Frustration at not being able to get excellent results in outdoor conditions
1 view
Skip to first unread message
Juan
Nov 21, 2008, 11:46:27 AM11/21/08
to AcrCalibrator
Hi,
First: I have ColorChecker and MiniColorChecker and I understand they
correspond to the same LAB values - is this true?
Second: I have, in controlled conditions (using a softbox at about 15
feet away from the target, the light perpendicular to the target, and
my camera placed at the exact same height as the center of the target
and also about 15 feet away but making a 25-35 degree angle with
respect to the softbox light direction) achieved DE 2000 of 1.33 which
is as I understand very close to exceptional. This is all very nice,
but when I take my target (either) outdoors on a sunny day, a
Third: what geometrical arrangement should I use to place my target
with respect to the sun (rays perpendicular to target?), and how or
where and how far should I place my camera? In looking for uniform
lighting I have come to the conclusion that the sun's rays must be
perpendicular to the target and that, since the camera cannot be
perpendicular to the target, then the closest approximation is to take
an angle (perhaps 25-35 degrees or what is your suggestion on this?)
and to move the camera away far enough so that the difference in the
angle made with respect to one side of the target and the angle made
with respect to the other farther away side, is minimized. Also, when
these two angles are almost the same then the sun's reflection from
each extreme should be very similar. Is this line of reasoning
correct? Should I instead place the target to make an equal angle to
the sun and to my camera? The problem I see with these is a lot of
glare in the image... Not sure though...
Fourth: if the sun is fully shining with no clouds beneath it, does it
matter is there are white clouds (not stormy or dark) in other places
in the sky?
Fifth: I keep getting very good matches in the hues of RGB patches but
the saturation constantly becomes a problem - the script has a lot of
trouble getting a good value for the RGB saturation. Why is this and
what can I do to obtain saturation values closer to ideal?
Sixth: I have been using the ACR_Calibrator_CC24 V6.4 by Rags Gardner,
which is also based on Fors' script, and my camera is Canon EOS 1Ds
MIII.
Thanks in advance to any responses. I have tried to be very precise in
my questions and what I have tried, in the hope that this might make
it easier to answer.
Regards,
Juan Dent
First: I have ColorChecker and MiniColorChecker and I understand they
correspond to the same LAB values - is this true?
Second: I have, in controlled conditions (using a softbox at about 15
feet away from the target, the light perpendicular to the target, and
my camera placed at the exact same height as the center of the target
and also about 15 feet away but making a 25-35 degree angle with
respect to the softbox light direction) achieved DE 2000 of 1.33 which
is as I understand very close to exceptional. This is all very nice,
but when I take my target (either) outdoors on a sunny day, a
Third: what geometrical arrangement should I use to place my target
with respect to the sun (rays perpendicular to target?), and how or
where and how far should I place my camera? In looking for uniform
lighting I have come to the conclusion that the sun's rays must be
perpendicular to the target and that, since the camera cannot be
perpendicular to the target, then the closest approximation is to take
an angle (perhaps 25-35 degrees or what is your suggestion on this?)
and to move the camera away far enough so that the difference in the
angle made with respect to one side of the target and the angle made
with respect to the other farther away side, is minimized. Also, when
these two angles are almost the same then the sun's reflection from
each extreme should be very similar. Is this line of reasoning
correct? Should I instead place the target to make an equal angle to
the sun and to my camera? The problem I see with these is a lot of
glare in the image... Not sure though...
Fourth: if the sun is fully shining with no clouds beneath it, does it
matter is there are white clouds (not stormy or dark) in other places
in the sky?
Fifth: I keep getting very good matches in the hues of RGB patches but
the saturation constantly becomes a problem - the script has a lot of
trouble getting a good value for the RGB saturation. Why is this and
what can I do to obtain saturation values closer to ideal?
Sixth: I have been using the ACR_Calibrator_CC24 V6.4 by Rags Gardner,
which is also based on Fors' script, and my camera is Canon EOS 1Ds
MIII.
Thanks in advance to any responses. I have tried to be very precise in
my questions and what I have tried, in the hope that this might make
it easier to answer.
Regards,
Juan Dent
steve sprengel
Nov 21, 2008, 3:10:15 PM11/21/08
to acrcal...@googlegroups.com
Something is missing from your initial list of questions in that it jumps from the middle of a sentence ending with "a" in First to the beginning of Third.
