5D Mark III Custom WB is 550K too warm an +10 too Magenta

[JimmysPhoto]

TELL ME WHY!

Custom WB test between 5D Mark II, 5D Mark III:
- Same studio strobe light
- Same Exposure/ISO
- Same Lens
- Same CF Card (swapped between bodies)
-Shooting RAW images
-ZERO processing in ACR or Lr

Essentially, all I am changing is the body. I manually meter, shoot with an Expodisc, perform a custom WB, and repeat to make certain that my exposure and WB are as precise as possible. Trade the lens with expo disc still attached, swap the CF card and do the exact same thing.

Results: WB "As Shot"
- 5D Mark II: Temp 5500 Tint -6
- 5D Mark III: Temp 6050 Tint +4

That's a variance of +550 kelvin and +10 Tint variance!
(I have RAW files if anyone wants to download & evaluate)
I called Canon Professional Services and spoke to a Tech- they argued that the sensor is so new & different that this is normal. I said that color temperature is NOT subjective (and noted that my 1DS Mark II Color Temp is almost the same as the Mark II) and that I had performed the above tests leaving no room for error. They still argued that this is normal, cited all the new sensor capabilities... but said that they would test the body if I wanted to send it in.

I hope others read this and add their test results because I feel like I am alone in this issue.

 

[Drizzt321]

You are correct that color temperature is not subjective. But you're missing the point that a different sensor will render the colors a bit different, especially if it is from a different line of sensors. Like the T3i & 60D & 7D all use more or less the same sensor, so the color rendering should be fairly similar. But the 5d2 and 5d3 sensors are actually different.

Now, do the colors look exactly the same when you apply the 2 different white balance settings? Remember, the RAW file is the data the sensor collects. You then need to add a white-balance to correct what the sensor sees, which means even when you 'view' it without any processing, you're really either viewing the embedded preview JPG, or the RAW viewers processing with the built-in settings based on what the camera said the WB should be.

 

[JimmysPhoto]

Thank you for your reply!

I think your missing my key point: Adobe Camera, Lr (essentially the same engine) with WB set to "As Shot" to view the in-camera set custom WB RAW files: So there is no parameter other than the cameras data to tell the program WHAT the WB is. I don't think the variance should be this big. 50-150K I might accept, but 550 and +10 tint is excessive to me.

 

[JimmysPhoto]

...and YES, skin tones look really warm. Add 550k to any portrait and tell me what YOU see.

 

[Drizzt321]

Thank you for your reply!

I think your missing my key point: Adobe Camera, Lr (essentially the same engine) with WB set to "As Shot" to view the in-camera set custom WB RAW files: So there is no parameter other than the cameras data to tell the program WHAT the WB is. I don't think the variance should be this big. 50-150K I might accept, but 550 and +10 tint is excessive to me.

What does LR/ACR do when you use the WB dropper? Is the correction the same value or different? I'll try and remember to do a similar test as you this evening after I go home, I have 5d2 & 5d3.

 

[JimmysPhoto]

GREAT question!

If I click as close to the center in each image as possible, there is no change to the Temp in either file and in the 5DM2 there is a slight +3 change to the Tint, so essentially Adobe is seeing it accurately as well.

 

[Drizzt321]

GREAT question!

If I click as close to the center in each image as possible, there is no change to the Temp in either file and in the 5DM2 there is a slight +3 change to the Tint, so essentially Adobe is seeing it accurately as well.

Interesting, so either Adobe is doing an initial WB off of what the camera says, and then adjusts from there, or it's engine sees about the same values from the RAW data. I'll see what I can get at home, although I don't have an Expodisc, I'll just a piece of plain white paper and the ambient lights I have in my apartment with a tripod.

 

[bycostello]

every camera even same model will give you slightly different results

 

[Drizzt321]

Ok, so did my tests. Not quite as scientific as yours, but I think fairly close. I have the 5d2 and 5d3, tripod against a piece of white paper next to some lights, so OK lighting. First shot was AWB and used that to set the custom WB, and took a 2nd. Pulled all 4 photos into LR4, and looked at the color temp & tint values, then took the 2nd shot and approximately the same area, used LR's WB dropper.

