wrong color temperature Godox ? Do I do something wrong

[rlast]

Your brightness is likely much to high. As a guess, I suspect a value closer to 30% would be where it should be. 7400K seems to be unreasonably high. I suspect that 5500K is a more reasonable value. If C1 is showing 7100K, I suspect that the colours in the image are fooling it.

Sorry Manfred, I don't understand this because I'm not judging and setting the WB based upon what I see, but let Capture One do it via the picker tool which corrects it based upon the RGB values captured.

[rlast]

I'm not sure what you try to investigate now: white balance or color temperature.

I opened a nef in 3 Nikon converters, in DxO and in RT. Nikon tells the color temp is 5601, DxO says 5400 and RT says 5367. I don't think it's hard science in photography.

I'm still puzzled with color temp. High color temp is white, low color temp is red. In photography it's the other way. More like a correction on the 'measured' color temp. That means there's something as the 'measured' temp.

George

George : I'm trying to understand - why my white balance needs to be 7400K, when I shoot a 5500K source (my flash) on a greycard …
The result is not matching my expectation, so either my thinking is not correct, my tools are not good enough, or my method is wrong.(problably a bit of all ;-) )

[xpatUSA]

Originally Posted by rlast so my expection was that a picture taken of the greycard with this setting should be rather neutral.
But if I compare it with real grey (background of C1) it looks very blue-ish, so a correction to 7200-7500K is enough.

If the colour does not look neutral, then ...

Too much emphasis on "how it looks" ... leading inevitably to the standard advice re: monitor calibration, etc.

We should be commenting on the file RGB values, not how it looks on-screen!

If a file data shows R=G=B for a pixel then that is neutral and is correct, no matter how it looks on one's screen.

[rlast]

Too much emphasis on "how it looks" ... leading inevitably to the standard advice re: monitor calibration, etc.

We should be commenting on the file RGB values, not how it looks on-screen!

If a file data shows R=G=B for a pixel then that is neutral and is correct, no matter how it looks on one's screen.

ExpatUSA: I'm sorry for the confusion. My experiment is without judging what I see on the screen. but just the the RGB numbers.
As simple as this:
* I take a photo of my grey card with WB set to 5500K
* The greycard is lightened with my flash, specified 5500K
* in Capture one I disable the std. processing (noise, sharpeing, color profiles)
* I see that B has a higher value than R
* I used the color picker tool of the whitebalance in Capture one, and after that the WB is set to 7400K
* the latest result surprised me ...

I could have done this on a black/white display (speaking figurative)

Hope this clarifies / summarized what I did.

Thanks!
Rudolf



Verzonden vanaf mijn iPhone met Tapatalk

[Manfred M]

Too much emphasis on "how it looks" ... leading inevitably to the standard advice re: monitor calibration, etc.

We should be commenting on the file RGB values, not how it looks on-screen!

If a file data shows R=G=B for a pixel then that is neutral and is correct, no matter how it looks on one's screen.

I understand where you are coming from Ted and at one point many years ago I felt much the same way. Colour vision is not just a matter of physics, but also has physiological and psychological components. What a digital camera records is pure physics.

For some photography, I absolutely agree we want to very precisely reproduce the colours that have been recorded; scientific photography or documentation photography, for instance. Corporations tend to like to see their logos and corporate colours reproduced "accurately". For other types of image making it can range from somewhat important to something that is completely unimportant.

What it looks like on screen is very important. Quite often a "correct" white balance is really little more than a very good starting point.

[Manfred M]

Sorry Manfred, I don't understand this because I'm not judging and setting the WB based upon what I see, but let Capture One do it via the picker tool which corrects it based upon the RGB values captured.

That suggests to me that one of two things is likely happening:

1. The colour temperature of the light hitting the target is not pure and the contamination is showing a higher colour temperature than you would expect. This could for instance come from light from your flash illuminating something with a lot of blue in it; or

2. The C1 algorithm failing because it was not written for the lighting conditions in the scene you are shooting. I suspect it is not your flash.

[xpatUSA]

George : I'm trying to understand - why my white balance needs to be 7400K, when I shoot a 5500K source (my flash) on a greycard …
The result is not matching my expectation, so either my thinking is not correct, my tools are not good enough, or my method is wrong.(problably a bit of all ;-) )

So, Rudolf, trying to keep up, I looked at your first screen capture which confirms what you've been trying to tell us all along.

As far as I can tell, it shows the C1 window immediately after clicking in the middle of the image with color-balance picker:



It seems that the picker set the temp to 7458K and the tint to -8.8, the result of which set the RGB to 150, 149, 151 and set a fourth number in white of 150 which might be Intensity (R+G+B)/3. Those numbers tell us that the color balance function worked, in that the file numbers for the pixel(s) in question are close enough to neutral (and the image on my uncalibrated screen also looks neutral ... ).

