gsutiono@alias.com (Guno Sutiono) (04/05/91)
We are looking for some information regarding the geometry of video camcorder optics, specifically field of view determination from physical camera settings. Any help would be appreciated... Q1. Is there a standard for CCD size specification in a video camera ? For instance, if my video camera has a 1/2" CCD, does this imply that there are specific dimensions for CCD size? If so, what are they ? In what standard dimensions are CCD's commonly available? Q2. Based on the pinhole camera model, the field of view (FOV) can be determined by the film size (e.g. 35mm film) and the associated lens' focal length (e.g. 50mm) by using the equation: FOV = 2 * arctan( 0.5 * film size / focal length ) When will this pinhole camera break down in reality ? Q3. In the case of video camera, can we just substitute the "film size" variable in the previous equation with the "CCD sensor size" ? Q4. The commercial CamCorders that are available nowadays use terminology such as 6X, 8X, or 10X to specify the power of the zoom factor they support. Is there an industrial standard method for mapping these zoom factors to range of focal length ? e.g. It seems that 8X implies an 8.5~68mm lens and 10X implies an 8~80mm lens in a 1/2" CCD video camcorder. Q5. Is the image captured from a video camera appearing on a TV monitor a cropped representation of what is on the CCD (specifically, is it possible for the optical centre of the image on a TV monitor to be offset in image space from the optical centre of the image falling on the CCD) ? If so, how can we determine the offset ? Are there any other transformations which are applied to the image falling on the CCD during its conversion to a video signal? Q6. Can we generalize the above approach for obtaining field of view to a movie camera ?
goldberg@oasys.dt.navy.mil (Mark Goldberg) (04/05/91)
In article <1991Apr4.200358.2223@alias.com> gsutiono@alias.com (Guno Sutiono) writes: >Q1. Is there a standard for CCD size specification in a video camera ? In what standard dimensions are CCD's commonly available? 1/2" and 2/3" CCDs are the defacto standard. Consumer and some compact pro cameras use 1/2". A few consumer units used 2/3" CCD (like the JVC GFS-1000 SVHS camcorder) but now 1/2" is the usual. CCDs are also described in terms of pixels (picture elements). The important figure of merit is EFFECTIVE PIXELS. That is because the image are that's used is really smaller than the total number on the chip. Video has a 3V:4H aspect ratio like super-8 and 16mm film. While I personally don't know that actual effective image dimensions for video camera CCDs, the 2/3" units basically use 16mm film optics and the 1/2" units are similar to super-8. I think the 1/2 or 2/3 inch refers to the image raster diagonal, similar to expression of TV screen size. >Q2/3. Based on the pinhole camera model, the field of view (FOV) can be determined by the film size (e.g. 35mm film) and the associated lens' focal length (e.g. 50mm) by using the equation: FOV = 2 * arctan( 0.5 * film size / focal length ) When will this pinhole camera break down in reality ? It basically holds. But you must know the effective image dimensions on the chip. You would need to have three equations: one each for vertical, horizontal, and diagonal fields of view based on those respective distances on the chip for an effective focal length. Realize that the true focal length or a fixed lens or zoom at a particular setting does vary slightly with focussed distance. When estimating the coverage of my lenses for a job, it's easier for me to ditch the trigonomentry and use simple ratio-proportion: raster height on CCD coverage height -------------------- = ----------------------- focal length cam-to-subj distance > >Q4. Is there an industrial standard method for mapping zoom factors to range of focal length? It seems that 8X implies an 8.5~68mm lens and 10X implies an 8~80mm lens in a 1/2" CCD video camcorder. Zoom factors (eg 8x, 10x, 16x, ...) express the focal length ratio maximum (at telephoto) to minimum (a widest setting). A particular zoom ratio DOES NOT imply a specific minimum f.l. For example, A Panasonic camcorder may have 8-64, while a Sony Hi-8 model has 11.5-95. Professional video lenses are designated without this ambiguity, like "J15x9.5" or "A12x10.5" meaning 15x ratio with 9.5 mm minimum focal length and 12x with 10.5mm minimum f.l. respectively. >Q5. Is the image captured from a video camera appearing on a TV monitor a cropped representation of what is on the CCD (specifically, is it possible for the optical centre of the image on a TV monitor to be offset in image space from the optical centre of the image falling on the CCD) ? If so, how can we determine the offset ? Are there any other transformations which are applied to the image falling on the CCD during its conversion to a video signal? TVs always have a certain degree of overscan - that is, a cerain amount of cropping occurs. That's why studio always place graphics and captions within the "safe title area." Some studio systems supimpose the safe lines in the image on their viewfinders and monitors to aid in proper composition. Professional monitors have an UNDERSCAN function which shrinks the scanning raster, enabling technicians to see the whole transmitted image. Since overscanning is not necessarily geometrically uniform, the true picture center may not be consistent when viewed at home or on a consumer VCR/monitor. Also realize that everything that the a sensing chip has a particular output vs. illumination curve. There is a saturation point on top and a pedestal on the bottom. A CCD is basically an analog sensor. As the signal is transmitted and processed, numerous alterations and distortions occur. IMHO, the greatest of these is the encoding or the luminance and chrominance into a composite video signal. That's why contemporary video production systems try to postpone this encoding as long as possible and work with a Y/C or Y/R-Y/B-Y COMPONENT signal. > >Q6. Can we generalize the above approach for obtaining field of view to a movie camera ? The geometric computation works, no matter what the medium. Even in film, some inevitable cropping will occur in the post production steps if optical printing is used. =========David Taylor Research Center (a US Navy lab) - Annapolis, MD========== /|/| /||)|/ /~_/\| |\|)[~|)/~_ | "Everyone's entitled to MY opinion." / | |/~||\|\ \_/\/|_|/|)[_|\\_/ | goldberg@oasys.dt.navy.mil "Poor is the man whose pleasure depends on the permission of another."-Madonna