the problem?

Well, the problem usually isn't immediately obvious. Sometimes these new focus screens do an OK job at manual focus. But at other times, they flat out fail.

The problem begins with the desire for nice bright finders in our DSLRs. Most people want bright finders. But getting that in a modern DSLR presents obstacles. Since most DSLR formats are smaller than 35mm film, most DSLRs use more screen magnification than 35mm film cameras. More magnification leads to a darker screen. The autofocus sensors (now a standard requirement) need to be able to look at the image at the same time you are using the finder to compose your picture. So light must be diverted to them for that purpose. The reflex mirror that sends the lens image up to the focus screen and your eye is partially tranparent to allow some of the light from the lens to find its way to the autofocus sensors. That results in less light getting to the focus screen and means a dimmer screen. More and more people use slow or moderately fast zooms instead of faster primes these days. Oops. There goes some more light.

So slower lenses, a smaller format and the needs of autofocus all work against the engineer's goal of delivering a bright finder to you. The poor engineer has to find solutions, and there are only so many photons coming through that slow lens to work with dang it!

So the engineers put microlenses into our focus screens. These microlenses are really efficient at passing light at a paricular lens f-number or range of f-numbers. For lower priced cameras, the camera engineers assume the user will mostly be using lenses with apertures of around f/3.5-f/5.6 - so those cameras have screens that are efficient at that range. They expect buyers of more expensive cameras to be using lenses that are somewhat faster. Maybe f/2 - f/3.5 is more common for that group of buyers. So they use screens that are more efficient at that range for those cameras.

This all works pretty well. Most finders are fairly bright. But there's a price to be paid when you start doing stuff that the engineers weren't counting on. Like mounting that legacy f/1.4 manual focus lens.

I'm not going to try to explain exactly how these microlenses work. That's mostly because I'm not fully sure of the details. But the basic idea seems to be that they are very efficient at passing light that hits them from certain aperture sizes. This works great when you have a lens of that aperture range mounted. You get a very bright image since the screen is so efficient at passing light at those apertures. But if you mount a faster lens, you don't get a proportionally brighter image. The microlenses don't pass that extra light very efficiently at all.

Larry J. Clark ran tests on the Olympus E-1 and E-500 using a one degree spot meter to measure screen brightness at different apertures and his result show the non-linearity quite nicely. Click here to see a graph of the results. More details can be found here.

Now from the standpoint of viewfinder brightness, this is no big deal. If the screen is bright enough at f/4, then it will also be plenty bright with an f/2.0 lens mounted. Maybe not as bright as the more expensive camera that has a screen tuned for f/2.8 or faster, but still plenty bright. And it allows you to have a pretty bright image with your inexpensive slower lens. That's a good thing!

But from the standpoint of manual focus, it isn't necessarily so good. The rays from the larger aperture that are being blocked from getting through are the rays that show the most misfocus. This means that the focus screen has the effect of showing screen focus that is about the same as if you had stopped your lens down to f/4!! Now if you are shooting at around f/4 or slower, this isn't a problem. The screen focus will be very similar to the focus delivered by the lens. But if you are shooting at f/1.4, it's a big deal. You simply can't see the point of sharp focus. So when shooting at a large aperture with shallow DoF - precisely the place where you tend to need good critical focusing capabilities - you have a focus screen that shows too much DoF and makes the image appear to be in-focus when it isn't.

Lots of people don't believe this. It is counter-intuitive. And if they never focus with fast lenses at fast apertures, they won't be able to notice it. But if they mount a legacy 85mm f/1.4 to the typical DSLR and shoot at fast apertures, they may soon be wondering if maybe their eyes need checking. After all, it sure looked in-focus. It really did! Blink - blink!!

Demonstration
The following images demonstrate this effect with an Olympus E-330. I've observed this with an E-500 and and E-300 as well. I'd expect less of this effect with an E-1 since it was designed for professional use and its finder caters to the use of faster lenses. But I would still expect to see some of this happening at apertures faster than f/2 and maybe faster than f/2.8. It should be expected from other makes as well - especially the lower priced models.

I used an Olympus C7070 to take pictures of an image as seen through an E-330's optical viewfinder. I shot at different focus and aperture settings of the E-330. At the same time, I also took images with the E-330. This way I can show what the viewfinder displayed and compare that to what the camera lens actually recorded.

I used a 50mm f/1.4 Takumar SMC as the test lens. This lens is fairly soft wide open on a four-thirds camera, though it sharpens up nicely at f/2.0. I used the E-330 because it has 10x LiveView Mode B. This lets you focus directly from the camera's main imaging sensor with very high accuracy. You see exactly what the sensor sees - at 10x magnification. Way cool!

This first image set shows a 100% crop of a sewing needle eye. The lens has correct focus. The image next to it is what the C7070 saw through the E-330's optical viewfinder.