Therefore, either the raw capture was not neutral (lighting, whatever) or the conversion from raw to RGB was not neutral (a capture one problem). Manfred has already suggested similarly.

Is it possible to post the RX_07785.ARW file somewhere (e.g. DropBox) so that we could download and look at it in another converter (e.g. FastStone or RawTherapee or IrfanView)?

That could cut down on all this guess-work and bring us closer to the cause ...

[xpatUSA]

Manfred mentioned the possibility of the color-balance function not having enough range in his post #26.

By accident, I noticed this today:

White Balance range : I can work with IR easily as the white balance range is simply very wide. Lightroom requires lots of hoop jumping to get rid of red cast in ~700nm IR images. Not here

source: https://www.reddit.com/r/photography...ly_its_really/

Admittedly, IR is "opposite" to too much blue but hopefully the point is clear: if C1 lacks color-balancing range, maybe RawTherapee has more ...

[rlast]

That suggests to me that one of two things is likely happening:

1. The colour temperature of the light hitting the target is not pure and the contamination is showing a higher colour temperature than you would expect. This could for instance come from light from your flash illuminating something with a lot of blue in it; or

2. The C1 algorithm failing because it was not written for the lighting conditions in the scene you are shooting. I suspect it is not your flash.

Manfred, I will prepare a RAW file, hoping that one of you could confirm my results, what could exclude the C1 processing or not!

[rlast]

Ted, thanks for your excellent suggestion, and I will do that.
At the moment I have very little time, so it will be during the weekend …
Rudolf

[rlast]

Manfred, Ted + others …
I redid the shot, as good as possible with my equipment.
Like before shot with my Sony RX10, WB 5500K, and Godox V860ii - 5500K spec'd

Please find below links to the RAW + JPEG file (out of camera).

This is how it looks like in capture one (crop applied, to exclude vignetting)


The files can be downloaded via wetransfer (popular service in the Netherlands)
https://wetransfer.com/downloads/273...0130916/c1d2f7

Interested to hear, what your software thinks about the WB needed for a neutral response.
Thanks for your help and effort!
Ruud

[george013]

When I open the raw in DxO the wb is 5154K and -7 tint and looking blueish. When I use the picker the wb changes to 7492K and 0 tint. The average pixel value is 160. RT is not yet installed on this pc.

Just opening shows me a histogram with an average value for R=165, G=185 and B=200. Far away from what supposed to be gray. Three different columns hardly overlapping each other. Strange.

George

[george013]

Just some thoughts.
Shooting a gray card should give you a mostly gray image. The gray pencil is used for further editing. So the focus must be placed on thge image as it comes out of the camera.
Could you try to shoot it again with awb? The exif says manual white balance. And with A or S mode? Exposure is now manual and the place in the histogram is to much to the right.
Maybe it would be wise to try it not with spot metering.

George

[xpatUSA]

Manfred, Ted + others …
I redid the shot, as good as possible with my equipment.
Like before shot with my Sony RX10, WB 5500K, and Godox V860ii - 5500K spec'd

Please find below links to the RAW + JPEG file (out of camera).

This is how it looks like in capture one (crop applied, to exclude vignetting)


The files can be downloaded via wetransfer (popular service in the Netherlands)
https://wetransfer.com/downloads/273...0130916/c1d2f7

Interested to hear, what your software thinks about the WB needed for a neutral response.
Thanks for your help and effort!
Ruud

Hello again, Ruud,

First I looked in EXIF and meta-data for the raw and the OOC JPEG.

I see that you shot at 220mm focal length, field of view 3.4 [sic] deg, hyperfocal distance 1098m. Seems a little extreme ...

I see that neither the raw nor the JPEG has an ICC profile meaning that no color-management will be applied when viewing in my editor (RawTherapee).

Still, each one in RT renders with a cyan cast 210 deg at about 20% saturation (HSV) brightness < 70%.

Clicking the WB picker raised the temp and changed the tint similar to your experience.

a) I'm suspicious of the focal length used - at that shooting distance **, surely extraneous light incident onto the card will have a greater ratio to the illuminance from the flash unit and thereby have a greater effect.

** not saying that you shot at 1098m !!

b) The absolute WB numbers given by any editor must be suspected in the absence of an embedded ICC profile in either image. The profile shown in your screen-shot must be the converter's default working color space for your camera; RawTherapee uses something completely different ...

Thanks for posting the images for us to look at!

[xpatUSA]

The exif says manual white balance.

The EXIF actually says Color Temp 5500, Color Comp 0.

Could you try to shoot it again with awb? And with A or S mode? Maybe it would be wise to try it not with spot metering.

Why not also just throw in different ISOs and Color Modes and Spaces? After all, the more variables, the better ...

