[Anodrac]

I would like to profile my camera, and I should take a photograph of a colour chart target when the sun provides the right temperature (5000K).


I have read the instructions in the pilxs.us forum. The author of the post says that,when he sets a custom white balance using a grey card, his camera displays which temperature the profile has. He doesn’t mention his brand/model but it is not Pentax.


I have tried to find this info displayed in my Pentax K-1 when setting a manual WB, but I couldn’t. Is this feature missing in the K-1 or I am missing something?


In case it can’t be done with my camera, does anybody knows other ways to know the exact (or really accurate) temperature of the scene?


I suppose I could tether the camera to a laptop and see the temperature in Darktable or Rawtherapee, but I should do this in the field and I am looking for a more practical alternative.

---

Last edited by MarkJerling; 08-28-2019 at 02:29 AM. Reason: Typos corrected to make more easily readable.

[08amczb]

I think you can download the image to your phone through wifi and get the information there. I have no better idea.

[Anodrac]

That might be a solution, I've never done it. Thanks for your quick reply. I have also seen that there are color meters, but they are very expensive as I would only use it once.

[K1N8]

There is some white balance info in the EXIF data


Pentax Tags

"WhiteLevel" would range from 12291 to 16318 in my collection, that might be something comparable for color reference.


And there seems to be some color temperature readings:

Pentax Tags

"White balance Blue/Red gains as a function of color temperature."


[...]

"KelvinWB_Daylight": "5205 0 2.1156005859375 1.5146484375",
"KelvinWB_01": "2500 0 1.137939453125 2.7635498046875",
"KelvinWB_02": "2630 0 1.202392578125 2.65625",
"KelvinWB_03": "2780 0 1.283203125 2.525634765625",
[...]

Please share if, during your calibration, you can make any sense of that data.

[stevebrot]

There is some white balance info in the EXIF data

There is, but it is in the form of factors (four integers, RGGB) to be applied directly to the sensor voltages during processing to actual image data (pixels). The values are set to the Pentax:WhitePoint tag*. There is a set of algorithms for conversion between those factors and K° + tint, but I don't have it handy. IIRC, there may be a way to extract the value using dcraw or similar.

The KelvinWB tags are a lookup table. I have never confirmed, but believe they are always the same.


Steve


* ExifTool tag name for that portion of the Pentax makernotes

[stevebrot]

There is a set of algorithms for conversion between those factors and K°, but I don't have it handy.

Here is one version that may be applicable...have fun.

Color temperature - Wikipedia


Steve

[stevebrot]

I would like to profile my camera, and I should take a photograph of a colour chart target when the sun provides the right temperature (5000K).


I have read the instructions in the pilxs.us forum. The author of the post says that,when he sets a custom white balance using a grey card, his camera displays which temperature the profile has. He doesn’t mention his brand/model but it is not Pentax.

Welcome to the Pentax Forums!

It would help if we knew what you are trying to do and why. A link to the instructions would be helpful.


Steve

---

Last edited by stevebrot; 08-28-2019 at 06:28 AM.

[K1N8]

The KelvinWB tags are a lookup table. I have never confirmed, but believe they are always the same.

Ah, shame. I have all my EXIF data collected in a database and did a quick check, Yes, across ~16200 photos they are the same.

[Anodrac]

K1N8:
Using exiftoolsoftware under linux I can list all the exif data like this:
exiftool -a -u -g1 File_name.DNG


There I can findthis tag: Color Temperature : 5150
Whenever Darktable displays 5203K, and Rawtherapee displays 5227K for the same file (Different algorithms taking into account other exif data related totint?? I don’t know)
The photo was taken with Auto White Balance.


The tag «Whitelevel»definition: «This tag specifies the fully saturated encoding levelfor the raw sample values. Saturation is caused either by the sensoritself becoming highly non-linear in response, or by the camera'sanalog to digital converter clipping».
I don’t understandit.