5d2:
Shot 1, AWB: 4350K, -2 tint
Shot 2, Custom: 4200K, -5 tint
Shot 2, LR Custom: 4250K, -3 tint

5d3:
Shot 1, AWB: 4650K, +7 tint
Shot 2, Custom: 4650K, +6 tint
Shot 2, LR Custom: 4650K: +6 tint

The output on my screen (high quality TN, 95% NTSC calibrated panel) looks pretty darn close to each other, even though they are off by ~300K and 10 tint, more or less. Both cameras have, at one time or another, been through CPS for cleaning/servicing. So, I'd reinforce, it's sensor/electronics differences that is the difference, and not a problem with either camera. Don't worry about it, it's all just fine.

 

[drmikeinpdx]

JimmysPhoto:

When I switched from my 5D classic to the Mark III there was a definite shift in the color that I see when imported into Lightroom. It sounds roughly similar to what you are reporting, although my shooting conditions are different.

So far I have just been moving the sliders manually, but one of these days I need to figure out how to get Lightroom to do that for me automatically when the images are imported. It is a minor issue for me, but if I were a high volume photographer it would need to be dealt with right away.

 

[TrumpetPower!]

Use an application such as Raw Digger to see what's actually in the raw files. If the raw files are essentially the same, then it's the auto white balance algorithm in the camera that's changed. If the raw files are different, then there's something different about the sensor, such as a different color filter array or different sensitivities in the electronics or the like.

Either way, it doesn't really matter. If your goal is to match white balance between the two cameras used on the same set, you don't do that by picking the same color numbers. You might be able to do that by picking the same presets (daylight / tungsten / etc.) on both, but the real way is to set a custom white balance off the same target in the same place under the same lighting. And if color is that critical, including matching between cameras, then you need to be profiling (ICC, DNG, Canon Camera Style, whatever) both under the same lighting conditions.

Cheers,

b&

 

[Drizzt321]

JimmysPhoto:

When I switched from my 5D classic to the Mark III there was a definite shift in the color that I see when imported into Lightroom. It sounds roughly similar to what you are reporting, although my shooting conditions are different.

So far I have just been moving the sliders manually, but one of these days I need to figure out how to get Lightroom to do that for me automatically when the images are imported. It is a minor issue for me, but if I were a high volume photographer it would need to be dealt with right away.

That's pretty easy, either in Library mode select all the photos, Right Click->Develop->Auto White Balance, or create a custom preset with Auto White Balance selected and select that preset to apply on import. Or use the dropper function of the LR White Balance to select the point you want to set the WB off of.

 

[dudaferraz]

Hi Jimmy, I have the same issue of a magenta looking image in 5D Mark III. The funny thing is that I use my camera for shooting video and the image looks good on the LCD, but not in any of my computers.

I also work with Canons 70D, 60D and T2i, but all these other models have no such issue.

Did you solve this problem? Thanks!

 

[Maiaibing]

TELL ME WHY!

My guess is that light metering/WB has deteriorated slightly since the emergence of digital cameras - because its less critical than it used to be.

Having used light meters a lot in the film days - when we were dependent on getting the lighting 100% correct from the outset - it seems to me that metering was more reliable on the top bodies back then.

Of course my 5DII is not Canon's "top body" so maybe this is the obvious reason. Just like your 5DIII combo is not.

On balance I can understand if manufactures think that with digital they can go a little cheaper on the metering. And the truth is that I'm still doing better overall with my WB today due to the ability to adjust in post.

 

[StudentOfLight]

TELL ME WHY!

My guess is that light metering/WB has deteriorated slightly since the emergence of digital cameras - because its less critical than it used to be.

Having used light meters a lot in the film days - when we were dependent on getting the lighting 100% correct from the outset - it seems to me that metering was more reliable on the top bodies back then.

Of course my 5DII is not Canon's "top body" so maybe this is the obvious reason. Just like your 5DIII combo is not.

On balance I can understand if manufactures think that with digital they can go a little cheaper on the metering. And the truth is that I'm still doing better overall with my WB today due to the ability to adjust in post.

If I'm shooting RAW I don't care too much as the color can be twerked tweaked as required in post~. On the odd occasion I shoot JPG or mRAW then I spend some time taking a few test shots and adjusting both Kelvin and Tint with the RGB histogram. On most of my cameras K and color shift are easily accessible via the [Q] info screen so it's no big deal to quickly make an adjustment.