... the place in the histogram is too much to the right.

Why? The sensor was about 1/2 exposed (according to sRGB gamma).

[rlast]

Just some thoughts.
Shooting a gray card should give you a mostly gray image. The gray pencil is used for further editing. So the focus must be placed on thge image as it comes out of the camera.
Could you try to shoot it again with awb? The exif says manual white balance. And with A or S mode? Exposure is now manual and the place in the histogram is to much to the right.
Maybe it would be wise to try it not with spot metering.

George

George, I will try to shoot with AWB later. I'll get back to you.

The exposure is indeed to the right, but maybe this is also because of the wrong / strange WB?
In a next shoot, I can make it more precise and see if that makes a difference. I don't expect it, but optimizing the setting never hurts! :-)

Thanks for looking into this!
Ruud

[rlast]

Thanks Ted for looking into this!

Hello again, Ruud,

First I looked in EXIF and meta-data for the raw and the OOC JPEG.

I see that you shot at 220mm focal length, field of view 3.4 [sic] deg, hyperfocal distance 1098m. Seems a little extreme ...

I see that neither the raw nor the JPEG has an ICC profile meaning that no color-management will be applied when viewing in my editor (RawTherapee).

Still, each one in RT renders with a cyan cast 210 deg at about 20% saturation (HSV) brightness < 70%.

Clicking the WB picker raised the temp and changed the tint similar to your experience.

a) I'm suspicious of the focal length used - at that shooting distance **, surely extraneous light incident onto the card will have a greater ratio to the illuminance from the flash unit and thereby have a greater effect.

This RX10M3 is a bridge camera, with a 1" sensor, and fixed zoom lens from 9mm tot 220mm (24 to 600mm FF eq.)
https://www.dpreview.com/reviews/son...t-dsc-rx10-iii
(See this website for more reading)
The minimum focus distance @ 220mm is 73cm - which is quite exceptional, and this is what I used to have the greycard filling my frame.

** not saying that you shot at 1098m !!

b) The absolute WB numbers given by any editor must be suspected in the absence of an embedded ICC profile in either image. The profile shown in your screen-shot must be the converter's default working color space for your camera; RawTherapee uses something completely different ...

I believe the ICC profile is included in the file, but I manually set the profile to linear, (I though this was best for this WB test).
This was part of my action to set all processing to zero (sharpening, noise reduction) to avoid any impact from processing in this quest. (Normally I have all these settings enabled, including the ICC curve)

Thanks for posting the images for us to look at!

Thanks for your help!

[xpatUSA]

This RX10M3 is a bridge camera, with a 1" sensor, and fixed zoom lens from 9mm to 220mm (24 to 600mm FF eq.)
https://www.dpreview.com/reviews/son...t-dsc-rx10-iii
(See this website for more reading)
The minimum focus distance @ 220mm is 73cm - which is quite exceptional, and this is what I used to have the greycard filling my frame.

That could bring us to the question of flash power, about which I know very little.

Your flash has a guide number of 60 meaning 60 meters at 100 ISO:

http://www.godox.com/EN/Products_Cam...sh_V860II.html

What I don't know is which power (out of the 22 steps from 1/128 to 1/1) should be set or occur automatically when shooting at 0.73m. Others here should know.

I believe the ICC profile is included in the file ...

I can assure you that neither the raw nor the JPEG contains an embedded ICC color profile of any kind, which means that ...

but I manually set the profile to linear

... I don't know what is meant by that, sorry.

(I though this was best for this WB test).
This was part of my action to set all processing to zero (sharpening, noise reduction) to avoid any impact from processing in this quest. (Normally I have all these settings enabled, including the ICC curve) !

My gut feeling is for an error in how the flash itself was used, rather than there being a problem with the flash.

As to the profile, if no profile is found, viewers assume that the file image content is sRGB which I assume is what your Sony is set to ...

[george013]

This is the histogram from DxO. Just from the raw image. If you did shoot a gray card with the right settings, the 3 colors should be equal or about.



George

[george013]

The EXIF actually says Color Temp 5500, Color Comp 0.



Why not also just throw in different ISOs and Color Modes and Spaces? After all, the more variables, the better ...



Why? The sensor was about 1/2 exposed (according to sRGB gamma).

No, the exif says wb manual. Probably he used a value of 5500 for the manual wb.
The main point is that when I shoot a gray card under normal exposure, I expect a histogram with the three channels being equal and somewhere in the middle. That would be my start point.
A further investigation might deal with the color temp of the used flash.

As you can see in DxO the color temp is 5154 and tint -7.

If I change the wb setting to flash, I get a gray image, the three channels are covering each other nearly. If I go higher the covering of the channels is getting better. His manual wb setting is wrong.

George