You mean you are relying on the meter to work right under dynamic, changing lighting and contrast conditions? LOL! You are too funny! 8^) But seriously, you're just experiencing the problem with ALL TTL metering schemes, the meter is not smart enough to know what you're doing nor what your photographic intent is. That whole B.S. about taking thousands of images and using them in the meter's program is all marketing B.S., they have been claiming this since before there were microprocessors capable of doing a comparison of even 1 or 2 scenes in the time it takes to you to press the shutter and the shutter to release! Pure marketing B.S., even with today's processors I sincerely doubt that there is enough memory and processor power in the camera to perform a comparison with more than 1-2 scenes in the time is takes for the shutter to release.
On the other hand all those sample images could be analyzed for patterns for a given set of meter segment readings and a generalized model could be developed considering contrast preferences, scene dynamic range, camera orientation etc. Then this information model could be programmed into the meter algorithm to produce a bias for the meter to use in conjunction with the selected AF point, camera orientation and photographers intent. But I don't think anyone does all this no matter what the marketing people say.
Hi Chip,
I'm quite sure they don't really compare the actual segment readings to those of thousands of images. This is marketing B.S., as you call it ;-) Probably they used those thousands of classified readings to feed them into some auto learning system, like a neural network. This network could be quite small and the resulting classifier would be fast and also easy to implement in either hard- or software. Of course it sounds much more impressive when they say "the camera compares the readings to thousands of images" or something alike ;-)
Thomas Bantel
* **** ******* *********************************************************** * For list instructions, including unsubscribe, see: * http://www.a1.nl/phomepag/markerink/eos_list.htm ***********************************************************
