# The depth of field issue

2539 Views 68 Replies 13 Participants Last post by  John Mason
In the SHO HD Disappointing there were some very interesting comments about depth of field in HDTV video vs. film. Some thought they were OT there so I decided to start this thread. I'll just copy some interesting quotes.

mmost:
Quote:
 What I think you're talking about is depth of field. Depth of field is determined by a number of factors, none of which are whether the system recording the image is photochemical (film) or electronic (tape). The factors affecting depth of field are the length of the lens, size of the aperture, and size of the image plane. 35mm film has an image plane that's about 5 times larger than a 2/3 inch video CCD imaging device (used in just about every professional standard and high definition video camera today). It therefore has a shallower depth of field than video. If there were a video camera with an imaging chip equal in size to that of 35mm film, and its sensitivity were about the same as a specific film stock, it would have the same resulting depth of field when shot with the same length lens. This is only one of many characteristics of the resulting image that are different between the two systems.
PrimeTime:
Quote:
 Depth of Field, or hyperfocal distance, is a function of F-stop number. F-stop is the ratio of the focal length (the distance from the lens focusing plane to the imaging plane) to the iris aperature width. It is independent of the size of the imaging plane.
mmost:
Quote:
 Although I doubt that there are very many people here who find this whole discussion in the least bit interesting, I would say that I know what hyperfocal distance is. However, the image plane comes into play in the sense that to get the same field of view with a smaller imaging plane, a wider lens (shorter focal length) must be used. Since the distance between camera and subject is often limited by the physical space available, a camera with a smaller imaging plane (i.e., a video camera) will almost always be using a much wider angle lens for, say, a medium shot or a master shot than a camera with a larger imaging plane, say, a 35mm film camera. So for the same field of view, given the physical constraints of a typical set, the system with the smaller imaging plane will almost always have a great deal more depth of field than the one with a larger imaging plane. Although not part of the technical equation, in a practical sense it is certainly a factor.
Well I personally do believe the discussion is interesting. That's of course why I started a new thread for it. And I was not previously aware of the effect of the size of the imaging plane in all this.

But I think that control of depth of field (or lack thereof) is important for film makers and one of the reasons there has been some resistance to HD video cameras.

If you can have (optional) shallower depth of field then you can focus the viewers attention where you want and not have to pay as much attention to possibly distracting background details. Movie makers take advantage of this.

But there is a second advantage for shallow depth of field. If the background is out of focus this acts as a blur filter, removing high frequency details and allowing the movie to compress better in MPEG-2. This is especially important when the background is moving since a fast pan over a detailed background is very hard to compress and can result in great loss of foreground detail and nasty macro blocks as the encoder gets bit starved. I don't know how much movie makers worry about macro blocks, but maybe they do. Sometime it's good to have certain things blurry.

All this is why I'd like to see more control of depth of field on HD cameras.

Can anyone shed more light on this issue/problem? For instance, is it possible to get special lenses for HD cams that optionally decrease depth of field in some way. Or is it possible to shoot with a wider aperture on HD cams? Or is there any other way to do it?

mmost?

- Tom
See less See more
Status
Not open for further replies.
1 - 20 of 69 Posts
The one thing that can be controlled is the amount of light reaching the imaging device. This can be done by placing a neutral density filter in the optical path. This increases the size of the aperture necessary to achieve the same exposure, thereby reducing the depth of field.
Then I wonder why they don't do that in some cases to make it appear more film like. Do they?

I've often thought the difference in "film like" maybe had as much to do with this issue as it does with grain. Video's great depth of field is great for a travel documentary where you want the viewer to be able to choose to look anywhere in the picture and feel like "being there". But the same thing probably hurts in a drama.

- Tom
Quote:
 Originally posted by trbarry Can anyone shed more light on this issue/problem? For instance, is it possible to get special lenses for HD cams that optionally decrease depth of field in some way. Or is it possible to shoot with a wider aperture on HD cams? Or is there any other way to do it?
You can try lighting tricks such as lighting the background to be darker, thus showing less detail (that's what Bobby Byrne did on "Titus"), but the only thing you can really do, as spwace has already said, is change the amount of light hitting the image plane. This requires that on a video camera, you need to shoot with the aperture "wide open" in order to obtain the shallowest depth of field with a given lens. This in turn means that the lens must be "fast" enough (i.e., transmit enough light through the various lens elements) to allow that low light level to make an acceptable picture. The various professional lens manufacturers (Panavision, Zeiss, Canon, Fuji, et. al.) have responded by designing a new generation of lenses that are, in fact, the fastest lenses ever made, and this has allowed cameramen to shoot 24fps at f-stops as low as 1.6. This does, however, come at a price: the amount of light that passes through the lens at f1.6 is a bit below the normal "sweet spot" of the lens (usually around f2.8-f4.0 on most professional lenses), so the images are not quite as sharp as the lens is normally capable of. In addition, it is normal for all lenses to have sharper falloff at the corners than in the middle, and this is accentuated when operating at such wide open stops.

