Re: Color of a rose...
Re: Color of a rose...
- Subject: Re: Color of a rose...
- From: Robin Myers <email@hidden>
- Date: Wed, 3 Jun 2009 13:53:28 -0700
To clarify things a bit, there are NO naturally fluorescent flowers,
so that is not the problem. The problem is observer metamerism
failure; the camera does not have the same spectral sensitivities as
the human vision system. Measuring the rose color with a spectrometer
will, once compared with the spectral sensitivities of the camera,
give you a clue as to what wavelengths are causing the problem. Then
you can possibly add a filter to the lens to change the sensitivity
and give more accurate results.
There is a paper detailing a similar problem with artists' pigments at
http://betterlight.com/downloads/whitePaper/wp_color_accurate_photo.pdf
Regards,
Robin Myers
On Jun 3, 2009, at 12:21 PM, Sean wrote:
Hello all,
First thanks for this great list. I'm a photographer in the Los
Angeles area and I joined this list a few weeks ago in the hope of
improving my knowledge of color in general, and particularly color
as it relates to my photography. So far, mission accomplished! The
archives are a treasure trove.
I would really like to hear your thoughts on rendering red flowers
(first on monitors).
Recently I've been doing some camera tests to suss out the
limitations. For the test, I'm using a Nikon D3, and processing my
exposures with either photomatix or ACR. I have not calibrated the
D3, but I am viewing on a calibrated NEC 2690. Debayering or HDR is
done into 16 bit ProphotoRGB.
Here's the rub: A rose will render on the NEC (or cinema display) to
a kind of deep orang-ish red, rather than to the deeeep (almost
purple-ish) red I see when I look at the roses in the real world.
Now, it's pretty obvious that when shooting most objects for non
scientific/non archival purposes the goal is - for me anyway - more
of an artistic interpretation of the color than a strict
(spectroscopic?) reproduction of it. And that approach works very
well in general for most subjects that I shoot (flesh-tones,
landscapes, music venues).
Roses are a bitch.
Okay, so here's my thoughts.
Flower: I don't really know much about flowers, but I imagine that
part of what makes a rose so spectacularly red is possibly due to
florescence. I'm not putting any filtration on my glass (in this
case a Nikkor 17-35) but my guess is that by the time the light hits
the infrared filter on the sensor much of the "real world" chromatic
information has already been altered quite a bit. (I do plan on
picking up an achromatic lens for this flower project going
forward...)
Camera: other than the fact that the camera has relatively fewer
red filtered photosensors I would imagine that potentially, both the
infra red filter, and the red photosensor filters combine to have a
cutoff that simply blocks out the reds I'm looking for. Also the
red channel is quick to clip which is why I am experimenting with
HDR. The benefits of local contrast doesn't hurt here either.
ACR: perhaps this is more an issue of debayering than the inherent
ability of the sensor. I only understand debayering in a very basic
way, but it seems to me that companies like Nikon would be inclined
to tweak their algorithms so that they're optimized for "general
photography" (e.g. photographing people). To that end I would
think skin tones would be a priority - which (I don't know) might
explain the orange-ish rendering of roses? - just guessing here.
Monitors: The NEC is a great monitor. I'm happy with it under most
conditions and it renders colors beautifully to my eye so I won't be
spending 6 or 12 grande on anything "better". Though I'm sure
better exists. That said my final guess here is that my monitor
(like most monitors for that matter) simply can't reproduce the reds
that I'm looking for.
So - the question is: is it possible to render the reds in a rose on
a monitor any better? I did a quick search for "rose" on google and
the resulting images confirm my observations; it's not just me! I
guess I'm relieved (sort of), but the artist in me is picky. ;)
The hue of the roses on google is that same kind of orange-ish red -
again - when compared to the real thing... If I rest my eyes and
don't look at the real thing for a while my interpretation is good.
I like it. But I want better! :)
Thoughts? Am I going down the rabbit hole here? Is it possible to
profile a rose petal (perhaps using my X-rit spectrophotometer?) To
nail down what it's doing chromatically? So that then I could at
least quantify the limitation here - to see where the rose exceeds
my devices capabilities?
Sorry for the length of this post, and thanks for reading.
Regards,
Sean
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden