White Balance Question

[Luke_Miller]

Since I always shoot raw (or DNG with my M8.2) I normally set white balance to Auto and make any necessary corrections during post processing. The other evening I shot a family get together with my M8.2 and Nikon SB-25 flash. Since flash was to be the primary light source I set white balance to the Flash preset. During post processing in both Capture One and Lightroom I noticed that the as-shot Kelvin reading was changing across my exposures. Thinking that perhaps this was because I did not have a Leica dedicated flash mounted I did a test today with my SF-24D in TTL and got the same results. It appears that there is an Auto White Balance component even with the Flash WB preset. If I manually select 5600K that is what I get. Is this normal and I just did not get the memo?

[wstotler]

Originally Posted by Luke_Miller

Since I always shoot raw (or DNG with my M8.2) I normally set white balance to Auto and make any necessary corrections during post processing. The other evening I shot a family get together with my M8.2 and Nikon SB-25 flash. Since flash was to be the primary light source I set white balance to the Flash preset. During post processing in both Capture One and Lightroom I noticed that the as-shot Kelvin reading was changing across my exposures. Thinking that perhaps this was because I did not have a Leica dedicated flash mounted I did a test today with my SF-24D in TTL and got the same results. It appears that there is an Auto White Balance component even with the Flash WB preset. If I manually select 5600K that is what I get. Is this normal and I just did not get the memo?

Luke, interesting. AWB for the M8 on auto has certainly been an issue and returned varying results, as you've described. (There's thread after thread about it.)

You're right to think that setting the M8 to the Flash preset would eliminate the behavior. . . .

Can't test for this right now but I will later tonight and let you know, if nobody else pops in to confirm.

Thanks,
Will

[wstotler]

Luke,

I attached the flash via the SC-28 cord and tried some test shots.

On my M8 (classic, latest firmware) the Flash WB preset is not honored, as you discovered. E.g., setting the M8's WB to Flash has the same result as setting the camera to AWB.

-->Can another forum member or two please confirm this finding?

Luke: Are you also claiming that the Manual WB setting is not "sticky" either?

Thanks!
Will

[Luke_Miller]

Originally Posted by wstotler

Luke: Are you also claiming that the Manual WB setting is not "sticky" either?

When I manually set a specific kelvin temperature that number is honored. Lightroom will display 5600K as 5950K and +9 Tint. (Lightroom and ACR convert the kelvin temperature to values that display the correct colors.) Capture One will display it as 5616K

With Capture One V5.01 the M8.2s white balance preset of FLASH is correctly reported in the vendor area of the EXIF data. Notwithstanding that, the as-shot white balance varies when displaying the images. One example. Camera white balance set to Flash preset. Flash image taken under fluorescent lighting displays as-shot white balance of 3626 +3 Tint. Image has blue tint. Using the Capture One flash white balance preset (4778 -2) restores normal colors.

Lightroom 2.6 does not report the white balance preset in the EXIF and also varies the as-shot white balance when displaying the image just as Capture One does. Using the same image described above, Lightroom displays it with as-shot white balance of 4000K +20. Image has a blue tint like the Capture One version. Applying the Lightroom flash white balance preset (5500 +0) restores normal colors.

It seems as though both Lightroom 2.6 and Capture One 5.01 ignore the flash preset my M8.2 specifies in the EXIF. If I specify a kelvin number as the camera white balance, both programs appear to honor it.

[Luke_Miller]

More Information (I love a mystery)

I just did a series of shots using the Fluorescent WB preset. Both Lightroom and Capture One honored that setting. While I have not tested with the other WB presets I suspect (other than for Flash) I will see the same thing. I am guessing that my M8.2 is trying to do a white balance based on the combination of the flash illumination and ambient light when set to the Flash WB preset. My flash work has been at slow shutter speeds so ambient light is a factor in the images. I'll try a set where that is not the case and see if I get different results.

[wstotler]

Originally Posted by Luke_Miller

It seems as though both Lightroom 2.6 and Capture One 5.01 ignore the flash preset my M8.2 specifies in the EXIF. If I specify a kelvin number as the camera white balance, both programs appear to honor it.

Thanks for the additional info--manual WB settings are "sticky."

* Flash firing is written into the EXIF.

I'm figuring LR/C1/Aperture *shouldn't* automatically use a flash WB temperature just because the flash fired, so their behavior seems OK. (You might want to use AWB with flash--just 'cause the flash fired doesn't mean you want that temp. What if the flash is gelled and AWB is a better option than Flash WB?)

