Odd Color Temperature results during RAW analysis from Flash photos.

[Goochy]

I am using a Panasonic Lumix DMC-G2 with a couple good grey cards and RAW format files.

When I do a test using 5500 K rated CFL lights, then process these in a RAW processor (Silkypix, Photoshop RAW, or Raw Therapee), sampling the card gives a result pretty close to 5500 K. Great!

However, when I take Flash pictures of the card (various angles, both internal and hotshoe Metz 50 AF-1 flash...), using the dropper to sample the grey card yields a color temperature calculation of about 6900 to 7900 ... perhaps 7600 very common.

The resulting adjustment actually "looks right", but I have not done many tests, or tests of color charts (CMYK, etc.).

The flash is high quality and spec'd for 5500 K and I was hoping it would balance well with the CFL lights.

Any idea what is going wrong? Why is the flash so "hot"?

[William W]

Is the Flash Fired in a BLACK room?

WW

[Colin Southern]

Is the Flash Fired in a BLACK room?

WW

As a good test of Bill's idea, consider taking a photo of the card with the flash outside (perhaps with the camera pointing up to the sky) where no reflected light can hit the card.

[William W]

[Aside]
I hope all is OK with you and yours Colin - and to all other Kiwis here at CiC.
Most tragic and serious couple of weeks.


Bill

[Goochy]

Is the Flash Fired in a BLACK room?
WW

(Though I have also tested with the 5500 K lights on, with the flash overpowering them. I will also test in a low power "fill flash" mode...). ... oh... wait;

You mean a room with black walls (no reflections or only spectrally neutral reflections from various colored walls and objects...)...
Then the answer is no. The room for a few of the tests had dark red walls and painted "white" ceilings.
I can try to eliminate this variable by photographing in a very large room or outside in near 5500 K or dark conditions...

In the meantime, though, I could photograph the walls and ceilings under the 5500 K ("known") lights and see how much of an color temperature impact they have on a spot check... ? I guess this just would add more confusing numbers.


You really think that a short range directed flash to the card will affect the result by 2000 K to 2500 K in an unlit room with white walls?

Thanks,
...Terry.

[Goochy]

While waiting for some appropriate darkness (or some other way to ensure "black room" conditions), I have uploaded a couple of RW2 (Panasonic RAW) and adjusted DNG files for anyone who is curious.

The files are in the 10mb to 20mb range, so I have them on my own website for your option to download and analyze...

http://cosmicpuppy.com/CCPOST8379/

These are shot in a room lit only by cloudy daylight, with white shades drawn; but the light is reflecting around the room...).

The "NaturalLight" files give spot color temp ranges from 6000 to 6300 off the test card.

When I add the Metz 50 Flash (should overpower the daylight, but may have some room reflections...):
The "NaturalLight+Flash" files give spot color temp ranges from 6900 to 7800 off the test card.

I think the wider "range" can be attributed to the close firing range of the flash using the zoom mode.

But the topic question remains as to "why is the temperature so much higher than the expected 5500 K normal of a flash unit"...

Thanks!
...Terry.

[William W]

Thanks for answering my question about the black room - yes I meant a BLACK room - i.e. all walls floor and ceiling black and you dressed in black, too.

Yes, thanks.
I understood the question when it was posted in the original font. (noted, now edited by DH, thank you)
I was attempting to answer it for you - but I required more information to make an informed opinion. For expedience I shall make assumptions.



"I take Flash pictures of the card (various angles, both internal and hotshot Metz 50 AF-1 flash...), using the dropper to sample the grey card yields a color temperature calculation of about 6900 to 7900 ... perhaps 7600 very common."

This means you have the file in your computer and you are running the eye dropper tool over it.

*

"The resulting adjustment actually "looks right" "

This means that the flash image actually looks OK on the monitor - and looks similar if not the same as the image using the 5500 K rated CFL lights.

*

You have a calibrated monitor.
Your viewing room has efficient, homogenised and balanced calibrated lighting.
You are not colour blind in any manner.

*

My first opinionis is, you have a small, but intense and also varying colour spike.

It is from a reflection from some surface in the room.

