Art Pete Pete Ashton is (becoming) an artist
The problematic depth of fields
Why are the corners of these two TTV photos taken with identical equipment so different?
This post is a little technical so if you’re only interested in the pretty pictures you might want to skip it. If you’re interested in how the pretty pictures get made then read on.
In December I switched cameras for my TTV work. For the last three years I’d been using a Fuji Finepix S7000 to shoot through my Duaflex. This is essentially the same as a compact camera only chunkier and with more buttons. The lens is fixed, the viewfinder is digital (rather than letting you see through the lens) and any SLR-type functions are if not actually present then digitally mimicked. I found to be a good bridging camera between my crappy point-and-shoot and my Nikon D70s, giving me a taste of what could be done with more control over my photography and a hunger to do it properly. I switched to the D70s for “proper” photography around the same time I discovered TTV and since I’d decided to build the digital camera into the contraption and the S7000 was just sitting there…
When I shot TTV with the S7000 I had it set to Macro mode. I don’t fully understand how this works optically but I found I was essentially focussing on the surface of the viewfinder. It was rather like the image was being projected onto a screen and I was taking a picture of that screen. Roughly speaking. This meant there was no sense of depth to my images and everything was in focus no matter how far away it was.
(Actually, to be completely accurate there was a point about six inches from the Duaflex within which the focus failed so you could say the depth of field was six inches to infinity, which for a macro lens is kinda wrong, but like I said, I don’t understand it.)
But whatever was going on I didn’t mind. The combination of cameras made nice enough photos.
And then on December 21st disaster struck as I slipped on the ice and the S7000 went skittering across the concrete. The fixed lens is not quite so fixed any more – it just sort of hangs there like the broken nose of a sad dog. So I had to build a new contraption and use a different camera, my Nikon D70s with a 50mm f/1.8 lens. (I also use a +4 close-up filter but all this does is allow the D70s to get closer to the Duaflex – it doesn’t change fundamentally change the optics to the best of my knowledge so I won’t mention it again.
I’d better do a quick lesson in How Cameras Work here. If you know How Cameras Work feel free to skip this.
A camera makes a picture my letting a specific amount of light fall on a light sensitive object, be it film or a digital sensor. The amount of light let in is controlled in two physical ways, shutter speed and aperture.
The shutter speed is simple – it’s the length of time the shutter is left open and is usually expressed in fractions of a second. So a shutter speed of 1/60 means the camera is open for one sixtieth of a second. With all else being equal (and we’ll make things unequal in a minute) you need a fast shutter speed for bright conditions (say 1/1250) and a slow shutter speed for when it’s dark (say 1/15). Other than letting the right amount of light in the speed of the shutter also affects the image itself. If the subject is moving a fast speed will be crisp and sharp while an overly slow speed will be blurry.
Aperture is a little more complex, but only a little. This is the size of the hole through which light is let through. The bigger the whole the more the light, obviously, and this is, again, measured in fractions. Kinda. All you need to know is the lower the number, the bigger the hole, so f/1.8 lets in way more light than f/8 which lets in more than f/16. Here’s a handy diagram on Wikipedia. Again, there’s an optical effect beyond simply letting in light. The smaller the hole, the deeper the depth of field. Or to put in another way, shooting at f/16 will show more of the image in focus than shooting at f/4 as shown in this illustration on Wikipedia:
Actually, the Wikipedia article on Depth of Field is worth reading as there’s a lot of handy diagrams in there. There’s also a really need camera simulator which lets you see the effect of shutter speed and aperture on the image (for the sake of this post don’t worry about ISO for now).
Those of you who knew all that can start reading again now.
So what we’re after is a fast enough shutter speed and enough of the image in focus as possible. With the S7000 this wasn’t a problem as, due to the nature of whatever was happening, I was working with what appeared to be infinite depth of field regardless of the aperture. With the D70s, however, it’s more complicated than that.
It would appear that having previously been photographing the viewfinder of the Duaflex I’m now literally shooting through it. And that means the aperture is suddenly a lot more important. Here’s a diagram.
(Pictured is my new Argoflex 75 which hasn’t been covered in duct tape and turned into a contraption… yet…)
That very rough but you hopefully get the notion. Here’s some examples of it in action.
This photo was taken at f/10. The whole image is in focus and the frame of the Duaflex glass has a clear edge.
This photo of the same fence taken at the same distance was at f/5. The centre of the image is still in focus but it’s starting to blur at the edges. However, the frame is still acceptably sharp.
Finally, f/2.2 where the circle of blurring is very distinct.
Now, some TTV photographers like the very blurry image. When I was moaning to Stu about this he seemed surprised it was a problem. But I’ve been having my whole frame in focus for years now and it doesn’t aesthetically please me. The framing is the thing, particularly when I can get effects like on this photo.
This might not seem a problem to you. Surely if I just shoot at f/5 or above everything will be okay? Well, yes, assuming there’s enough light. You might have noticed that all my TTV since Christmas has been outdoors on relatively sunny days. In order to get avoid camera shake and moving objects blurring I need to have a shutter speed of at most (or is it least?) 1/20 which means I need to have as big a hole as possible. But in order to have everything in focus I need as small a hole as possible.
