Something about Coloraccuracy

[KIM MIN-SU]

I agree your opinions,
The noise is so critical issue for me and others.
but, I want to say about the primary color performance.
I got a sampe photo in my room.



captured with My SD9
In case of sd9, more accuracy mixed color



captured with My S2Pro
In case of s2pro, more accuracy color

can't display the clean red color in my photo taken by SD9.
I have been tried to get the clearn red, blue and green color.
but, in lighter area, color is changed to the other color feeling.
and in dakrer area, color is red, but too dark and dirty,

Ok,
I can undertand your opinions.
and I have a TRV2000 (sony, 3CCD DV) and I have been test the same
image without trouble.
Color temperature is not same as the target via media such as film,
CCD, CMOS or others.

BUT, I want to say about

(1) Abnormal color changing via intencity
(2) Abnormal gray color changing in high intencity area.

AND, I want to say about

(1) How can understand about the color difference
between low intencity area and high intencity area.

Check this sample image



It's a test pattern to check the color



It's a decompressed raw data.
In this test, can't found the color change. OK?



It's a relative intencity of sensor information with green light
source



It's a relative intencity of sensor information with red light source



It's a relative intencity of sensor information with blue light source

The relative intencity via light level and properties were measured
with the colored light source have no problems.
but, color changing is a problems (be occured by the color
separation process).

Color problems is appreared only in HIGHER light level.
To avoid this problems, Default light level of SD9 is so low.

I am falling in puzzle,
(1) to get the color, I overcome the noise problems
(2) to get the clean image, I overcome the color problems.

27-70mm/2.8, ISO100, 1/10S, F4.0
color temperature of source = 6800k

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

--
SD9 with 24-70DG2.8

 

[KIM MIN-SU]

^^

I have been used two camcoder

1. trv900 (old 3ccd model)
2. vx2000 (old but good 3ccd model)

model name is vx2000, trv2000 is a miss spell, sorry.

captured with My SD9
In case of sd9, more accuracy mixed color



captured with My S2Pro
In case of s2pro, more accuracy color

can't display the clean red color in my photo taken by SD9.
I have been tried to get the clearn red, blue and green color.
but, in lighter area, color is changed to the other color feeling.
and in dakrer area, color is red, but too dark and dirty,

Ok,
I can undertand your opinions.
and I have a TRV2000 (sony, 3CCD DV) and I have been test the same
image without trouble.
Color temperature is not same as the target via media such as film,
CCD, CMOS or others.

BUT, I want to say about

(1) Abnormal color changing via intencity
(2) Abnormal gray color changing in high intencity area.

AND, I want to say about

(1) How can understand about the color difference
between low intencity area and high intencity area.

Check this sample image



It's a test pattern to check the color



It's a decompressed raw data.
In this test, can't found the color change. OK?



It's a relative intencity of sensor information with green light
source



It's a relative intencity of sensor information with red light source



It's a relative intencity of sensor information with blue light source

The relative intencity via light level and properties were measured
with the colored light source have no problems.
but, color changing is a problems (be occured by the color
separation process).

Color problems is appreared only in HIGHER light level.
To avoid this problems, Default light level of SD9 is so low.

I am falling in puzzle,
(1) to get the color, I overcome the noise problems
(2) to get the clean image, I overcome the color problems.

27-70mm/2.8, ISO100, 1/10S, F4.0
color temperature of source = 6800k

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

--
SD9 with 24-70DG2.8

--
SD9 with 24-70DG2.8

 

[Dominic Groß]

The S2Pro routines sometime just produce too rich color (Laurence mentioned it in another thread as well). But nothin is unfixable with Photoshop, especially if it concerns a rich color. Here is my version:



I think the "wood" of the Hifi system looks better now. And it looks better than the S2Pro picture for sure, it has wrong WB and the midtones look strange.

captured with My SD9
In case of sd9, more accuracy mixed color



captured with My S2Pro
In case of s2pro, more accuracy color

can't display the clean red color in my photo taken by SD9.
I have been tried to get the clearn red, blue and green color.
but, in lighter area, color is changed to the other color feeling.
and in dakrer area, color is red, but too dark and dirty,

Ok,
I can undertand your opinions.
and I have a TRV2000 (sony, 3CCD DV) and I have been test the same
image without trouble.
Color temperature is not same as the target via media such as film,
CCD, CMOS or others.

BUT, I want to say about

(1) Abnormal color changing via intencity
(2) Abnormal gray color changing in high intencity area.

AND, I want to say about

(1) How can understand about the color difference
between low intencity area and high intencity area.

Check this sample image



It's a test pattern to check the color



It's a decompressed raw data.
In this test, can't found the color change. OK?



It's a relative intencity of sensor information with green light
source



It's a relative intencity of sensor information with red light source



It's a relative intencity of sensor information with blue light source

The relative intencity via light level and properties were measured
with the colored light source have no problems.
but, color changing is a problems (be occured by the color
separation process).

Color problems is appreared only in HIGHER light level.
To avoid this problems, Default light level of SD9 is so low.

