> >Obviously once the flash has fired there's little time to do an "auto-colour-balance" on the fly. So if it's looking at ambient lighting before the flash ... > > I think maybe we both missed this, but I reread this again according > to your wish, and I'd like to say that AWB is something that is kind > of unknown. But, we do know the 1Ds does a nice job because it has a > separate special AWB sensor built in. The 10d does not and I would > guess uses some kind of software sensing on the overall image. So it > does do it on the fly, as well as the 1Ds, but just in a different > way. I think most digital cameras likely do it on the fly too. In > other words, it uses the data from the image just captured to decide > how to balance it before writing it to the memory card. Jim: bottom posted as usual this time ;o) I've visited a few technical pages just now - the trouble is in the digital world manufacturers don't share technical stuff with you only the sanitised "concept-ware" versions they think consumers might understand (remembering the technical manuals that came with my first camera 30-y ago ;o). FWIW. It depends! 1) If (and it's a very big IF) internally the WB settings adjusted the gains of the R, G and B sensor AD converters then (and only then) would the WB setting dialled in to the camera affect the full-bit-depth data initially transferred to the camera's buffer. I'm not sure any cameras actually do this: it's a possibility. If it were so in any model then possibly too post-processing of raw images may not be able to correct ... When it comes to auto-white-balance though: the decisions would have to be made at the time of capture.Of course, it is still almost impossible to have this work for flash (especially tailflash) because the WB could not be complete till the image was complete. I'm guessing it does not work that way. I'm guessing auto-WB happens totally post-capture. 2) Internally digital cameras ADCs produce 10, 12, ... 16(?) bits per sensor site. A possibility (caveat) occurs if the number of bits stored in the raw file was somehow less than that initially available to the camera. I doubt they would be that dumb. 3) If your raw file really does contain ALL the information from which the camera can create bitmaps (RGB - 8-bit-per channel) then post processing the raw files (as you do) MUST be able to produce better results and be less critical over white balance. Post-processing routines are better because they can afford to be slower - offset by the fact they can use much faster processors of your PC / Mac. I can't understand why anybody avoids the use of raw files - unless they are only taking holiday snaps - IMO ;o) Bon