The only way video cameras will achieve shallower depth of field without image compromises is to create new cameras with larger image planes. This is, in fact, being done - a company called Dalsa is showing a prototype this week at NAB of a camera that uses an array of 4 CCD chips per color channel to achieve an imaging plane that is effectively about the size of 35mm film - higher resolution and much more control of depth of field at the same time. It is, however, only a prototype, an "NAB product," if you will, and these things usually take at least a year (or more) before they begin to evolve into something real.
See less See more
Quote:
 Originally posted by mmost You can try lighting tricks such as lighting the background to be darker, thus showing less detail (that's what Bobby Byrne did on "Titus"), but the only thing you can really do, as spwace has already said, is change the amount of light hitting the image plane.

Or use one of the new HD video cameras that has the same sized film exposer area (sensor) as a regular 35mm camera.

http://www.dalsa.com/dc/dc_sensor.asp
Quote:
 One of the obvious advantages of a huge sensor is the increased resolution. DALSA's sensor has a resolution of 4096(H) x 2048(V). Eight million full pixels. Four times more resolution than HD. In fact, no projector available today, either digital or film-based, can project the camera's awesome full resolution. Downsampled for compatibility with today's most advanced digital postproduction workflows, our system still far exceeds any HD system...all the while providing a path to the future.
Whoa!

"NAB Product" or not, that sounds like an impressive camera. Taking a downsampled 1080p image from 4096(H) x 2048(V) sounds quite nice. And it claims to handle the crushed color ranges also.

Though I'm still curious if anyone uses the idea of neutral density filters just to lower depth of field.

- Tom
Quote:
 the amount of light that passes through the lens at f1.6 is a bit below the normal "sweet spot" of the lens (usually around f2.8-f4.0 on most professional lenses)
Quite the opposite - the lower the F number, the wider the opening, thus more light passes. Lenses with low F numbers are called "fast", because they allow more light to pass which in turn allows higher shutter speed.

The relation is like this: lower F - more light and less depth of field, higher F - less light and more depth of field.

Hope it helps,

Kirill
Quote:
 Though I'm still curious if anyone uses the idea of neutral density filters just to lower depth of field
Tom,

There is no relation between ND filters and depth of field. See my post above.

Kirill
It's too bad we can't see in ultraviolet ?;~)

Tom, the whole deal with using filters is that

1) The greater the brightness (using microscope terms again, sorry if confusing), the greater (in general) the resolution can be...as that photons carry data and the more of them..., depth of field is basically the vertical area that is mostly focused as well as the actual plane of max focus. The higher degree of magnification (or sharpness if you wish) means that the *depth is ***similarily*** detailed and is proportional to the degree of magnification, holding working distance length and aperture sizes constant. If you resolve details microns apart, then on the up/down planes, you can see mebbe about a few times or an order of magnitude in depth of field. Now, when you want to *see* into a picture, it would suck if the picture was so sharp that you couldn't tell how far away the mountain actually is, even if you could tell exactly where each house is located on it. We solve that problem by decreasing brightness (filters), which decrease resolution, to one we are happy with in exchange for depth of field.

2) Using filters will also decrease chromatic abberations and other optical faults.

One comment to make about those hi -rez cameras. I'm not sure we will *ever* see it at that high rez, when it comes right down to it. I think what will likely happen is that the center of the camera will highly utilized and the pixils farther out will get much less data. Thinking on that, film simply will increase the noise wheas the chips will show wierd imaging artifacts that will probably have to be filtered--sperical and color distortion probably would happen quite easily. (how big physically are the chips?) The biggest problem would be that at that resolution, I can *easily* imagine that it might be *too* clear. Why would that be a bad thing? Simple, just imagine that you just *lost* an eye. A non-stereo view of a movie or documentary at that high resolution would probably be *highly* annoying. OMG! Deer! Run for your life! That LION is gonna EAT YOU!! oh wait, he's 200 meters away, never mind. Or the requisite stalking scene in a movie...or in any scene where it's important to convey distance.