I believe the two targets are supposed to be the same as far as colors. I wouldn't be surprised if they use the same pigments for each.
The matte finish of the target will pick up color from anywhere in front of it, so trees and grass and blue sky and sun and your clothes and the color of buildings, etc.
There is no single correct answer to what angle the target and lightsource and camera should be although there are probably some obviously wrong configurations: no glare, no shadows, no uneven illuminations, minimize envronmental reflections unless they will be significantly contributing to the overall lighting in the actual situation--tilt up toward the sky more than down toward the grass (unless your subject is very close to the grass or tree or green plants so their skin will and away from, not toward a large green tree or brick house, etc. Realize that the lighting that you are optimizing for is usually a mixture of sky and sun. To minimize glare I try to make the angle of the camera-target-sun so that if the target was a mirror then the imagined reflection would only be blue sky away from any sun and away from any trees or buildings.
Outdoors I usually do things in one of two ways, either:
1) With the sun relatively high, I lay the target flat on the pavement/sidewalk, away from any trees or buildings, and usually shoot it somewhere between 45 and 60 (closer to over-top) and have the sun over to one side quite a ways just to minimize any glare. I have it pointing straight up so I only have sky and sun light on it, and no grass or non-neutral pavement coloring the light. If it's cloudy then the angle to the sun isn't important, but any environmental things like trees and buildings are so I'd want the open cloudy-sky as
2) With the sun relatively low, I hold the target in one hand at a 45-to-60-degree angle (more vertical than horizontal) with the sun over my shoulder 30 degrees to the side to keep my shadow out of it and hold the camera and take the picture with the other hand. I use this method for low sun because the color of the sun will be different and there will likely be glare coming off the ground/grass/pavement between the sun and the subject that will color the light even more so I can to accurately measure and calibrate for this color.
Another special outdoor situation would be within a forest where the light is diffuse and colored by the greenery around or a variation of this would be in a flower garden where there are lots of plants and many different colors of flowers. You would want to probably keep the target more vertical to pick up the green being reflected from the ground as much as the blue from the sky. Of course this doesn't apply if you are using a large reflector below, so the reflected sky and sun is what is illuminating the face. The same sort of thing might be true if your subject was standing on any very non-neutral-colored surface such as a brick path, or next to a brick building, although the red from bricks might just make the skin look better, so wouldn't need to be accomidated for, but green grass or forest folage are another story--you don't want the skin to look green, usually.
In a studio situation, which means I have control over all the lighting and the environment (walls floor ceiling) is neutral (black/gray/white) colored, and where I am calibrating for the particular lighting, specifically, I would face the target relatively perpendicular to (flat to) the camera (perhaps slightly off just to keep any reflections from me out of it) and have the lighting at wherever angles they are going to be at. This assumes the part of the subject or object facing the camera is the most important part. If I were doing something odd, such as having the key light very far over to the side, I'd probably want to angle the target more toward the key light just so it was still the main light source being balanced for.
Nowadays, in a studio-lighting situation, instead of doing a custom calibration for each lighting setup, I would create a custom DNG camera profile (LR 2+, PS CS3+ with latest ACR) for each light source. I would create it with both color-tables using the same CC photo so no interplation would be occuring.
And also, nowadays, for non-studio things, I have a two light-source (Incandescent and Cloudy) DNG Camera Profile I have created with the two color tables based on two different RAWs and don't use the camera calibration sliders, nor these scripts, anymore.
One unfortunate thing about the ACR/LR profiles you're trying to calibrate against, is that they are dual light-source profiles that are for approximately 2700K (Tungsten) and 6500K (Cloudy) and the effective camera profile that is being used for a particular white-balance in between those two color temperatures will be an interpolation of those two internal profiles, and you have to have a different calibration for each WB between those two color temperatures--at least have to have several. If your WB is at or below 2700K or at or above 6500K then only one profile would be used and a single calibration would be valid for a wider range of WBs. In practice I have found that Sunny (5500K and above calibrations are very similar so I would tend to use a Cloudy (6500K) calibration for sunny and bluer, and only worry about multiple calibrations for lighting redder than Sunlight, except for very blue lighting, such as moonlight or twilight.
