Unscientific but interesting: AR7, RX1, D800, M240

[Mark Gowin]

Tim, I have always had the same assumption as you in that the camera white balance setting didn't matter if shooting raw and white balance is corrected in the raw conversion software. Your preliminary test seem to bear this out. My own experience has been similar as I have never had any problem correcting white balance to my satisfaction in post (except when using a Lee Big Stopper filter). I am far from an expert in these matters so what is acceptable to me may not be acceptable for someone else.

I admit that Jono's comment, or rather my understanding of his comment, about the camera WB setting causing a bias for the raw editing software sure got me thinking about the possibility he is right. I appreciate your efforts in testing this out. I don't have an opportunity to do the testing myself right now.

 

[tashley]

And just to round out my tour d'horizon of the fascinating subject of White Balance when shooting RAW, I shot a series under very poor tungsten light, with the camera set to each of the main in-camera WB options. I then set them all in LR 5.3rc to the same value, which was determining by clicking the dropper on the relevant WB calibration target in the first image and pasting it to all the shots… and the exposure as calculated by the camera remained identical for each frame. So in future, I will merrily shoot with whatever WB happens to be set if I'm in too much of a hurry, in the reasonable assurance that it makes no eventual difference - though just to be sure I will see what happens with shading correction turned on when I get a native FE lens…

In the same order as above:

 

[pozzello]

Oh tsk, this isn't about images!

Battersea Power station, London...

Power stations and incinerators are all I can think of these days - thanks for the tip

 

[Taylor Sherman]

OK, so there are two questions here it sounds like.

1. Does the RAW image change depending on what white balance you select in-camera?
2. Does the application of a particular WB in post, to a RAW image, change depending on what white-balance you select in camera

For 1: generally I've always assumed the answer is "no", the RAW does not change. However, what I don't know is how the camera meters - does it meter purely based on the raw data, or does it apply the WB, and meter from that? If it does the latter, you will get different exposures. And, different exposures will slightly affect the eventual color balance (though I don't know in what, if any consistent, direction).

For 2: I don't believe so. LR's application of "Daylight" to the RAW file is a fixed, mathematical transformation. IF the RAW pixel data is actually the same, then the output is the same no matter the camera's WB setting that is embedded in the RAW.

For practical purposes, I'm operating on the assumption that medium-sized WB issues (eg using Cloudy instead of Daylight, or using AWB) will have no visible effect on the end result I am capable of getting by applying the "proper" WB to the RAW.

Shooting with Tungsten in daylight, or using a crazy custom WB (both of which I used to do a lot of, actually, but haven't for a while) could possibly cause under- or over-exposures that you wouldn't have gotten before (but again, I'm not sure of that, it's just to be on the safe side).

 

[tashley]

Hi Taylor,

The two sets of tests I did both involved frames that were exposed as being at the far end of the 'spectrum of normal possibility' from the actual light in the scene and yet all the balanced files in both sets metered and converted exactly the same as far as I can see - so I think one will likely be safe even with big disparities between the WB as set on the camera and the actual temperature of the light. At least on this camera… but I am curious to bottom out if there are cameras or circumstances where that comforting rule of thumb will lead one into trouble...

 

[tashley]

Power stations and incinerators are all I can think of these days - thanks for the tip

The one in the photo has appeared on a Pink Floyd album cover and is a great iconic London building but it you like it, hop a plane quickly: they are finally. after many years of argument, about to re-develop it, albeit within the existing shell...

 

[Floyd Davidson]

For 1: generally I've always assumed the answer is "no", the RAW does not change. However, what I don't know is how the camera meters - does it meter purely based on the raw data, or does it apply the WB, and meter from that?

I can't speak to all cameras, but specifically the Nikon D800 (as do all Nikon DSLR's) uses and entirely different sensor for metering, and it is never adjusted for White Balance or any of the other configuration options for producing a JPEG image.

I think it is safe to assume that is true for other DSLR cameras by other manufacturers.

 

[peterv]

Tim, first off, thanks for sharing your tests.

I think the ‘confusion’ about the WB stuff comes from this:
If you use AWB, even on the same location in the same light, your colours are going to differ from shot to shot because the camera calculates the AWB from what it ‘sees’ in each specific shot.
Now if you start post-processing these shots and apply a new AWB in LR on the each shot in my experience one does not get a consistent WB-look for that shoot.

I’m not completely sure, but I think that Jono feels that if you apply a custom WB in LR to the first shot you took with AWB on a specific location under more or less the same light and than ‘sync’ (apply) this custom WB to all other shots, your WB is going to vary from shot to shot, as if LR only applies the correction value, which was ‘right’ for the first shot you corrected, but should be a different value for all the other shots, because they have different AWB measurings to begin with.

On a side note, I know you mentioned the landscape shots comparing the Sony’s, Leica and Nikon, are not scientific and all, but it looks to me the D800 is focussed on the foreground. So that would explain the shockingly big difference in resolution between the 7R and the D800.

 

[tashley]

Thanks Peter!

The D800E file was taken with magnified LV focus: if I were being scientific I would have bracketed (as I would have done were it a shot that mattered!) but in the unscientific spirit I thought I'd do a 'best efforts' with the systems as they stand - including the terrible magnified LV of the D800, which is often still preferable to the dodgy AF of that camera… but I wouldn't assume that it is focussed on the foreground because the distance mark was close to infinity, though I guess I could have jogged it with my hand as I held it to shoot. Maybe I'll try that one again - but the field of focus in that frame is odd: the extreme foreground is in focus, then some of the mid ground is OOF and then the distance looks like it is more or less in focus but without great resolution. It is s tricky lens to be sure so I will get it on a tripod and shoot it with focus done a variety of ways and report back… and in any event its soggy edges are often a problem, due to field curvature.


Per the White Balance: in the bridge series above, I took all the frames as shot at their widely differing in-camera WB settings, clicked the BW eyedropper on the reference patch of the White Balance frame and then copied that via Sync to all the others, and indeed each frame showed the same Temp and Tint and looked the same.

So it seems to me that LR applies an absolute value in Sync, not a relative shift.

I don't use AWB in LR often enough to have a view but I did just try it on those files and it did give each one, however shot, exactly the same Temp and Tint. However I have yet to shoot a series using AWB and then ask LR to use its own AWB on each of the series to see what happens!

 

[Taylor Sherman]

Tim, first off, thanks for sharing your tests.

I think the ‘confusion’ about the WB stuff comes from this:
If you use AWB, even on the same location in the same light, your colours are going to differ from shot to shot because the camera calculates the AWB from what it ‘sees’ in each specific shot.
Now if you start post-processing these shots and apply a new AWB in LR on the each shot in my experience one does not get a consistent WB-look for that shoot.

I’m not completely sure, but I think that Jono feels that if you apply a custom WB in LR to the first shot you took with AWB on a specific location under more or less the same light and than ‘sync’ (apply) this custom WB to all other shots, your WB is going to vary from shot to shot, as if LR only applies the correction value, which was ‘right’ for the first shot you corrected, but should be a different value for all the other shots, because they have different AWB measurings to begin with.

I'm pretty sure that's not the case. If all the shots really do require basically the same WB (EG, it is a proper application to create one WB in LR on the first shot then apply it to all of them), then it doesn't matter what the camera-applied WB was. At all. The WB is just applying to the RAW images, and ignoring the camera's WB setting that's embedded in there.

In other words, LR is not applying a "correction" to the in-camera WB. It is applying the full WB algorithm to the original sensor detections.

 

[cunim]

Peter, don't judge it until you see it shot with a native FE lens and a better raw developer... If you want close to MF quality in a briefcase, the a7r and the Rx1 are the only games in town IMHO.

Thanks, Tim. Got the camera yesterday with the 35 FE. Putzed about a bit, determined to be unscientific as per the topic. The 35 will probably go back because it is very nice, but I am a 50 sort of guy. Appears that my Biogon 35/2 beats the FE at center but then turns to utter rot as you go outwards. I guess this is the sort of thing many others have observed. My own theory is that microlenses are a two-edged sword. If the ray paths hit them appropriately, they work well. If not (rays either too obtuse or too straight) the mircolenses degrade the image. IOW, telecentricity may not be an entirely good thing. If this theory is valid, the chip will be temperamental.

What gobsmacked me was a tired old mid '60s Summicron R f2 - sort of like watching grannie enter a break dance contest.

Here's an image of the front yard, with a crop from the top right corner. Just your recommended sharpening and clarity 20. What's neat is that the ISO is 1000.

