Originally posted by asp1880 Leave highlight protection on in good light. Turn it off in low light, as it increases noise. That's not a problem at low iso but it might be at high iso.
Anyway, that's what I do.
Regards,
--Anders.
This is true, but much less so for the K-5 and other cameras with the 16 MP Sony APS-C sensor than for other cameras. That is why the K-5 has such tremendous dynamic range at low ISO:s - the read noise is low, and therefore, the deep shadow information stays intact when low ISO settings are used. When a picture is taken at low sensitivity (low ISO), underexposed, and later pushed in post processing, the read noise is of course also amplified, whereas if a higher ISO is used, the amplification takes place in the sensor circuitry, before the read noise is added. With other sensors, that have high read noise, the amplified read noise in a pushed low-ISO image would make it worse than the same image taken at a higher ISO. This is of course also true for the K-5, but the effect is very minimal. And if ISO:s (without highlight protection) over 1600 are used, it is a digitally amplified ISO 1600 image anyways, since the analogue circuitry of the sensor is limited to 1600 (as far as I know, all sensors have a similar limit).
In other words, for ISO:s higher than 1600 (3200 with highlight protection turned on), the camera just amplifies and throws away information. What it should do is of course to save an under-exposed ISO 1600 file and just have a variable in it that tells the raw converter how much to increase the exposure by default. This should give the exact same result (with the bonus of getting rid of the built-in raw noise reduction above ISO1600), but with much more highlight headroom in the raw file. I don't understand why camera manufacturers don't do it that way.
Actually, I think there should be an option of the camera always saving images at the lowest ISO, and with a variable in the raw file telling how much underexposed it is, instead of using the analogue amplification (since there is so little to gain from it with the K-5). In other words, if I take an ISO800 picture, the camera exposes correctly for ISO800, but the sensor is set to ISO80, and then there is a variable in the raw-file that tells the raw program that the picture is 3.5 (approximately) steps underexposed. The jpeg in the case of raw+jpeg shooting should of course get this amplification included in the processing too, so that it looks right. This image-mode would lead to high-ISO images with almost the same quality as normal ones, but tremendous highlight headroom. Of course this is possible to achieve by setting the camera to the lowest ISO, and then using exposure compensation instead of higher ISO, for later push in post processing. However, such an approach has several draw-backs: very dark preview images in camera, no usable jpegs out of the camera, the push in post processing has to be done manually and so on. I would like a firmware update with this option included, so that each photographer can chose if s/he prefers the lowest possible noise at higher ISO (this is the only option available now) or always keeping the entire dynamic range of the camera, and only shifting it into the highlights when higher ISO:s are used.