[HDRI] Question about the overall correction factor for raw2hdr

[Zhe Kong]

Dear all:
Sorry to bring up this topic again, but I run into the same issue Coltide
has (Coltide's original post is here: http://radiance-online.org/
community/mailing-lists/archives/hdri). I also use Canon 5D 2 with
SIMGA Fisheye lens 8mm f/3.5. I want to figure out the -C factor when using
raw2hdr to combine LDR images. I took the images (f/11) under different
circumstances: two HDR images were taken along a side window, two
images were taken under electric lighting condition, and the last one were
taken far away from a glazing. For each scene, 12 images were taken and
assembled. The luminance values from the HDR images and the lighting meter
are below (cd/m2):
Side window: 123.0(HDR) 225.0(measurement) 1.83(-C factor)
Side window: 101.0(HDR) 195.4(measurement) 1.93(-C factor)
Electric light: 1.9(HDR) 3.3(measurement) 1.74(-C factor)
Electric light: 1.4(HDR) 2.6(measurement) 1.79(-C factor)
Away from glazing: 9.8(HDR) 18.5(measurement) 1.88(-C factor)

Questions:
1) I can average the results (1.84) as the correction factor, but is it
correct? Some values have great difference (1.74 VS 1.93).
2) I notice that I use the same model as Coltide, however, our results are
very different. Is this normal? (The production of each camera and lens
cannot be exactly the same?)
3) Does lighting source influence the accuracy of the factor? Focusing on
daylighting data collection, should I take images under consistent electric
lighting conditions or daylighting conditions? (Daylighting conditions have
dynamic changes, but electric situations offer lower results.)

Any suggestion or comment is appreciated.

[Raquel Viula]

Hi Zhe,
I'm using a commercially calibrated camera (LMK mobile air) so my experience
with calibration is limited. But I¹ll try to answer your questions.

1) I can average the results (1.84) as the correction factor, but is it
correct? Some values have great difference (1.74 VS 1.93).
I don't think you should use an average. My understanding is that you should
always use the calibration factor for the main light source in your scene.
The spectrum of the different light sources is different and that affects
the measurements.

2) I notice that I use the same model as Coltide, however, our results are
very different. Is this normal? (The production of each camera and lens
cannot be exactly the same?)
Apparently yes, they can be different. Not sure if it¹s the production that
is different or it is simply the fact that mechanical systems do not always
behave in the same way. Even your own system will not always produce the
same result. That¹s why you should also test the reproducibility of your HDR
luminance value. You test that by doing not one but several captures and
using the average of those for calibration rather than the result of one
single capture. In that way you would be accounting for the camera's
mechanical uncertainties.
It looks like there are also sources of error relating to the luminance
meter measurement itself, as described in Inanici 2006, Evaluation of high
dynamic range photography as a luminance data acquisition system. So that
might add to the differences between yours and Clotilde¹s calibration.

3) Does lighting source influence the accuracy of the factor? Focusing on
daylighting data collection, should I take images under consistent electric
lighting conditions or daylighting conditions? (Daylighting conditions have
dynamic changes, but electric situations offer lower results.)
Different light sources will provide different calibration factors. I was
told by the seller of my system that a measurement under white LED has
errors of 5 to 12% in relation to the measurement done with a halogen-based
calibrated camera. But LED has a lot of flicker, doesn¹t it? I¹m not sure
how people using your calibration method deal with fluctuations of the
lighting conditions during capture, in the case of daylight scenes from
variability of the sky and in the case of the electrically lit scenes due to
flicker. Your LDR shots are done over a period of time and light can change
from one second to the next. I wonder when does your luminance spot
measurement take place in relation to the LDR captures and if that might
affect the result? Do you take illuminance measurements during luminance
collection to see how light varies?
In any case, I think you should always make you collection under very stable
lighting conditions, whether daylight or electric light. One option is to
calibrate for the most stable electric light source that you can find and
then make the correction for the daylight spectrum independently, if you
have access to a spectrometer.

Hope it helps.

Best Regards

Raquel



Raquel Viula

PhD candidate



TU Delft | Faculty of Architecture and the Built Environment |

Architectural Engineering and Technology

Julianalaan 134, 2628 BL Delft, The Netherlands | P.O. Box 5043 2600 GA
Delft
M [phone] | Email ***@tudelft.nl


From: Zhe Kong <***@gmail.com>
Reply-To: High Dynamic Range Imaging <***@radiance-online.org>
Date: Monday 27 February 2017 22:25
To: <***@radiance-online.org>
Subject: [HDRI] Question about the overall correction factor for raw2hdr

Dear all:
Sorry to bring up this topic again, but I run into the same issue Coltide
has (Coltide's original post is here:
http://radiance-online.org/community/mailing-lists/archives/hdri). I also
use Canon 5D 2 with SIMGA Fisheye lens 8mm f/3.5. I want to figure out the
-C factor when using raw2hdr to combine LDR images. I took the images (f/11)
under different circumstances: two HDR images were taken along a side
window, two images were taken under electric lighting condition, and the
last one were taken far away from a glazing. For each scene, 12 images were
taken and assembled. The luminance values from the HDR images and the
lighting meter are below (cd/m2):
Side window: 123.0(HDR) 225.0(measurement) 1.83(-C factor)
Side window: 101.0(HDR) 195.4(measurement) 1.93(-C factor)
Electric light: 1.9(HDR) 3.3(measurement) 1.74(-C factor)
Electric light: 1.4(HDR) 2.6(measurement) 1.79(-C factor)
Away from glazing: 9.8(HDR) 18.5(measurement) 1.88(-C factor)

Questions:
1) I can average the results (1.84) as the correction factor, but is it
correct? Some values have great difference (1.74 VS 1.93).
2) I notice that I use the same model as Coltide, however, our results are
very different. Is this normal? (The production of each camera and lens
cannot be exactly the same?)
3) Does lighting source influence the accuracy of the factor? Focusing on
daylighting data collection, should I take images under consistent electric
lighting conditions or daylighting conditions? (Daylighting conditions have
dynamic changes, but electric situations offer lower results.)

Any suggestion or comment is appreciated.

[Axel Jacobs]

Hi Zhe and Raquel,

my thought on this below...

Best regards

Axel

Post by Raquel Viula
Hi Zhe,
3) Does lighting source influence the accuracy of the factor? Focusing
on daylighting data collection, should I take images under consistent
electric lighting conditions or daylighting conditions? (Daylighting
conditions have dynamic changes, but electric situations offer lower
results.)
Different light sources will provide different calibration factors. I
was told by the seller of my system that a measurement under white LED
has errors of 5 to 12% in relation to the measurement done with a
halogen-based calibrated camera. But LED has a lot of flicker, doesn’t
it?

I looked into flicker some time ago, and presented my conclusions in a
presentation at the 2012 Radiance workshop:

http://www.jaloxa.eu/mirrors/radiance_workshops/2012/Talks/Jacobs-AJ09-HDR_Radiance_WS-2012.pdf

Skip to page 23 for flicker.

In short: If you are seeing the actual artificial light sources in the
photograph, then you need to use short exposure times to capture them
accurately. In this case, it is safe to assume that the lights flicker
(simply because most do, particularly LEDs), and are not accurately
represented in the HDR. If HDR glare measurement is what you want to
do, this is obviously a problem.

You probably won't have the equipment to measure flicker. What you can
do is take 5 to 10 photographs with the shortest exposure time you need
for your sequence. Keep all camera settings the same. If all
photographs look identical in their brightness, the light source is
flicker free, or flickers at such high frequency or with such a low
amplitude that you can ignore it.

Hope this helps

Best

Axel