So the basic problem seems to be this:

* When Flash WB is set, the kelvin value for flash isn't written into the EXIF. You'd expect the M8 to do this?

Thanks,
Will

[wstotler]

Originally Posted by Luke_Miller

I am guessing that my M8.2 is trying to do a white balance based on the combination of the flash illumination and ambient light when set to the Flash WB preset.

And that would be different from what AWB does now . . . ? (I didn't see a difference when I was tinkering--doesn't mean there isn't one.)

Hate to do this, but I'm going to chuck this into the mix (from the manual):

Flash preset is "for pictures with (prevailing) electronic flash illumination" -- that plays to your mixing hypothesis. How it's different from AWB I can't tell.

From a corresponding NOTE: "When using an electronic flash unit that satisfies the technical requirements of the System Camera Adaption (SCA) for the System 3000 and an SCA-3502 adaptor (from version 4 onwards), the automatic white balance setting (A) will render the correct colors. If however other flash units are used, which are not specially dedicated to the LEICA M8, the camera white balance will not be automatically adjusted, and the setting should be used."

Thanks!
Will

[Luke_Miller]

I just did a set of images using my SF-24D and M8.2 white balance set to the Flash preset. Shutter speed was 1/180, aperture f16, ISO 160. If I shoot in a room where the flash is the only source of illumination I get a very consistant Capture One as-shot WB reading of 5906 +1. When I shoot in other areas where ambient becomes somewhat of a factor I see the as-shot WB change. It looks to me that some sort of auto white balance is taking place in camera with the Flash preset. At this point I think that if I want a consistant flash white balance I will have to use a manual kelvin setting.

Since Auto White Balance automatically adjusts for flash I would think the Flash preset would produce a single WB value. Unfortunately in my limited test the automatic change that occurs away from the white balance of 5906K +1 produces an image that requires correction back to 5906K +1.

[wda]

I thought that when shooting raw there is no influence of WB in camera. That is why LR2.6 doesn't show it in EXIF. The raw file allows maximum control in post processing.

[ho_co]

Luke, Will--
My reading of the manual is that
1) when a "system-compatible" ('dedicated') flash is used, the flash automatically sets the camera to "Flash" preset;
2) from this I assume that "Flash" preset means a given Kelvin temperature is written to the file;
3) and this color temperature would be read by the RAW converter.

Clearly, that's not happening here. Why should there be a manual "flash" preset available if it doesn't adjust the image to a standard? That is, what good is "flash" preset if it is merely another version of AWB?

Two lines of attack:
1) Try jpg+DNG. Maybe the "flash" preset is active only in jpg?
2) Is your lens coded? Are you using "Lens ID on"? P 86 of the manual says that "Flash exposure and flash reflector control also uses the lens data" and that "When using a lens without 6-bit coding, the recognition function of the camera must always be turned off, to prevent malfunctions."

Both those questions are just for completeness' sake. There should IMHO be no difference between JPG and DNG, and there should be no automatic white balance going on if you've made any preset.


David--
Technically, you're right. White balance is completely adjustable in RAW. But I think the camera writes its idea of what the white balance should be to the file, simply as a starting point for the RAW converter.

[Luke_Miller]

Originally Posted by ho_co

Two lines of attack:
1) Try jpg+DNG. Maybe the "flash" preset is active only in jpg?
2) Is your lens coded? Are you using "Lens ID on"? P 86 of the manual says that "Flash exposure and flash reflector control also uses the lens data" and that "When using a lens without 6-bit coding, the recognition function of the camera must always be turned off, to prevent malfunctions."

Both those questions are just for completeness' sake. There should IMHO be no difference between JPG and DNG, and there should be no automatic white balance going on if you've made any preset.


David--
Technically, you're right. White balance is completely adjustable in RAW. But I think the camera writes its idea of what the white balance should be to the file, simply as a starting point for the RAW converter.

Great questions, Howard.

The lens is 6-bit coded and Lens Detection is On.

Just shot three DNG + Jpeg images with my SF-24D mounted and WB preset set to Flash. Capture One V5.01 reports all three jpegs as having an as-shot white balance of 5000K 0 Tint. The corresponding DNGs report out at 4400K -2, 5721K 0, and 5427K 1. In each case the DNG white balance is closer to accurate than the 5000K 0 in the jpegs.

So it appears some auto adjustment of the Flash white balance is occurring in the DNGs, but not the jpeg versions. Curious.