I wouldn't use Rags' or Fors' scripts because the only use three of the color patches (R, G, B) to calibrate the color againt. I would use Tindemans' where you can have it use all the color patches.
I'm not sure what you're saying regarding saturation, but maybe it is that saturation is a function of contrast so if your contrast or black point or contrast (toning) curve settings are different between your calibration and your real-world shots then your real-world shots will have a different saturation compared to the calibration shots. If you always use the ACR defaults for your RAW conversions then you can calibrate to those and have less change, but the calibration default for the toning curve is something non-linear which makes the contrast of midtones higher than the contrast of darks and lights so calibrating using only three mid-tone RGB patches will make the shadows and highlights less contrasty and look less saturated, although perhaps that is what you'd usually want to occur. I know Fors' script resets the toning curve to linear and so using any Fors calibratoin with a non-linear toning curve will have overly saturated colors. I usually like this look, as my shots are nature not people, but your situation may be different.
I believe the two targets are supposed to be the same as far as colors. I wouldn't be surprised if they use the same pigments for each.
The matte finish of the target will pick up color from anywhere in front of it, so trees and grass and blue sky and sun and your clothes and the color of buildings, etc.
There is no single correct answer to what angle the target and lightsource and camera should be although there are probably some obviously wrong configurations: no glare, no shadows, no uneven illuminations, minimize envronmental reflections unless they will be significantly contributing to the overall lighting in the actual situation--tilt up toward the sky more than down toward the grass (unless your subject is very close to the grass or tree or green plants so their skin will and away from, not toward a large green tree or brick house, etc. Realize that the lighting that you are optimizing for is usually a mixture of sky and sun. To minimize glare I try to make the angle of the camera-target-sun so that if the target was a mirror then the imagined reflection would only be blue sky away from any sun and away from any trees or buildings.
Outdoors I usually do things in one of two ways, either:
1) With the sun relatively high, I lay the target flat on the pavement/sidewalk, away from any trees or buildings, and usually shoot it somewhere between 45 and 60 (closer to over-top) and have the sun over to one side quite a ways just to minimize any glare. I have it pointing straight up so I only have sky and sun light on it, and no grass or non-neutral pavement coloring the light. If it's cloudy then the angle to the sun isn't important, but any environmental things like trees and buildings are so I'd want the open cloudy-sky as
2) With the sun relatively low, I hold the target in one hand at a 45-to-60-degree angle (more vertical than horizontal) with the sun over my shoulder 30 degrees to the side to keep my shadow out of it and hold the camera and take the picture with the other hand. I use this method for low sun because the color of the sun will be different and there will likely be glare coming off the ground/grass/pavement between the sun and the subject that will color the light even more so I can to accurately measure and calibrate for this color.
Another special outdoor situation would be within a forest where the light is diffuse and colored by the greenery around or a variation of this would be in a flower garden where there are lots of plants and many different colors of flowers. You would want to probably keep the target more vertical to pick up the green being reflected from the ground as much as the blue from the sky. Of course this doesn't apply if you are using a large reflector below, so the reflected sky and sun is what is illuminating the face. The same sort of thing might be true if your subject was standing on any very non-neutral-colored surface such as a brick path, or next to a brick building, although the red from bricks might just make the skin look better, so wouldn't need to be accomidated for, but green grass or forest folage are another story--you don't want the skin to look green, usually.
In a studio situation, which means I have control over all the lighting and the environment (walls floor ceiling) is neutral (black/gray/white) colored, and where I am calibrating for the particular lighting, specifically, I would face the target relatively perpendicular to (flat to) the camera (perhaps slightly off just to keep any reflections from me out of it) and have the lighting at wherever angles they are going to be at. This assumes the part of the subject or object facing the camera is the most important part. If I were doing something odd, such as having the key light very far over to the side, I'd probably want to angle the target more toward the key light just so it was still the main light source being balanced for.
Nowadays, in a studio-lighting situation, instead of doing a custom calibration for each lighting setup, I would create a custom DNG camera profile (LR 2+, PS CS3+ with latest ACR) for each light source. I would create it with both color-tables using the same CC photo so no interplation would be occuring.
And also, nowadays, for non-studio things, I have a two light-source (Incandescent and Cloudy) DNG Camera Profile I have created with the two color tables based on two different RAWs and don't use the camera calibration sliders, nor these scripts, anymore.