So until I figure this out I don’t think I’ll be doing much indoor or night-time TTV photography. Which is a shame, so I’m determined to figure it out. (I suspect I might need to solidly fix the D70s to the contraption so the frame is always in focus even if the subject is blurry, like here or here)
It does, however, explain the blurring that happened with my Bubblebottle contraption as seen in this montage from 2007:
But more about that another day.
I think you’re still focussing on the screen of the view finder.
I’m guessing the Fuji aperture only opens up to f4 or so ? which would partly explain it.
Also SLRs have a bigger sensor than compacts/’prosumers’ which is generally good as it means more light so less noise. But it does also mean that the depth of field is more noticeable, which can be good if you want to use that. So, not 100% about my facts here, but I think two cameras with same aperture, but different sensor sizes will have different depths of field – so a compact will tend to have more things in focus, which also explains what you’ve found in swapping from the fuji to the slr. I think if you tried a full frame slr (sensor size same as 35mm film) it would be even more pronounced. So even at f4 the fuji will have a bigger depth of field than an slr. Does that make sense ?
If I’m right, doesn’t really help, but at least explains it – be interested what you find from anyone else.
This is an interesting subject and my thoughts may be of interest. The viewfinder camera (the Argoflex in your picture) can be focused but has a fixed, wide open, aperture. This lens is not as good as the other, picture taking lens that is situated below it, and typically will have some edge distortion. Even the viewing lens on my Rolleiflex has some distortion which is understandable since, until TTV, no-one ever took pictures using the viewing lens. It was simply a tool for making the composition.
The Nikon D70s, on the other hand, has a high quality lens with very little edge distortion at all. But the depth of field, as you rightly point out, is deeper the more the aperture is stopped down. This means that any inaccuracies in the registration between the two cameras will be lessened the more you stop down the Nikon lens. This presents problems in low light but will give you sharper images. These images will never be sharper than the viewing lens of the TTV camera will provide.
Personally, I like the edge effects and I like the framing. Not so sure about the dust, but that comes from years of using the Photoshop stamp tool to remove specks from scanned 35mm slides.
I agree with Mike, I can’t image the system working at all unless you are focusing on the screen itself (this is exactly how “depth of field” adaptors like this – http://www.pstechnik.de/en/digitalfilm-pro35.php – work).
Is the viewfinder screen in any way curved or could the DSLR lens be not perfectly centred on it? If the “taking” lens (the 50mm, which incidentally will be equivalent to about 40mm with a +4 diopter) is set to f/1.8 and the “subject” (the viewfinder screen) is say, 5 inches away; the depth of field is going to be measured in millimeters.
With such small depth of field, even without curvature in the viewfinder screen just the distance caused by the geometry between the centre of the DSLR lens and the centre of the viewfinder screen, compared to the centre of the DSLR lens and the corners of the viewfinder screen, is going to make a big difference.
That’s also leaving out the slight edge sharpness you would expect to see from a diopter filter unless you spent thousands on an astronomy-grade piece of glass.
I would also expect that this situation is worse with the DSLR compared to the point and shoot, as the former’s got a bigger sensor, translating to less depth of field.
… a better way of imagining what I just described is to visualise the plane of focus from the DSLR as a *sphere* with its centre roughly on the middle of the lens, rather than the typical DOF diagram showing an isosceles triangle with the point on the camera sensor and the flat plane of focus on the subject.
Thank you! this is excellent and further to your mail to help me out…as soon as I see the sun again I’ll put all this into action!
Just to add, pretty sure the screen will be quite curved (so fits in with what Ed’s saying), think its called “bubble glass” instead of “ground glass” which is flat and used on more high end view finder cameras. So, although I didn’t explicitly say before I’m assuming the depth of field is to do with the centre of the screen being higher than the corners.
I think your picture of the logs(think thats what they are) backs this up – the blurring isn’t because the logs on the edge are further away – assume they’re all in the same plane?
@edmoore is absolutely right
What would be interesting is if you could move each of the elements individually, in much the same way as you would in a view camera, you could do some pretty interesting things with the perceived depth of field.
@Matt, if there was a way of moving the lens on the front of the viewfinder camera, you’d effectively have a tilt-shift system.
@edmoore Exactly, and it’d be nearer to a view camera than a tilt/shift lens as you’d have full control of the film/sensor plane as well (and descend into the world of the Scheimpflug principle). In many ways you could take the TTV concept further by building a contraption from scratch, stripping the TLR down and rebuilding it with some sort of bellows system. Although this might take some of the fun away…
In order to have exact registration between viewing lens and taking lens the last thing you need is tilt/shift. Unless you want a tilt shift effect which would be akin to the lenses that are available to make a real scene look like a model.
Pete, is the viewing screen on your camera ground glass or “bubble” glass? If it’s ground glass you should be able to get perfect registration in which case any edge effects will be entirely attributable to the quality of the viewing lens optics.
Another whole way of doing this would be to remove the glass viewing screen altogether and couple the camera lens to the viewfinder lens, rather like a camera is attached to a microscope. But that might just negate the principle of TTV photography. The screen is an integral part of the art form.