Edited by Josh Silfen, 29 December 2005 - 09:50 AM.
Jump to content
Electronic vs. Optical viewfinder
15 replies to this topic
Posted 29 December 2005 - 09:49 AM
I know that many people seem to think an optical viewfinder is the way to go for the new 'Digital Cinema' cameras, (is the Arri D20 the only one that currently has one?) but some people don't agree. I'm wondering what the advantages and disadvantages of each system are.
Edited by Josh Silfen, 29 December 2005 - 09:50 AM.
Posted 29 December 2005 - 10:02 AM
With an optical system you see in color, accurate focus and estimate the depth of field. No power required when setting up!
Posted 29 December 2005 - 10:08 AM
Also, with an optical system, you can see what is just out of frame allowing you to anticipate when an actor or object might enter frame.
Posted 29 December 2005 - 10:13 AM
Less so with people shooting S35 and S16 !
Posted 29 December 2005 - 11:19 AM
The Dalsa Origin also has an optical finder, an excellent one manufactured for them by P+S Technik.
Posted 29 December 2005 - 11:40 AM
It's just two different ways of shooting -- do you want to see the actual image coming from the lens, as clear as possible to see focus, etc. plus see off the edges of the frame? Or do you want to see the actual signal coming off of the chips, with actual exposure information? There are advantages to both. I almost wish it was an optical viewfinder with an electronic image that you could flip into view as needed.
Posted 29 December 2005 - 11:57 AM
So does the optical viewfinder work just like a film camera, with a mirrored spinning shutter, or is it some other system? If so, do cameras with electronic viewfinders have spinning shutters as well?
Posted 29 December 2005 - 12:02 PM
Yes for the first and maybe for the second. An optical viewfinder usually involves a spinning mirror to split the light from the sensor or film to the viewfinder. The other option is something like a prism, which is rare these days because it robs some light to the film and sends a dim image to the viewfinder.
A video camera usually does not have a physical shutter, just an electronic one -- although the frame transfer CCD design of the Viper (and I believe the Dalsa, maybe the D20???) require a short mechanical shutter to briefly cover the sensor so it can discharge.
Edited by David Mullen, 29 December 2005 - 12:02 PM.
Posted 30 December 2005 - 05:09 AM
By simply using a mechanical shutter to blank the chip during the frame readout period, the design of the CCD chip can be drastically simplified, allowing more silicon to be devoted to gathering photons and so giving better performance. It also completely eliminates the dreaded vertical smear.
Philips/BTS used this to great effect in their 3-chip cameras, although there seems to be a certain almost "superstition" about having a mechanical device attached to an electronic camera. Never mind all the whirring and clanking bits and pieces in the attached Betacam recorder!
Ironically, the D20 uses a CMOS sensor which uses direct readout of each pixel, (rather than shunting them out "Bucket Brigade" fashion) so it would be completely unaffected by vertical smear anyway.
Meanwhile the Genesis, which is, doesn't!
Edited by Jim Murdoch, 30 December 2005 - 05:12 AM.
Posted 30 December 2005 - 11:13 AM
Ah so thats what the problem was with the candle shots from the Genesis, in the D20/Dalsa/Genesis shoot-out!
Posted 30 December 2005 - 01:31 PM
I like the optical viewfinder on a film camera, but let's look at it from another angle. How would you set the aperture and_be_confident_in_the_aperture with an optical viewfinder on a D-20? How can you be sure? I have a Sekonic L-608, but have never found it helpful for determining f/stops with digital. It is an excellent tool for measuring FC and LUX, of course.
Posted 30 December 2005 - 01:44 PM
If you were to test a D-20, and establish your ASA I think you could expose using a meter. Its done every day with DSLR's.
Posted 30 December 2005 - 02:23 PM
I sure would like to test a D-20!
Posted 30 December 2005 - 04:57 PM
I asked Dalsa about this, and was laughed out of the room by the then-head of the ASC, thus forever lowering my estimation of that august organisation.
The idea was to have an optical viewfinder with a heads-up-display style reflective superimposition of a monochrome LCD to include exposure information (zebra stripes), framelines, and other camera data. I thought this would be the best of both worlds and offer a way to at least check highlight problems on a medium where that's particularly critical.
The idea has two major drawbacks:
- Dalsa didn't want to admit that their video camera was a video camera and had the same comparative dynamic range issues as any digital imaging device, despite the fact that this is unavoidably and self-evidently so;
- Steven Poster believes he can foresee and control specular highlights to within 1/1024 of lograithmic Dmax in three channels using a handheld light meter, a superlative feat which I invite him to prove at the earliest opportunity.
Ain't it a shame how politics wrecks good ideas?
Posted 01 January 2006 - 12:59 AM
Well yes and no/maybe?
See, the "straight" Frame Transfer technique as used by Dalsa dates back to the early 70s. A true Frame transfer chip has two almost identical blocks of CCD elements, one situated directly adjacent to the other like this:
The top set ("A"s) have a transparent covering and is what gathers the light, while the bottom set
("B"s) contains the readout circuitry and has an opaque cover.
During the frame blanking period (24 times per second) the contents of all the As get shunted down into the corresponding Bs, one CCD element at a time in a "Bucket brigade" fashion (in fact CCDs were also known as "Bucket Brigade Delay Lines" in the early days.
The As are then free to accumulate new charges while the Bs are having have their individual charges shunted out during the visible picture period to form the video signal. The advantage of using this chip design is that each CCD element also doubles as the readout device, so the whole piece of CCD "real estate" serves as a light gatherer.
However, an obvious problem is that the As will still be gathering photons as their contents are being shunted down into the Bs and that's what produces vertical smear. The only real cure with this type of chip is to use a mechanical shutter to block the light off during the readout period.
Most imagers these days use the so-called "Frame Interline Transfer" (FIT) technique, where the second set of CCD elements is mounted physically underneath under the main light gathering ones. During the frame readout period, it's more as if a whole lot of "trapdoors" open up at once and dump the accumulated charges onto the "floor below". Because the second set of elements are buried below the surface of the silicon, theoretically they're shielded from incident light and so there should be no vertical smear. (The disadvantage of the FIT design is that to do this, extra electronic elements have to be added to each CCD element, so less silicon is available for light gathering, a problem which is even worse with CMOS imagers).
In practice, silicon is slightly transparent, particularly to infrared, and so bright IR sources can penetrate the silicon and produce the familiar vertical lines with (usually) IR-rich incandescent light sources. (IR is more likely to get through the red optical filter, which is why vertical smear often tended to be red on earlier cameras.)
The only solution to this is to bury the readout CCDs deeper in the silicon, which makes the chips harder to manufacture. I would imagine that the Sony chip used in the Genesis is an FIT, otherwise the smear problem would be a lot worse. So in trying to make a chip with 12 million pixels that can all be accessed 24 times a second, something had to give!
Posted 01 January 2006 - 01:30 PM
Wouldn't a separate HD monitor and waveform monitor be a better way to determine exposure than an electronic viewfinder anyway? I don't know why anyone would want to set exposure using a light meter with a digital camera. It seems like an optical viewfinder is better for operators and a digital one would only appeal to a DP who is also operating, but even in that case it seems like it would be easy enough to use the monitor for exposure and enjoy the benefits of the optical system while operating. These cameras are not intended for ENG-style production where the operator is going to be racking the iris himself during the shot.