One unfortunate thing about the ACR/LR profiles you're trying to calibrate against, is that they are dual light-source profiles that are for approximately 2700K (Tungsten) and 6500K (Cloudy) and the effective camera profile that is being used for a particular white-balance in between those two color temperatures will be an interpolation of those two internal profiles, and you have to have a different calibration for each WB between those two color temperatures--at least have to have several. If your WB is at or below 2700K or at or above 6500K then only one profile would be used and a single calibration would be valid for a wider range of WBs. In practice I have found that Sunny (5500K and above calibrations are very similar so I would tend to use a Cloudy (6500K) calibration for sunny and bluer, and only worry about multiple calibrations for lighting redder than Sunlight, except for very blue lighting, such as moonlight or twilight.
I wouldn't use Rags' or Fors' scripts because the only use three of the color patches (R, G, B) to calibrate the color againt. I would use Tindemans' where you can have it use all the color patches.
I'm not sure what you're saying regarding saturation, but maybe it is that saturation is a function of contrast so if your contrast or black point or contrast (toning) curve settings are different between your calibration and your real-world shots then your real-world shots will have a different saturation compared to the calibration shots. If you always use the ACR defaults for your RAW conversions then you can calibrate to those and have less change, but the calibration default for the toning curve is something non-linear which makes the contrast of midtones higher than the contrast of darks and lights so calibrating using only three mid-tone RGB patches will make the shadows and highlights less contrasty and look less saturated, although perhaps that is what you'd usually want to occur. I know Fors' script resets the toning curve to linear and so using any Fors calibratoin with a non-linear toning curve will have overly saturated colors. I usually like this look, as my shots are nature not people, but your situation may be different.
On Fri, Nov 21, 2008 at 10:46 AM, Juan <juan...@hotmail.com> wrote:
Hi,
First: I have ColorChecker and MiniColorChecker and I understand they
correspond to the same LAB values - is this true?
Second: I have, in controlled conditions (using a softbox at about 15
feet away from the target, the light perpendicular to the target, and
my camera placed at the exact same height as the center of the target
and also about 15 feet away but making a 25-35 degree angle with
respect to the softbox light direction) achieved DE 2000 of 1.33 which
is as I understand very close to exceptional. This is all very nice,
but when I take my target (either) outdoors on a sunny day, a
<missing text>
Juan
Nov 21, 2008, 7:14:53 PM11/21/08
to AcrCalibrator
HI,
Thanks for such a complete response. I have some doubts about your
suggestions.
when you say "usually shoot it somewhere between 45 and 60 (closer to
precise as to what you mean by have the sun over to one side?
Say your camera makes an angle of 45-60 with respect to the ground and
say south (just for reference) of the target, does the sun need to be
north or towards the east or west (whichever side the sun happens to
be)?
How high over the horizon do you mean by "the sun relatively high"?
More than 30 or 45 degrees above horizon?
Now, if environmental things like trees, are going to be in front of
the subject, then shouldn´t we place the target to include those trees
- in order to really get the same lighting that the subject will get?
Now, the other point: "With the sun relatively low, I hold the target
ask but how are you measuring your angle? Do you mean between the sun
rays and the target surface? In that case, placing the target
perpendicular (fully facing the sun) would make a 90 degree angle. If
I interpret this correctly, then 45-60 angle will make the target
approximately face the horizon assuming a low sun. This is a concern
because it would take light from the ground as well as the sun and
then we are back at my previous question: should we face the target at
all light emitting (or reflecting) sources that will also reach my
subject?
Also, when you say: "bluer, and only worry about multiple calibrations
profiles specific for these uncontrolled lighting, not trusting the 2
profile system that Adobe has come up with? You say calibrations but I
guess you meant DNG profiles, right?
And, "I'm not sure what you're saying regarding saturation, but maybe
tone should be linear in order to get more precise calibrations with
the scripts and I always do the calibrations with linear tone curve.
Yet, I see the scripts having great difficulty in reaching a
satisfactory saturation. I don´t quite understand this??
Finally, and probably the most important question: what ProPhoto RGB
or LAB values should I expose the white patch for? Is it 241,241,241?
Again, thanks a million for your preciseness and patience.