Fix it? At the same sharpness? Ummm would you like a nice panavision P&S style blurring all the time during certain scenes? Betcha they start trying to get people interested in high-def stereo movies (hey, it's not THAT far away), or get serious about moving holograms.

Darius
See less See more
I have a question-

why does some material shot with 1080p/24 HD cams still look like film as far as depth of field is concerned as compared to 1080i/30 HD cam work? Take the ABC sitcoms for example that are shot with 1080p/24 HD cams. They look nothing like what you might see on say, HDNET.

Interested..

Andrew B.
First of all, sitcoms and some dramas are totally nondemanding as far as depth of film are concerned. We usually have have a set that is only a few feet across. Second, they do not use the full resolving power of the cameras, even before downrezzing to 720p.

And they still don't look that great...Max Brickford is a classic example. As far as the Tonight Show is concerned, the set is usually *tiny*, thus there is no real need to worry about depth

Darius

edit: feets!=correct english
Read more about depth of field here:
http://www.c-j-ball.ndtilda.co.uk/field.html

and here:
http://www.cs.mtu.edu/~shene/DigiCam...-of-field.html

Hope it would clarify things

Kirill
Quote:
 Originally posted by shah8 First of all, sitcoms and some dramas are totally nondemanding as far as depth of film are concerned. We usually have have a set that is only feets across. Second, they do not use the full resolving power of the cameras, even before downrezzing to 720p. And they still don't look that great...Max Brickford is a classic example. As far as the Tonight Show is concerned, the set is usually *tiny*, thus there is no real need to worry about depth Darius

But The Tonight Show is so much crisper all the way throughout the image than the aforementioned ABC material. I am just going to chalk it up to using less than the full resolution.
The smaller the set, the better for using the full rez...

Darius
Quote:
 There is no relation between ND filters and depth of field. See my post above.
I did see your post above but I thought that using an ND filter would alow (require) you to also use a larger lens aperture and this would decrease depth of field. Is that then not true?

- Tom
Quote:
 Originally posted by Kir Tom, There is no relation between ND filters and depth of field. See my post above. Kirill
Yes there is, placing a neutral density filter in the optical path requires a lower fStop to achieve the same exposure and depth of field is related to the fStop.
Appreciate Tom starting this because depth of field interests me. But, as others point out, don't think we've pinned down the 'crispness' distinction between HDNet/PBS HD-camera tapes and most telecined HDTV. Depth of field and resolution differ. I theorized about the 'smoothness' contrast between the two formats here recently. The original negative of 35mm productions, needless to say, can carry both extraordinary depth of field and resolution, if desired. -- John
Quote:
 Originally posted by Kir Quite the opposite - the lower the F number, the wider the opening, thus more light passes. Lenses with low F numbers are called "fast", because they allow more light to pass which in turn allows higher shutter speed. The relation is like this: lower F - more light and less depth of field, higher F - less light and more depth of field. Hope it helps, Kirill
While everything you say is correct, the original comment meant to convey something completely different. Simply that lenses do not "perform" as well completely wide-open. They are generally less focused than they are at smaller apetures and vignetting is generally worse. Link on vignetting - mentioned already as "darkening of the corners".

http://www.vanwalree.com/optics/vignetting.html

I think mmost said that "...you need to shoot with the aperture "wide open" in order to obtain the shallowest depth of field... ". You said, "...more light and less depth of field..." I'm interpreting your use of "less" as being synonymous with "shallow", so you're both saying the same thing.

Thanks,

Ron
See less See more
John,

I know for a fact that Kodak has some insano (yeah, not a good word) algorithms that are being used to clean up grain as much as the artistic interests will let them before encoding. It's been used heavily on old films for DVD transfers, current motion picture productions and other transfers. (I know the guy who wrote them). Is it possible that something like that might be accounting for loss of resolution we think we should be seeing in film? That's really what we're talking about here - why we aren't as blown away by film sourced HD content.

Tom, sorry for the OT, I'm not sure if John is still checking the other thread.

Ron
On the resolution issue, it may well be that film grain is just too expensive to properly compress with MPEG-2 at HD bit rates. If they leave it in we get macro blocks and if they try to filter it out we get softness.

And I suppose each new generation of prints adds a new generation of grain. ??

- Tom
1 - 20 of 69 Posts
Status
Not open for further replies.