Note that the view through the focus screen taken with the C7070 give us a very high effective magnification - around 4x. Also note that the image taken with the lens on the E-330 is soft, which is charatisteric of this lens at f/1.4 with the E-330/300/500.

In the next pair, the lens was purposely misfocused a tiny bit. It doesn't take much at f/1.4 with a 50mm.

We can see the image taken with the lens is now clearly misfocused, but the view through the finder looks the same. It actually looks nice and sharp - slightly sharper the finder image above (because that first finder image apparently had some slight blur due to camera movement). The needle doesn't look out of focus at all.

In this last pair the lens is still misfocused, but it has been stopped down to f/4 which increases the DoF.

The needle looks fairly sharp in the actual image taken by the E-330 lens. The increased DoF at f/4 has helped us out. But the view through the E-330's camera finder recorded by the C7070 is essentially unchanged. The E-330 optical finder never gave us any clue that our focus was off or that stopping down the lens changed the DoF.

All of the images of the viewfinder taken with the C7070 were taken at the same manually set exposure of 1/6 second at f4.5. Even though the lens was stopped down a full three stops from f/1.4 to f/4.0, the screen brightness only changed slightly with a small visible difference only appearing after the change from f/2.8 to f/4.0. The background measurement changed from an average reading of 215 to 203 using Photoshop's picker. That works out to less than a third of a stop in brightness change when I compare it to a Stouffer 4110 transmissive step wedge image take with the C7070. A 1/3 step in the Stouffer image at an image value of around 215 shows up as a numerical value change of more than 15. Since most DSLRs meter from the focus screen, it is easy to see how using a fast lens can give exposure problems as well.

And if the screen is showing the image as though the lens was stopped down to f/4, that means it isn't showing depth of field correctly at fast apertures either.

What to do?
We can pretty easily see that critical focus at fast apertures simply cannot be seen directly through the optical finder of the E-330. And while I didn't demonstrate it, this is surely true of other DSLRs from other makers as well. Using a magnifying aid doesn't help. These images are very heavily magnified compared to what you see in the finder with your nekid eye (I estimate about 4x magnification sitting here at my laptop). And brighter lenses aren't the solution either. They actually are the problem.

Practical solutions vary. The obvious one is to just use autofocus lenses. The next obvious one is to not shoot at fast apertures.

Other possible solutions are:
1) Get a split prism focusing screen. This helps a lot. But the good ones from Katz Eye aren't cheap and they can be a pain or an expense to install. And critical focus with high resolution sensors having small pixel pitches, even though much improved, is still difficult with a split prism.
2) Get a screen optimized for faster lenses or a DSLR that has such a screen. This will help, but probably not solve the problem.
3) If you have a Canon camera, get an adapter that has a chip that enables the camera's autofocus assist. I'd guess this is about as good as a split prism focus aid, but I haven't used one. So I'm just guessing really.
4) Some cameras will give you autofocus assist without resorting to special tricks. But I've heard that some of these are too generous with the "window" of acceptible focus that they give. You'll have to investigate which ones do well on your own.
5) Get a camera with main sensor based live view. The current cameras that have this are the Olympus E-330, Panasonic DMC-L1 and the Leica Digilux 3. These are all four-thirds format cameras. BTW, the Panasonic DMC-L1 also provides manual focus assist using the cameras AF sensors. I have no idea how accurate this is, but this camera may very well be the best overall choice for using manual focus lenses since it also has liveview. The Olympus E-510, E410 (four-thirds) and the Canon 1D Mark III will be released soon and they too will have this feature. I expect this feature to become more common in the future. This gets you very accurate focus - though it isn't always handy to use.

As expected, live view has not only become more common, but could now be considered a standard feature. It is so common that it no longer makes sense to list the cameras that have it. I'm leaving the previous short list intact simply for historical reference. Live view has come a long way since a rather famous camera review site described it as, "a solution looking for a problem" when introduced on the E-330. That site now lists the absense of the feature as a "con" if it is missing.

Conclusion
The point of this article isn't to disuade you from trying manual focus lenses. I use legacy manual focus lenses all the time. The point is to give you fair warning of what to look out for in case you are considering mounting some of these fine old optics. The design of modern SLR and DSLR focus screens can have a significant impact on how easy it is to use legacy lenses effectively. Things in the DSLR's finder aren't always what they appear to be.


Revisions
Revised March 20, 2007 to add live view camera model names.
Revised April 20, 2007 to be more accurate about the available DSLRs with live view, including Panasonic (with MF assist) and Leica four-thirds cameras and distinguishing between cameras currently available and those not yet shipping.
Revised November 2, 2008 to add more comments on live view availability.

You are visitor # since 02-17-2007


If you have questions, comments, suggestions etc. please feel free to write me at:
jay@studio522.com


Copyright © 2007 by Jay Turberville