Best regards,
Juan Dent
Thanks for such a complete response. I have some doubts about your
suggestions.
when you say "usually shoot it somewhere between 45 and 60 (closer to
over-top) and have the sun over to
one side quite a ways just to minimize any glare" , can you be more
precise as to what you mean by have the sun over to one side?
Say your camera makes an angle of 45-60 with respect to the ground and
say south (just for reference) of the target, does the sun need to be
north or towards the east or west (whichever side the sun happens to
be)?
How high over the horizon do you mean by "the sun relatively high"?
More than 30 or 45 degrees above horizon?
Now, if environmental things like trees, are going to be in front of
the subject, then shouldn´t we place the target to include those trees
- in order to really get the same lighting that the subject will get?
Now, the other point: "With the sun relatively low, I hold the target
in one hand at a 45-to-60-degree angle (more vertical than horizontal)
with the sun over my shoulder 30 degrees to the side", I am sorry to
ask but how are you measuring your angle? Do you mean between the sun
rays and the target surface? In that case, placing the target
perpendicular (fully facing the sun) would make a 90 degree angle. If
I interpret this correctly, then 45-60 angle will make the target
approximately face the horizon assuming a low sun. This is a concern
because it would take light from the ground as well as the sun and
then we are back at my previous question: should we face the target at
all light emitting (or reflecting) sources that will also reach my
subject?
Also, when you say: "bluer, and only worry about multiple calibrations
for lighting redder than Sunlight, except for very blue lighting, such
as moonlight or twilight", do you mean that you create several DNG
profiles specific for these uncontrolled lighting, not trusting the 2
profile system that Adobe has come up with? You say calibrations but I
guess you meant DNG profiles, right?
And, "I'm not sure what you're saying regarding saturation, but maybe
it is that saturation is a function of contrast so if your contrast or
black point or contrast (toning) curve settings are different between
your calibration". What do I make of this? It has been told that the
black point or contrast (toning) curve settings are different between
tone should be linear in order to get more precise calibrations with
the scripts and I always do the calibrations with linear tone curve.
Yet, I see the scripts having great difficulty in reaching a
satisfactory saturation. I don´t quite understand this??
Finally, and probably the most important question: what ProPhoto RGB
or LAB values should I expose the white patch for? Is it 241,241,241?
Again, thanks a million for your preciseness and patience.
Best regards,
Juan Dent
Steve Sprengel
Nov 24, 2008, 11:35:38 PM11/24/08
to acrcal...@googlegroups.com
By "over to the side or "around to the side' I mean
at a different "heading" or "compass direction" than directly in front or
behind. Using the old-fashioned clock-face analogy, if the clock face is
flat on the ground, and the CC is in the center, and I am at the 6-o'clock
position, facing the center and hence facing the 12-o'clock position, then in
the example of the sun being high but in front of me (as if I am facing
generally south in the northern hemisphere) then over to the side would be
somewhere between the 10:00 and 11:00 or the 1:00 and 2:00 positions or between
30-60 degrees either counter-clockwise or clockwise. Directly in front of
me would be the 12-o'clock position and I would assume too much glare would be
reflecting to make an accurate reading. In the case when the sun is low
and over my shoulder then it would be between the 4:30 to 5:30 or 6:30 to 7
position.
By the sun being relatively high I mean closer to
mid-day than sun-rise or sunset.
If the trees in front of the subject are providing
most of the light, and the sun, nor blue sky are not, then sure, use the target
at the subject position facing somewhat upward, but the sun and even a cloudy
sky is quite a bit brighter than what would be reflected by nearby trees that it
would likely be a special case, unless you are actually under a tree canopy in a
forest. Of course, determining the correct white-balance is much more
important than having exactly the right calibration. You can test your
theory about whether the trees are messing much with the calibration by doing
one calibration with the CC facing the camera and one with is facing more upward
so the sky and sun are more of the light compared to the trees and see how much
different the calibration is.
As far as multiple calibrations for redder than
sunlight or very blue like twilight then I mean multiple calibrations for redder
than sunlight, assuming you are already using a dual 2650K-6500K profile, and
could mean either multiple calibrations or multiple DNG Profiles for something
very blue, which won't be covered in the range of a dual profile and is much
bluer than then Cloudy 6500K.
