RAW & WhiteBalance setting

[32BT]

Suppose the WB setting of the camera affects the analogue part of the (per channel) A/D conversion? That would suggest that it does make sense to set the WB even for RAW shooting. I presume the WB setting is totally irrelevant, but is it really?

[digitaldog]

Suppose the WB setting of the camera affects the analogue part of the (per channel) A/D conversion? That would suggest that it does make sense to set the WB even for RAW shooting. I presume the WB setting is totally irrelevant, but is it really?
[a href=\"index.php?act=findpost&pid=74202\"][{POST_SNAPBACK}][/a]

With respect to the RAW file, everything I've read or been told is it has no effect on that RAW data. It's simply EXIF data (a suggestion for rendering with respect to WB if you will).

[32BT]

With respect to the RAW file, everything I've read or been told is it has no effect on that RAW data. It's simply EXIF data (a suggestion for rendering with respect to WB if you will).
[a href=\"index.php?act=findpost&pid=74218\"][{POST_SNAPBACK}][/a]

Yes, and I don't have evidence on the contrary, but I was wondering out loud... I mean do we really know? Remember the allegations if you will about Canon RAW being not so RAW? I also recall some Nikon D200s suffering from an amplifier issue.

And, if at all possible, perhaps it is an interesting idea: adjusting the ISO for channels separately would possibly help WB response?

[John Sheehy]

Yes, and I don't have evidence on the contrary, but I was wondering out loud... I mean do we really know? Remember the allegations if you will about Canon RAW being not so RAW?

Canon's RAW files are about as RAW as they get, but there are a few modifications in certain models in certain situations.  Maybe the earliest Canon RAWs were different.

I also recall some Nikon D200s suffering from an amplifier issue.

And, if at all possible, perhaps it is an interesting idea: adjusting the ISO for channels separately would possibly help WB response?
[a href=\"index.php?act=findpost&pid=74238\"][{POST_SNAPBACK}][/a]

Perhaps, but unless each color had it's own read lines, there would probably more noise from switching the amplification between pixels.  Few, if any CFA-based cameras are using independent gain.  The D2X is alleged to, but I haven't seen the proof.

[32BT]

Perhaps, but unless each color had it's own read lines, there would probably [be] more noise from switching the amplification between pixels. 

Right, but that would come down to the age old discussion of;

what's better:
1 - higher iso but correct exposure, or
2 - lower iso but underexposure

[jule]

Suppose the WB setting of the camera affects the analogue part of the (per channel) A/D conversion? That would suggest that it does make sense to set the WB even for RAW shooting. I presume the WB setting is totally irrelevant, but is it really?
[a href=\"index.php?act=findpost&pid=74202\"][{POST_SNAPBACK}][/a]

Why not do some practical tests with the WB set at extremes to determine any differences?

[jule]

Right, but that would come down to the age old discussion of;

what's better:
1 - higher iso but correct exposure, or
2 - lower iso but underexposure
[a href=\"index.php?act=findpost&pid=74249\"][{POST_SNAPBACK}][/a]

Once again...Why not do some practical experiments with your equipment ?...and then scrutinise the results to see which is your preference.
Julie

[32BT]

Why not do some practical tests with the WB set at extremes to determine any differences?
[a href=\"index.php?act=findpost&pid=74408\"][{POST_SNAPBACK}][/a]

Well, that obviously was the first thing I tried, but like I said, I didn't find evidence of relevant differences. The whole thing came up when I litteraly dreamed about a completely customizable and controllable RAW converter. It exhibited controls for every minute step in the process, including a really interesting looking, equalizer-like control board to adjust the channel-cross-over parameters.

[michael]

"what's better:
1 - higher iso but correct exposure, or
2 - lower iso but underexposure"

When I first discussed the topic of "exposing to the right" with Thomas Knoll several years ago, he made the point that exposure at the camera's native ISO (usually the lowest) and with desirable bright areas just below clipping, produces the cleanest results.

The rest is just amplification.

Michael

[John Sheehy]

Right, but that would come down to the age old discussion of;

what's better:
1 - higher iso but correct exposure, or
2 - lower iso but underexposure
[a href=\"index.php?act=findpost&pid=74249\"][{POST_SNAPBACK}][/a]

If the weaker channel were the only issue, then the extra noise may not be enough to offset the gain, but you're going to get switching noise at every pixel in the channel that was already stronger, making it weaker in terms of S/N.

[John Sheehy]

"what's better:
1 - higher iso but correct exposure, or
2 - lower iso but underexposure"

When I first discussed the topic of "exposing to the right" with Thomas Knoll several years ago, he made the point that exposure at the camera's native ISO (usually the lowest) and with desirable bright areas just below clipping, produces the cleanest results.

The rest is just amplification.
[a href=\"index.php?act=findpost&pid=74430\"][{POST_SNAPBACK}][/a]

That's a given, but that's not really a trade-off; it's an ideal situation.

The lowest ISO is only going to give less noise when it is fully exposed.

Unless you're shooting architecture with a tripod, there is often a set of unavoidable compromises.  The trade-off between ISO and exposure varies with cameras, and different concerns apply to black noise and midtone and highlight noise (the former is affected mainly by readout and digitization processes, the latter by the literal sensor exposure in photons).

"Several years ago", there was a lot more gain-related readout noise than in current cameras.  The recent Canons have readout noise at ISO 1600 that is already weaker than shot noise with exposures 9 stops below RAW saturation.  If it weren't for the patterned line component of the noise, it would be even less visible.