I am falling in puzzle,
(1) to get the color, I overcome the noise problems
(2) to get the clean image, I overcome the color problems.

27-70mm/2.8, ISO100, 1/10S, F4.0
color temperature of source = 6800k

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

--
SD9 with 24-70DG2.8

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

 

[KIM MIN-SU]

Thanks for your kind working.

"wood" is ok.
and "clock" is almost ok, but not clear.
and unatural red color on back ground image.

And I want to say that
S2pro's photo is more same as my eye's viewing
(In the sample)

And I gree, S2Pro produce too rich color, BUT,
SD9 produce the abnormal color

I think the "wood" of the Hifi system looks better now. And it
looks better than the S2Pro picture for sure, it has wrong WB and
the midtones look strange.

captured with My SD9
In case of sd9, more accuracy mixed color



captured with My S2Pro
In case of s2pro, more accuracy color

can't display the clean red color in my photo taken by SD9.
I have been tried to get the clearn red, blue and green color.
but, in lighter area, color is changed to the other color feeling.
and in dakrer area, color is red, but too dark and dirty,

Ok,
I can undertand your opinions.
and I have a TRV2000 (sony, 3CCD DV) and I have been test the same
image without trouble.
Color temperature is not same as the target via media such as film,
CCD, CMOS or others.

BUT, I want to say about

(1) Abnormal color changing via intencity
(2) Abnormal gray color changing in high intencity area.

AND, I want to say about

(1) How can understand about the color difference
between low intencity area and high intencity area.

Check this sample image



It's a test pattern to check the color



It's a decompressed raw data.
In this test, can't found the color change. OK?



It's a relative intencity of sensor information with green light
source



It's a relative intencity of sensor information with red light source



It's a relative intencity of sensor information with blue light source

The relative intencity via light level and properties were measured
with the colored light source have no problems.
but, color changing is a problems (be occured by the color
separation process).

Color problems is appreared only in HIGHER light level.
To avoid this problems, Default light level of SD9 is so low.

I am falling in puzzle,
(1) to get the color, I overcome the noise problems
(2) to get the clean image, I overcome the color problems.

27-70mm/2.8, ISO100, 1/10S, F4.0
color temperature of source = 6800k

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

--
SD9 with 24-70DG2.8

--
http://www.pbase.com/dgross
ICQ UIN: 289647506

--
SD9 with 24-70DG2.8

 

[Ladehaug]

This is a captured image, P and AF mode with 24-70 F2.8 DG/DF
and staturation option +0.5, sharpen +0.5.

Kim,

Thanks alot for doing this test. This confirms what I've suspected for some time. The red hue is changing dramatically with intensity, going from almost a correct red in the darkest region towards orange and yellow in the other end. I can also see what's going on with the blue colors. The SD9 is not able to differ much in the intensity for the hues that are near pure blue. It appears that the blue color cannot be "blown out" to a lighter blue (or white) with increasing exposure (which it should do). This explains the unnatural deep blue sky and the blue blooming artifacts seen on some samples. The green, cyan and magenta is actually pretty good. The yellow is fairly good until it is abruptly being clipped, and this has been seen on many samples as well.

What does this mean?

1. Colors that have same hue in the real world, but with different intensity, will not appear with the same hue with the SD9.

This again will unfortunately lead to the following:

2. A large range of colors that are different in the real world will appear to have the same hue with the SD9.

Why is it unfortunate?

Making a general color profile that will correctly adjust all of the colors within a reasonable exposure range will not be possible to make, and the reasons for this is mentioned above. The best one can do is to make one or more basic hue adjustment profiles and adjust the colors as needed, depending on the exposure and the colors involved.



Monitor screenshot of the same testimage taken with my Canon S40. Program mode with average metering: F5.0, 0.4s, manually adjusted whitebalance (RAW). As one can see, the hues are much better preserved with intensity in this sample.



Just for the test, I also did a monitor screenshot of the Gretag Macbeth chart. Program mode with average metering: F5.0, 0.25s, manually adjusted whitebalance (RAW). Reading the hues, I find that the S40 does surprisingly well in this test. But then again, a color chart doesn't tell much about the relative color accuracy within detailed textures of an image. This is where the SD9 is clearly superiour to any other digital camera!

Geir R

 

[whereswade]

It is very difficult for any of us to comment in any way regarding the color of any image that is posted on the internet. If I look at this image on my monitor here at work, it will show the color cast that this monitor is set up for, if I look at it on my powerbook at home, It will present differently than if I look at it on my wife's HP. Saying "judge this color cast" unless every monitor is calibrated correctly, is like comparing the image that you see on a bar Television and your 15-inch kitchen TV. They are going to be different

There is, however, one way. If you know what the absolute values for proper rendition of skin tone or any color in the RGB gamut (or CMYK, for print purposes), then you can say that this image is tilting toward cyan or yellow or whathaveyou (which is a neat color).

That being said, the images themselves are nice. Very Krisp!