I wasn't quite sure what you mean when you say
"scripts having great difficulty reaching satisfactory saturation" so I just
made general comments saturation issues that you can have when you use a linear
curve to compute a profile but using something other than linear when you
actually develop. Maybe you can explain more what you mean and I can say
something else or may not actually know what to say.
If you look in the script you will see what the
white-patch values are in ProPhotoRGB and I don't know what those are off
hand. I wouldn't worry so much about the exposure being exactly 241, 241,
241 though if the whole thing is too bright or too dark. If you are
letting the script optimize the neutral-patches then it should do what you want,
otherwise maybe having the midtone close to what it's supposed to be is better
than having one end what it should be.
Juan
Nov 25, 2008, 7:15:03 PM11/25/08
to AcrCalibrator
Hi Steve and thanks again.
My comment on saturation and the scripts is that they here is where
the most deviation (largest errors) occur. By contrast, the hue
quickly reach the goals set in the dialog box of the script (and excel
them!).
I was very grateful for your careful explanation of your "over to the
side". In that regard, I wish you could confirm my interpretation:
1) I think you have said (please confirm) that as long as the sun
is high enough, the CC target is to be placed flat on the ground,
facing the sky
2) I also interpret that when the sun is low, the CC target is to
be placed more near vertical (although not completely vertical) so as
to look towards the low sun
Does it matter if the camera is not facing the CC target exactly
perpendicularly?
Does it matter if the sun is not facing the CC target exactly
perpendicularly?
On the subject of evaluating CC shots: how can we measure the shot´s
quality objectively? The scripts give some idea of how much of a good
job one has done, but the DNG profile editor does not give any
feedback as to how good (i.e. evenly lit, free of color casts, etc.) a
shot of the CC target we have done!
Again, much appreciated.
Best regards,
Juan Dent
> ...
>
> read more »
My comment on saturation and the scripts is that they here is where
the most deviation (largest errors) occur. By contrast, the hue
quickly reach the goals set in the dialog box of the script (and excel
them!).
I was very grateful for your careful explanation of your "over to the
side". In that regard, I wish you could confirm my interpretation:
1) I think you have said (please confirm) that as long as the sun
is high enough, the CC target is to be placed flat on the ground,
facing the sky
2) I also interpret that when the sun is low, the CC target is to
be placed more near vertical (although not completely vertical) so as
to look towards the low sun
Does it matter if the camera is not facing the CC target exactly
perpendicularly?
Does it matter if the sun is not facing the CC target exactly
perpendicularly?
On the subject of evaluating CC shots: how can we measure the shot´s
quality objectively? The scripts give some idea of how much of a good
job one has done, but the DNG profile editor does not give any
feedback as to how good (i.e. evenly lit, free of color casts, etc.) a
shot of the CC target we have done!
Again, much appreciated.
Best regards,
Juan Dent
>
> read more »
Steve Sprengel
Nov 26, 2008, 2:40:39 PM11/26/08
to acrcal...@googlegroups.com
"The CC target *is to be* placed flat on the
ground" is a bit strong. My yard has big trees, but south of me there
aren't so I typically place my CC flat on the ground and stand north of
it, with the sun south of it, over to the side, and take my calibration
shot. In a different location I might prop it up a bit at and angle with
the important thing in whatever the situation is, is that the sky is what would
be a reflection if the CC was a mirror instead of a matte-finish color-target,
and that the sun is not glaring directly into the camera from the target, which
can be judged by the glare not coming from the shiny black border area between
the colored squares.
Except when the sun is low and the glare from the
ground is also coloring the light, then I hold the target so it
is almost
glaring from the sun and the ground. It probably depends a bit on the
orientation of your subject in this case, though. If you are photographing
a person near camera height their face toward the sun, then having the target
also mostly facing the sun would seem right. However, if you are
photographing someone laying down on the ground with the sun hardly illuminating
them at all, and the light is mostly coming from the darkening blue sky then it
is more like shade lighting and I'd do that calbratoin with the target facing
mostly upwards like their face is.