[Luke_Miller]

Originally Posted by fotografr

You can avoid the whole issue by simply putting a filter over the flash to warm it to a tungsten balance, then setting your camera WB to Tungsten/Incandescent. Then you don't end up with a daylight balanced forground and an orange background. With fluorescent lighting, you would of course put a green filter over the flash and set the camera WB to Fluorescent.

I agree, Brent.

I do this with my Nikon DSLRs and SB-800s. As a digital M newbe I am just trying to understand how it works. It seems to work differently from my Nikons, which is OK as long as I understand the differences.

[fotografr]

Originally Posted by Luke_Miller

I agree, Brent.

I do this with my Nikon DSLRs and SB-800s. As a digital M newbe I am just trying to understand how it works. It seems to work differently from my Nikons, which is OK as long as I understand the differences.

Sorry for pulling my post. I did it because my response wasn't really an answer to your question. If you're interested in trying this out with your Leica, Nikon makes a set of filters that fit quite nicely over the Leica flash heads. I use them all the time.

4764 Nikon SJ-1 Color Filter Set, Eight Filter Light Balancing Set for the SB-800, SB-600 & SB-80DX Flashes

[ho_co]

Luke--
Very interesting and very peculiar thing you've discovered. Haven't heard of it before. Sounds like a bug to me.

If it's supposed to work this way, why do DNG and JPG work differently?

If it's supposed to work this way, I'm sure Leica has a good explanation as to why so.

And I hope they bother to tell us.

Thanks for the heads-up!

[scott kirkpatrick]

Haven't used flash much, but I remember the discussions when the AWB was upgraded to its present form. My understanding of the impact of the white balance is that it is passed to the raw file development program as data that expresses how a neutral grey will look in the light that the present scene is experiencing. But no correction is applied to the raw data, only to a jpeg computed at the same time. So to sort this out, there may be a contribution to the confusion coming from the different ways in which different programs (LR, C1, etc) expose the information.

Where the shots using flash lit only by the flash, or was there mixed lighting?

scott

[Luke_Miller]

Originally Posted by scott kirkpatrick

Where the shots using flash lit only by the flash, or was there mixed lighting?

scott

Both. In the case where little to no ambient light was present the DNG as-shot white balance reported out became a consistant value. That value changed as ambient was added to the mix.

[Luke_Miller]

I did another series with my M8.2 (firmware 2.04) and SF-24D. I shot a ColorChecker card illuminated by a tungsten lamp. Camera was mounted on a tripod to eliminate any variable due to different framing. Shots were in DNG + Jpeg Fine and processed in Capture One V5.01. I shot a series in TTL flash and manual flash with camera white balance set to Auto, Flash, and 5400K. Results were interesting.

All jpegs are reported by Capture One as having a white balance of 5000K regardless of the camera white balance setting. So Auto, Flash, and a 5400K Kelvin setting all produced jpegs at 5000K (as per Capture One.) And they are clearly cooler than their DNG counterparts, which are reported about 500K warmer. Correction - Just noticed that Capture One reports all M8.2 jpeg as-shot white balance as 5000K regardless of actual white balance.

With DNG all images taken with the 5400K setting were reported by Capture One as 5418K. Images taken with Auto white balance and Flash white balance had essentially identical white balances as reported by Capture One.

My conclusions so far:

If you shoot DNGs the Flash preset produces the same result as Auto. If you want other than Auto white balance with flash you must set a specific value in the Kelvin menu.

None of this should be taken as a criticism of how the M8.2 functions. Nor am I suggesting a "bug" in its firmware. As a long time Nikon digital shooter I was trying to determine if my M8.2 differed from my Nikons in how white balance was handled. It does and now I have a better understanding of how to use it.

[ho_co]

Luke, it's great that you're doing this research. What you're discovering is quite interesting.

When I said above that one might see this behavior as a bug, I meant only that "it doesn't work like the others." As you said, that's neither good nor bad in itself; one simply has to know one's tools. Whether it's a bug depends on whether Leica intended for the white balance to work this way or not.

Thanks for introducing me to aspects of the camera I was unaware of!

[wstotler]

Luke,

Thanks for the additional follow-up work! Very handy to know, proof-positive.

I've E-mailed Leica (Solms) about the M8's Flash WB behavior vs. AWB behavior and directed them to this thread for reference.

Hopefully they will comment here or get back to me directly, in which case I'll post their answer.

Thanks!
Will