>> Is chromatic metering bias on bodies without RGB meters, like >> the N70, a relatively constant factor? > Modern sensors are fairly linear -- I've never had any > problems using colored filters with black & white. Sorry, the sensors used in cameras do not have a uniform spectral response (even with filters applied to the sensor). What matters is the difference between the sensors response and the film response. BUT the difference for B/W film should be well within the exposure latitude of the film. As has been extensively discussed recently, if you are using colour slide film then you may get an error. >> Does a deep red filter (I've used both a Nikon R60 and Tiffen >> #25) throw off the N70 AF sensor due to a similar chromatic >> bias? > That's not a problem with the passive AF systems that most SLR's > (all Nikon SLRs) use. The camera focuses on the same range of > wavelengths that expose the film. I am not convinced that this is always true in all situations. There is plenty of evidence that the AF system will work with the extreme red filters often used for IR films (with an autofocus assistance light). But most film is not responsive to these wavelengths. If this is the case (and not due to another effect), then in some unusual situations you might get a spectrally induced bias. BUT this will not be due to your filtering (using normal photographic filters), but rather to an unusual scene spectral distribution such as molten steel in which case I would be bracketing in any case. > The AF system needs a lot of light to work with a dark red filter. > Adding a 3x filter factor to your f1.8 lens leaves you with the > same amount of light as an f5 aperture -- only 1/3 stop from the > AF system's recommended limit of f5.6. That's slow enough to make > AF marginal in dim light. This is wrong (in the details). The f/# limit of an autofocus system is set by the design of the wedge prisms used within the system. The faster the limit is set the better the focus that can be obtained, but the smaller the range of lenses that can be used. The choice of f/5.6 by Nikon (and many others) is a consequence (or cause :-) ) of the speed range of the "consumer" zooms. It is quite possible to design a system that works at f/11 - but it would be less accurate even if you were actually using a faster lens. There are two limits related to light level. Fundamentally there are intensity limits. If the light level is very low there is too much relative noise (signal to noise problem). If there is too much light the sensor will saturate. There may also be electronic limits due to the presence (if used) of A/D converters, automatic gain controls etc. Using EV terms, the lower intensity limit is around EV 0 (ISO100) and the upper intensity limit is around EV19 (ISO100). The other limit is on contrast. If the scene does not contain sufficient contrast at a spatial frequency range matched to the sensors used in the AF system, then it will not be possible to determine if the image is in focus. This is related to the "beer bottle effect" - the bottom of a beer bottle gives a perfect image of a large blank wall (totally useless, but true)! One of the compromises of having a large area AF sensor is that you do not know on which part of the sensor the system will select to focus (I assume it is biased towards the centre, but I don't know). If you read the manual for the N/F70 you will find that it is possible to set it to "spot" auto-focus and I would suggest that you try this in similar situations (although with the filter/contrast level you are using you may find that the light levels are too low) - This is what I use when taking candid snapshots of my children playing in the garden. Andy Shaw