As far as perpendicular, the target and the sun
never are for me, and if they both were, then my shadow would be on the
target. It probably doesn't hurt to have the sun perpendicular to the
target, but the camera probably shouldn't be, at least outside where the light
is coming from above, because the camera and photographer would likely be
blocking quite a bit of the sky from the target and giving the target more of a
reddish sun-color compared to the blue-sky color where "sunny" should be a
mixture of the two. About the only time I can think when the camera and
the target are relatively perpendicular is indoors in the studio where there is
balance lighting coming from both sides, and the model is holding the target
mostly toward the camera, although it still wouldn’t hurt to have it angled up
slightly since the lighting is coming from slightly above--in the normal case,
at least. Both the scripts and the DNG Profile Editor can handle a
non-rectangular shape so having the camera perpendicular is not
necessary.
I don't know if your saturation-error experience is
different than mine or not, except to say that if your numbers are ending up at
-100 or +100 then something is wrong. It probably depends, somewhat, on
the particular camera. the particular color of light, the age and condition
of the color-checker you're using, and the emphasis Adobe chose to put on
various aspects of the color-profile you are calibrating against. If you have
used LR 2 or ACR 4.6+ with the Adobe's camera-matching profiles (now just
released from being non-beta) then you will see how much variation there is
between different ones that try to emphasis certain colors, either to make them
more saturated or to make them more accurate. I have my default
calibration in Lightroom and ACR set to a dual-color custom profile that I've
named 2700K-6500K (for obviously reasons) but sometimes I switch to one of the
camera-matching profiles: Neutral or Landscape or Portrait, depending on
which I like the best for the colors in the shot. The Landscape profile
typically emphases certain colors that make outdoor shots without people look
more like the bright colors of a picture-postcard postcard but would make people
look much too colorful so I don't use it with indoor shots or portraits,
usually.
As far as measuring the accuracy of the
calibration, the numbers you get out of the scripts as their running are for
just the patches being compared, which for Tom and Rags are only three, and not
the whole chart, but I think Rags has a script that looks at all the colors and
gives some overall number and perhaps per-patch numbers, although I'm not
exactly sure.
I use the less-expensive (but still not cheap)
version of Imatest for gauging color accuracy of my calibrations or
profiles. It gives several overall numbers representing color-error
measured in several different ways, plus it gives a chart that shows the
direction and size of the error for each color patch. It even has a way to
look at a 3D model of the color-error. The color-checker image does need
to be relatively rectangular, so I usually have to do some perspective
correction in Photoshop before giving the chart to Imatest to analyze.
During its analysis it will warn if the illumination is uneven and one of the
plots shows the line across the neutral squares where if you see the tops of the
"bars" as slanted, then you know there was vignetting or some other source of
unevenness, and another one of the plots shows the exaggerated color of the
neutral strip so you can see what colro casts there might have been, or how
non-neutral the squares actually are. I think there is a trial version of
Imatest you can download. The UI is a little weird because the program can
do all sorts of fancy anlaysis and the developer wanted the best math libraries
he could get, which didn't have the best UI, but it is usable. Imatest,
itself, does all sorts of other analysis about lens sharpness and noise levels
in images and things so that's why it is so expensive, where I only use it for
the color-error because I don't want to spend the money on the professional test
charts you need to gauge sharpness from.
You have, no doubt, seen these Imatest color-error
plots on my webpages for:
the ACR Calibrator script color error:
and the DNG Profile Editor color
error:
The main takeaway is that the
colors won't be perfect no matter how well you profile so don't be overly
obsessed with it, except for the fun of it.
Juan
Nov 26, 2008, 3:35:52 PM11/26/08
to AcrCalibrator
HI Steve,
In case you are online: you say camera-matching profiles are now just
released from being non-beta, do you have a link where to download
them? I haven't found them...
Thanks for the Imatest suggestion.
One last question (for a while): targets shot under tungsten lighting,
are they incapable of being as accurate as shots in daylight/cloudy/
softbox (5500K-6500K)?
I ran Rags script on 2 shots at 2650K and their DE 2000 error is
around 4.0 - horrible!!
Is this to be expected because the light is just not as complete?
Thanks!
Best regards,
Juan
> ...
>
> read more »
In case you are online: you say camera-matching profiles are now just
released from being non-beta, do you have a link where to download
them? I haven't found them...
Thanks for the Imatest suggestion.
One last question (for a while): targets shot under tungsten lighting,
are they incapable of being as accurate as shots in daylight/cloudy/
softbox (5500K-6500K)?