Stevebrot
This is the link ofwhat I’d like to do:PIXLS.US - Profiling a camera with darktable-chart
According to theauthor, individually profiling your camera can provide better colorresults. To be honest it is not something that I absolutely need, but anyway.
It is not simple, but the guide is detailed enough to give it a try.
p { margin-bottom: 0.25cm; line-height: 120%; }a:link { }

[jbinpg]

Note there is an error in the command given to extract the info using exiftool. It should be (using one of my K3's files):

exiftool -a -u -g1 K3A_0237.DNG

Note the space between -g1 and the filename.

Jack

[Anodrac]

You are right. Thanks for your correction, I will edit my post as soon I can. Shoudn't be an edit option?

[stevebrot]

This is the link ofwhat I’d like to do:PIXLS.US - Profiling a camera with darktable-chart
According to theauthor, individually profiling your camera can provide better colorresults. To be honest it is not something that I absolutely need, butanyway.

Cool! That is what I thought, but knowing the target software and the method helps a lot on this end. With any luck user @BigMackCam will drop in on this thread. They are a Darktable user and may have some helpful hints.


Steve

[photoptimist]

....

In case it can’t be done with my camera, does anybody knows other ways to know the exact (or really accurate) temperature of the scene?

There are a couple of obstacles to your quest:

1) "Color temperature" can never be accurate or exact except when the light source is a true blackbody radiator with constant emissivity and attenuation (e.g., atmospheric absorption) across all wavelengths. The sun is pretty close to a blackbody but even it is not exact due to spectral emissions and absorption lines). And once the sun's light goes through the atmosphere, it's light is only approximated by color temperature (dust and humidity ruin the blackbody assumption). Some artificial light sources (LEDs, fluorescent, mercury vapor, sodium, neon, etc.) are nothing like blackbody radiators and color temperature is a very poor approximation of the white balance of their light.

2) Most scenes have more than one color temperature because color temperature of each part of the scene depends on exactly what light is hitting that part of the scene. A bride standing outdoors next to a red car will have a range of color temperatures -- redder ("temperature" lower than 5000K) on the side facing the car, whiter ("temperature" closer to 5000K), or bluer ("temperature" above 5000K) in any shadows lit by blue sky.

The point is that color temperature is an approximation, not anything that can be accurate or exact.

[BigMackCam]

In reality, I've found that the target of 5000K colour temperature isn't critical. Just photograph your chart on a nice, consistently sunny, cloudless day with the sun directly overhead, and you'll be fine. It's far more important to ensure you take multiple shots under the same lighting conditions at slightly different EV compensations, so that you have at least one raw + JPEG pair where the neither the darkest grey and lightest white squares from your test chart are clipped.

An alternative, though perhaps controversial, approach to lighting a chart for colour profiling is to use an artificial lighting source with colour temperature control (one designed for photography / film-making). A little while ago, I bought an Aputure AL-MX pocket LED light panel, with stepped control of colour temperature from 2800 - 6500K. Tripod-mounting it close to my Colorchecker Passport target and angled so as to avoid glare, in an otherwise dark room (night time, curtains / blinds closed) it's a good way to illuminate the target with something close to the required temperature. I had my doubts as to the consistency and accuracy of this unit, but in practice I've found it to work well. I recently created some dual-illuminant profiles for use with RawTherapee, and they worked as intended. The beauty with this, of course, is that you don't need to wait for a specific time on a sunny day to take colour chart shots... You can do them any day you like, and at any time (so long as you can largely eliminate other lighting sources by darkening the room).

[Anodrac]

That was really helpful.

I didn't know about dual-illuminant profiles, so I have read the rawpedia and it sounds great, although I am new to this and I will have to do it in order to understand how it works.

1- Making a DCP dual-illuminant makes the DCP single-illuminant redundant or it is worthy having both?
2- I guess it can be also done under Darktable, isn't it? (I am far more familiarised with Darktable, although after watching Carafife's video series about Rawtherapee I am feeling the more and more confident)

The possibilities look very promising, as this profiles could also emulate the Pentax profiles and filters if I understand it correctly.