May 28, 2014

Depth of Field in Macro Focus Range Videos

A recent discussion on the Prototype Railroad Modeling Yahoo Group got me wondering about the effects of f-stop on depth of field when shooting videos in macro focus ranges. I made this short video to investigate how the sharpness varied at three different f-stops, f4, f22 and f29. I shot all using a Canon 70D, with 48mm focal length at the minimum focus distance.

I used a scene on my 00/009 Scale WWI layout, as it was already set up for lights and camera.  The camera was about 6 inches from the layout edge with the 18-135mm zoom. If you want to read more, I included an updated discussion of aperture and depth of field below the video. I recommend you watch in HD so you can see the effects of the parameters better. Note this is a new upload, the original upload did not allow HD. This one does.

Depth of field is probably the single most important parameter in model railroad photography. It is also important in model railroad videography. It is useful to understand why.

Two primary factors affect depth of field, aperture and focus distance in model railroad photography. A third factor, image format, is fixed by the type of camera you use. If are you comparing across different film or sensor  formats it usually means that the same image framing or composition will be at different focus distances.

The brackets in the middle row of
numbers represent the depth
of field for each f stop of this lens.
(source Wikipedia)
When we a photograph a prototype railroad scene with a locomotive, we usually need to stand well back from the subject to make sure that it can be fully captured in the frame. As the focus point of the subject moves away from the camera the effective depth of field increases. Most DSLR camera lenses have a scale engraved on the lens body to show how the depth of field changes with aperture and focus (see image at left from wikipedia). Eventually the focus point reaches the hyperfocal distance, the point of focus where everything from half that distance to infinity falls within the depth of field. Usually prototype rail photographs have the subject and background scenery in the hyperfocal region. The bottom line effect is that most of the scene is in sharp focus.

If we attempt replicate the same composition in a model railroad setting, we find that we are not in the hyperfocal region. In model railroad photography and videography depth of field is only a few inches deep. It is a property of camera lenses that the depth of field region shrinks as the focus point moves closer to the camera. This region shrinks faster than the scale distance diminishes. Therefore, not all of the image will be in sharp focus. This is particularly severe in N scale models because the camera must typically be placed very close to the models to replicate a prototype shot, such as a three quarters trackside view of a passing train.

The shallow depth of field one gets at these macro distances makes even the most realistic scenes look like miniatures. Furthermore, prototype scenes can be make to look like miniatures by manipulating the  depth of field either in post production  or with tilt shift lenses. For example see this tutorial on Vimeo

The Sandpit from Sam O'Hare on Vimeo.

To increase depth of field to simulate the hyperfocal region we use as small an aperture (also called f-stops) as possible since the smaller the aperture size - the greater the depth of field. Pinhole lenses were popular because the small aperture of the pinhole allowed for great depth of field. Unfortunately, pinhole lenses were hard to come by, tricky to make and difficult to use because of the dark image in the viewfinder.

Furthermore, there is a limit to the sharpness obtainable from small apertures. There is a point where the aperture gets so small that the light diffracts as it passes through the hole and the overall image loses sharpness. It is the size of the aperture that governs depth of field. Aperture sizes associated with an f-stop vary from camera lens to camera lens. In other words, f22 on one camera may not have the same size aperture as f22 on another.

To understand the process requires the use of the simple formula, f-stop (N) equals lens focal length (f_l) divided by the actual aperture size (D), N = f_l/D.

The table below shows aperture sizes for the two cameras that I have tested. You can see that the Canon Rebel XT at f29 and 27mm focal length has the smallest aperture at 0.931 mm, and therefore greater depth of field than the 50mm macro lens at f32.

Lens focal length (mm)
Aperture diameter (mm)
Camera –Lens Combination
Canon 5D with 50mm macro lens
Rebel XT with kit zoom lens

You can compute the aperture size for your camera by applying the formula. Be careful in comparing lens and f-stops as some manufacturers report equivalent focal length when describing their lenses and not the actual focal length, especially for the compact digital cameras. The equivalent focal length is the comparable focal length for a "conventional" 35mm film camera lens.

The recent arrival of focus stacking software  changes the calculus of depth of field for still images. Although the optical properties of the individual images do not change, focus stacking  post image processing of multiple exposures with different focus points to construct an image that is sharp all over. When used in model railroad applications, focus stacking allows us to create model railroad images that simulate the hyperfocal sharpness of a prototype image. I have found my macro lens very useful when combined with focus stacking. The ability of the macro lens to focus just an inch or so in front of the lens means that I can capture images where virtually the whole image is sharp, from immediate foreground to distant background

However, in videography, focus stacking is not an option. So we are back to shooting at small apertures. But, the problem arises that with small f stops, you need to slow the shutter speed to let in enough light. But, you can not use long time exposures in videography. If you are limited to 1/30th of a second exposure or faster, then you will need to either bump up the ISO or add more light. My Canon 70D can go over 51,000 ISO, but noise increases as the higher ISOs. I used 2000 ISO  in the video above and it shows acceptable results.

A better solution is to add more light. One 700W flood is usually enough for a small scene. Bigger scenes will require more light. But then you can shoot at wider f stops as you are unlikely to need to be in so close, so depth of field increases. Lighting is an important subject  and deserves its own discussion. But understanding the basic physics and performance of your lenses is important too.

Videography is much more complex than still photography. Many of the techniques needed for good model railroad photography still apply, but they are insufficient to insure you have a good video. This will be a topic I revisit in  upcoming blog posts.

No comments:

Post a Comment