I ran Rags script on 2 shots at 2650K and their DE 2000 error is
around 4.0 - horrible!!
Is this to be expected because the light is just not as complete?
Thanks!
Best regards,
Juan
>
> read more »
Steve Sprengel
Nov 26, 2008, 9:42:40 PM11/26/08
to acrcal...@googlegroups.com
The new, non-beta Adobe camera-matching
profiles install when you install the latest DNG Converter and perhaps CS4's/ACR
5.2. I became aware of this information because I subscribe to the
messages from the Adobe Lightroom Forums where someone posted a message
there:
with a link to the Lightroom Journal blog about
it:
http://blogs.adobe.com/lightroomjournal/2008/11/camera_raw_52_and_dng_converte.html
Which links to the Featured Updates page
where you have to change the drop-down-list at the
top to DNG Converter to get just it, and not ACR along with it unless you have
CS4 then the combined one is good. LR 2.2 will be out in a week or two, I
guess, but the profiles are visible in 2.1, just not the new camera support,
yet.
[That first part was easy, but you asked a question that requires more
explanation so I am taking more time to write it.]
As far as the error with incandescent lighting, I
think this has to do with the particular colors of the color-checker as being
across the entire spectrum and the scripts' internal "ideal" standard CC values
are for either D50 (Sunny) or D65 (Cloudy) lighting (I forget), but not for
Illuminant A (tungsten).
The main issue photographing something with
many colors (like the CC) under lighting with bias towards a particular set of
colors (reddish ones for tungsten) is that a colored light will neutralize
colors that are similar to it once you white balance for the color of the
lighting. White-balancing is basically draining the color of the light
from all the colors in the image, which will undersaturate the colors that are
similar to the color of the light and oversaturate the colors that are opposite
it and this will make the color error among the various color-patches uneven as
compared to the ideal D50 (or D65) standard values. A wild-card in all of
this is that Adobe's Incandescent end of their dual profile is likely optimized
for skin tones so it is hard to guess where the color-error will be but there
will likely be more for Incandescent than the Sunny/Cloudy because the
color-error is being computed with respect to the D50/D65 color values, not
Illuminant-A color values. It would be impossible to use these scripts to
compute an actual color profile because the internal standard is for D50 or D60
lighting only, but you can assume that the color profile Adobe provides has
already taken the lighting color into account--after all they can compute their
own standard values for the CC patches in different lighting using an actual
spectrophotometer and shouldn't have to rely on published values. So a
calibration script is merely correcting values that should be mostly corrected
for the lighting color already. However, as was previously said, the
Incandescent profile is likely optimized for skin tones so there is more
color-error in it, overall, and the scripts just can't neutralize it all.
You can assume that if you use the DNG Profile Editor and create a dual-color
profile that Adobe has used the ideal values for 2650K when you compute that
color-table and 6500K when you compute that color table so your DNG Profile will
likely be closer than the Adobe Standard that should be optimized for skin
tones.
The one lesson I can learn from doing my own calibrations is that I want to
include Yellow along with red, green, and blue, when I calibrate for
incandescent because yellow has been the most off after I calibrate if I
don't.
Look at the 2700K+1 color-error sequence and notice how far off the "top"
yellowish colors are after running Tom Fors' script -- the third panel over, as
compared to how relatively close things are if I optimize the color error in RGB
and CMY -- the fourth panel over:
You will notice how close I got using the DNG profile editor for 2700K
lighting, where mostly the Red square is a bit oversaturated (circle further
away from the center than the square), using either the DNG Profile editor dual
profile on the 2700K end, or a single color table profile made specifically for
2700K:
http://www.pbase.com/ssprengel/image/101322979 (2700K
end of a dual-color-table profile)
http://www.pbase.com/ssprengel/image/101308595 (both
color tables set to 2700K)
In the 6500K sequence you will notice the Tindemans got better results than
the DNG Profile Editor, but you will also notice that the circles directly in
the middle (which correspond to the neutral row) aren't aligned above each
other, so there was likely some color-cast that affected the dark areas
differently than the light areas, but 6500K is hard to get, exactly, using only
sun and clouds outdoors (which is all I have) so I haven't been able to reshoot
to see if that helps make the neutrals more in line.
Reply all
Reply to author
Forward
0 new messages