 

[tashley]

Thanks, Tim. Got the camera yesterday with the 35 FE. Putzed about a bit, determined to be unscientific as per the topic. The 35 will probably go back because it is very nice, but I am a 50 sort of guy. Appears that my Biogon 35/2 beats the FE at center but then turns to utter rot as you go outwards. I guess this is the sort of thing many others have observed. My own theory is that microlenses are a two-edged sword. If the ray paths hit them appropriately, they work well. If not (rays either too obtuse or too straight) the mircolenses degrade the image. IOW, telecentricity may not be an entirely good thing. If this theory is valid, the chip will be temperamental.

What gobsmacked me was a tired old mid '60s Summicron R f2 - sort of like watching grannie enter a break dance contest.

Here's an image of the front yard, with a crop from the top right corner. Just your recommended sharpening and clarity 20. What's neat is that the ISO is 1000.

Man can your granny shake it!

 

[fotografz]

I can't speak to all cameras, but specifically the Nikon D800 (as do all Nikon DSLR's) uses and entirely different sensor for metering, and it is never adjusted for White Balance or any of the other configuration options for producing a JPEG image.

I think it is safe to assume that is true for other DSLR cameras by other manufacturers.

I think that is right.

What can be tricky is what an overly warm, lower light image that appears well exposed does when you WB in post. Suddenly it looks under-exposed, sometimes badly under.

- Marc

 

[Vivek]

I can't speak to all cameras, but specifically the Nikon D800 (as do all Nikon DSLR's) uses and entirely different sensor for metering, and it is never adjusted for White Balance or any of the other configuration options for producing a JPEG image.

I think it is safe to assume that is true for other DSLR cameras by other manufacturers.

It is also true for the M in the "classic mode".

 

[peterv]

I'm pretty sure that's not the case. If all the shots really do require basically the same WB (EG, it is a proper application to create one WB in LR on the first shot then apply it to all of them), then it doesn't matter what the camera-applied WB was. At all. The WB is just applying to the RAW images, and ignoring the camera's WB setting that's embedded in there.

In other words, LR is not applying a "correction" to the in-camera WB. It is applying the full WB algorithm to the original sensor detections.

I’m sorry if my post created confusion, but here’s what I see happening in LR:

If I’ve shot on one location in consistent light with AWB in the camera the colours (WB-values) are going to be a bit different from shot to shot. Now when I use the white balance selector tool on one image and than sync other images, the numbers of the WB of the image I used the eyedropper tool on, are applied to all other images I’ve selected, the value stays the same.

When I apply an auto white balance in LR and than sync all the shots on that location, the colors (WB-values) are going to be different from shot to shot. LR will come up with somewhat different values, although the shots are taken in very similar light circumstances. Just like the AWB in a camera is not consistent.

 

[Floyd Davidson]

When I apply an auto white balance in LR and than sync all the shots on that location, the colors (WB-values) are going to be different from shot to shot. LR will come up with somewhat different values, although the shots are taken in very similar light circumstances. Just like the AWB in a camera is not consistent.

Isn't that just about exactly what should be expected?

Essentially what an auto white balance function does, whether it is in the camera, in a raw converter or in an image editor is the same. It basically blurs the entire picture to such a degree that it becomes all one color, and then it adjusts the RGB channel multipliers to make that color neutral gray. That is only slightly over simpllified...

The over all effect is that if the scene is illuminated by a more reddish light, the red channel will be reduced and the blue channel increase. But, alas the same thing will happen if 75% of the image is a solid read wall! Ouch.

The practical effect is that every time the camera is framed even slightly differently the AWB (in the camera or in your software) is going to see a slightly different average color, and will generate a slightly different set of channel multipliers.

Incidentally, none of them are "correct". The "correct" WB is whatever you, the photographer, happen to decide to use. And that is true even if the next person thinks you are color blind for having picked it. It's your picture and your choice.

 

[Jack]

When shooting in RAW it makes absolutely no difference at all. None. Zilch. Zip.

The raw sensor data is exactly the same regardless of how the White Balance is set.

To set the record completely straight: The raw data is the same regardless, but your histogram is not and associated exposure may be off if you impart a large WB change after the fact. So you can end up with some channels going from in bounds to completely clipped with a large WB change in post.