It is just enough to trick the eyedropper tool because of the continuous spectrum differential, when the eyedropper reads the image file of the grey card, compared to what the naked flash pushes out.

I expect that if you measured the FLASH OUTPUT with a CT Meter it would be 5500K.

As I understand this, you are "metering the CT" of a reflection from a grey card and you are measuring it with a computer tool which is reading a file.

The flash's CT output (in its specs) is measured with a CT meter at source.

Thsoe are two different measurements and two different measurement tools.

Further, I submit that the two different measuring tools are likely to react differently to the little spectrum continuity peaks (or holes).



WW

[Colin Southern]

But the topic question remains as to "why is the temperature so much higher than the expected 5500 K normal of a flash unit"...

That remains the $65,535 question

Can really only be one of two things, either that's the temperature of the light that the flash is putting out, or it's getting tainted by something it's reflecting off. My guess is the latter as I'm a little surprised by the variations you're getting off the card - and - the CT of the flash is pretty much "by design".

[Goochy]

I understood the question when it was posted in the original font. (noted, now edited by DH, thank you)

My first opinionis is, you have a small, but intense and also varying colour spike.
It is from a reflection from some surface in the room.
WW

Firstly, I apologize if my use of fonts was interpreted as frustration. I only like to emphasize the "point" of my postings from time to time, since I am often long-winded. I usually don't get this pointed out to me as a problem. But sorry anyhow.

Second, ... no -- I do not have everything calibrated as I will eventually want to; and I appreciate knowing all the variables involved.

The confusion I have with this "color spike from a reflection in the room" hypothesis, is that I get similar results in different rooms. I find it difficult to believe that the same nature of spike occurs in rooms with different colored walls and furniture, etc..

I posted an additional reply in this thread (up there somewhere), which refers to a daylight lit room and I uploaded the RW2 files just in case that helps (and someone wants to be generous enough to download the files and do some eye-dropper sampling...). Odd Color Temperature results during RAW analysis from Flash photos.

Regardless, your response has been helpful in that I know I have to do this test under more controlled conditions (either a black or spectrally neutrally lined room, or an open space) to eliminate colored reflections of the flash as much as possible, and see if that improves the sampled CT reading.

Thanks,
...Terry.

[William W]

Firstly, I apologize if my use of fonts was interpreted as frustration. I only like to emphasize the "point" of my postings from time to time, since I am often long-winded. I usually don't get this pointed out to me as a problem. But sorry anyhow.

It was not being pointed out as a problem. It was simply unnecessary to do.
If you are cool - I am cool - then let’s move on.

Second, ... no -- I do not have everything calibrated as I will eventually want to; and I appreciate knowing all the variables involved.

Not the biggest issue in this matter IMO – forget this for the minute.

The confusion I have with this "color spike from a reflection in the room" hypothesis, is that I get similar results in different rooms. I find it difficult to believe that the same nature of spike occurs in rooms with different colored walls and furniture, etc. . .

Nope – not necessarily the same nature of colour spikes (or dips) – but similar results from your eyedropper / computer testing set up.

1+1+1 = 3 and 2+5-4 = 3 and 3+6-5=4. These three sums give "similar results"

Do you understand my point?

Regardless, your response has been helpful in that I know I have to do this test under more controlled conditions (either a black or spectrally neutrally lined room, or an open space) to eliminate colored reflections of the flash as much as possible, and see if that improves the sampled CT reading.

I am not sure what this testing is to prove or nor to prove – or the need for it, save that of being fun to do.

I am not questioning why you want to do this - I am wanting to steer you to an output and then see if the testing has any relevence to an output or usable result

The final output is the finished product – print or file – and the CT impression of that will change, dependent upon wherever and however that end product is viewed.


So the practical application of CT is to match the CT of the Flash (at source) with the Ambient Light at source, such that you don’t get colour casts.

*

Notwithstanding that – have I clearly explained:

1. the TESTING equipment you are using (eyedropper tool) is different to the testing equipment used to measure the CT of a Flash by the manufacturer. Therein different algorithms and accuracies (or inaccuracies).

2. the Testing equipment you are using is NOT measuring the CT of the output of the Flash – it is measuring a sample of a file.

