 |
The setup as described below |
In October 2020 Roger Cicala published in D P Review a method for testing field curvature of lenses using grass as a test subject. Really. Here is the link:
I have for several years been aware that testing lenses by photographing a flat test chart of some kind does provide considerable useful information but cannot tell us about the three dimensional distribution of sharpness including any field curvature, asymmetry or tilt.
It turns out there is a simple way to display this.
All you need to do is photograph a field of grass then apply Photoshop [find edges] filter.
Method
I use grass as a subject. However any fairly flat surface will do as long as it has a clearly evident surface texture.
The area needs to be large enough to fill the frame even with a wide angle lens.
The focus point needs to be far enough from the camera to replicate normal photography, say, somewhere in the 2-5 meter range or more if possible.
The distribution of the in-focus field at close-up distance is very likely to be substantially different from that at normal photography distances.
To avoid giving a false impression of field asymmetry I have found it is necessary to ensure that the camera and a line across the middle of subject are both level. I failed to do this with my initial tests, leading to inconsistent and misleading results.
I lay a straight alignment stick about 2 meters long on my chosen grassy area and using a spirit level move it around until it is level.
I set up a focus target just in front of the center of the stick. I use anything suitable, a leaf or a short stick at right angles to the center of the long stick.
I find that keeping everything in alignment is easier if I mount the camera on a tripod.
I set up the camera a suitable distance from the focus target. This is focal length dependent. I guesstimate that about 100x focal length should be enough.
I place the camera so the optical axis is perpendicular to the middle of the alignment stick.
I activate the electronic level indicator in the camera’s display and use this to level the camera. I move the camera and tripod around until the level indicator is parallel to the alignment stick in the viewfinder.
I locate the smallest available focus area in the center of the frame in the viewfinder and right on the focus target.
I make an exposure at the widest aperture of the lens and more exposures at each whole EV step down to f22.
If the pictures are captured in Raw I open each in Adobe Camera Raw>Optics>Profile>Remove chromatic aberration and Use profile corrections.
Then I open each file in Photoshop and go to Filter>Stylise>Find edges.
This produces the kind of result you see in this post with the sharpest parts of the image represented by the darkest parts of the filtered image.
I can easily see how sharpness is distributed both side to side and from near to far.
Distance near-to-far in the subject is represented by the up-down dimension on the test images.
Example
For this post I show results with the Canon RF 50mm f1.8 lens on a Canon EOS R5 body.
Results and Discussion
I have to say the results were quite a surprise to me. I had not previously understood just how complex is the near-to-far distribution of greatest sharpness and how this changes as the aperture is closed down.
Here is the RF 50mm f1.8 at f1.8.
You can see the system has focussed exactly where I told it to. You can also see that the near-far distribution of sharpness is curved away from the camera towards the edges and also that there is a slight tilt of the zone of greatest sharpness.
RF 50mm f1.8 at f1.8
Next, the same lens at f4.
The distribution of greatest sharpness has changed remarkably.
We see that tilt a little more clearly and we see the edges have sharpened substantially.
But most obviously we see that the shape of the zone of greatest sharpness has changed into a strong W/moustache type pattern.
If a lens tester looked only at the center of the frame, they might conclude that this lens exhibits some focus shift, with the point of maximum sharpness at the center moving back a little. But that would not be an adequate or useful description of the lens’ behaviour at all.
In fact if we draw a line across the frame through the in-focus point we see that the subject is rendered decently sharp right across the frame. The net result of the complex optical events going on here is not focus shift as such but a means by which the lens designers have been able to prevent focus shift taking into account the full width of the frame.
RF 50mm f1.8 at f4
Now I show f8.
Here we see the zone of sharpness extending near and far by an approximately equal amount.
The slight tilt and the W shaped distribution of sharpness are still there but becoming obscured by the overall expansion of sharpness.
Note that as the zone of sharpness expands across the frame, peak sharpness decreases. The sharpest part of the frame at f4 is sharper than anywhere on the frame at f8.
RF 50mm f1.8 at f8
Summary
The way lenses render sharpness in a subject is considerably more complex than I had previously realised.
I would like those who regularly publish lens tests to include the process described in this post, or something similar, in future tests.