*

Thinking further on this matter and just comparing the results from your eyedropper tool measuring device on the Lights and the Flash: the effective Tv (shutter speed) will also play a role, as also might Reciprocity at High Shutter speeds.

When you shoot the grey card with the lights, your effective Tv will be the actual Tv used on the camera.

When you shoot the grey card using the Flash, your effective Tv will be closer to, or actual Flash's duration.


WW

[Goochy]

Notwithstanding that – have I clearly explained:

1. the TESTING equipment you are using (eyedropper tool) is different to the testing equipment used to measure the CT of a Flash by the manufacturer. Therein different algorithms and accuracies (or inaccuracies).

2. the Testing equipment you are using is NOT measuring the CT of the output of the Flash – it is measuring a sample of a file.
WW

My question is partly theoretical and partly practical. Indeed, the practical part is figuring out if it is ... practical ... to mix flash and continuous lighting in the same situation; and, if so, what parameters (pre-shoot, and post-processing) should be considered. I will have some control over the "studio" setting, and want to plan these controls.

But I want to understand the theory first ... i.e., "break down" the components of my understanding of white balance, RAW processing, etc.. And this discussion is helpful, but inconclusive so far.


I feel that I am using the post-processing "eye-dropper" grey point / spectral neutral point analysis correctly due to my results of a continuous lighting control test.

Under about 1500 effective watts of "Photo rated, 5500 K" CFL (compact fluorescent) lights in softboxes; snapshots of more than one brand of grey card and different brands of RAW processing software, all yield eyedropper evaluation results of 5500 +/- 100 Kelvin.

In other words, I get "very expected" results using continuous lights, which is comforting and easy to understand.
This gives me some confidence (or false confidence) that I am using the grey card and RAW processing properly.


So, unless my CFL results are just "lucky", I would like to be able to get to the same level of confidence and understanding of Flash color temperature.

I am being proactive so that I:

1. Plan and use the studio as best as I can with my available equipment.
2. Know in advance how complex my post-processing workflow will be.

...Terry.

[William W]

I feel that I am using the post-processing "eye-dropper" grey point / spectral neutral point analysis correctly due to my results of a continuous lighting control test.

Under about 1500 effective watts of "Photo rated, 5500 K" CFL (compact fluorescent) lights in softboxes; snapshots of more than one brand of grey card and different brands of RAW processing software, all yield eyedropper evaluation results of 5500 +/- 100 Kelvin.

In other words, I get "very expected" results using continuous lights, which is comforting and easy to understand.

This gives me some confidence (or false confidence) that I am using the grey card and RAW processing properly.

So, unless my CFL results are just "lucky", I would like to be able to get to the same level of confidence and understanding of Flash color temperature.

Yep, I understand the theory of your comprehensive testing. (i.e. I understand your methodology).

I also understand why you are doing this.

I have not thought that you are using the Gray Card or the RAW processing tool “incorrectly”. In my mind it is NOT about how YOU are using the gear. . . It is about the two point below . . .



Testing and Hypotheses Logic dictates that:

If you are NOT getting “consistent and expected” results with the Flash,
AND IF we ASSUME that the results with the Continuous Lights are constant, expected and reliable, then either:

1. the testing equipment is reacting differently to the Flash.
2. there is one or more differentials in the Flash Test, which is skewing the result.
3. Both of the above

I cannot think of any other mechanisms to cause the “inconsistent” results you are getting.

What I have done, in previous posts here, is to list all the possibilities I can think of, that could fit in category (1) or (2) above.

Colin’s point about the grey card is valid – and that fits in with my category (1). Or if you are using DIFFERENT grey cards then that is category (2).

You might well find that the “consistent” result for Flash – using your testing mechanism is: 7300K –

Try another Flash unit, that is another logical step.

WW

[Goochy]

Try another Flash unit, that is another logical step.
WW

Yup... Definitely will do more testing as methodically as possible before coming back with more results or conclusions. Thanks.

I have already tested with another flash ... that being the pop-up one built into the camera. Also gives the ~7000K CT.

I appreciate all the suggestions and hidden variables; though I was honestly hoping that I was missing something really simple and obvious.

Catch y'all in a few days or so, once I get through some "work" and then back to the photo hobby.