Color profiles and Celestia...

Discussion forum for Celestia developers; topics may only be started by members of the developers group, but anyone can post replies.
Topic author
ElChristou
Developer
Posts: 3776
Joined: 04.02.2005
With us: 19 years 9 months

Color profiles and Celestia...

Post #1by ElChristou » 01.09.2006, 18:53

Guys, while testing a new dds converter from Dirkpitt on osX, we found strong evidence that color matching has never been taken in account in the galaxy of files available to use with Celestia...

What can we do to create a norm to be sure that at last official maps (but also whatever external files) use the same color profile? (to be sure we all see the same thing)

Could Celestia integrate a color correction depending of a specified profile?
Image

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 3 months
Location: NY, USA

Post #2by selden » 01.09.2006, 18:59

ElChristou,

Exactly what do you mean by "color matching"?

Many DSC Nebula images originate from scientific images which are designed to present exactly the levels corresponding to the amount of radiation received in particular bandwidths. Changing their colors would damage this correspondance.

That said, I'm sure that most such Addons have been created without worrying about maintaining their scientific color accuracy.
Selden

Topic author
ElChristou
Developer
Posts: 3776
Joined: 04.02.2005
With us: 19 years 9 months

Post #3by ElChristou » 01.09.2006, 19:13

selden wrote:ElChristou,

Exactly what do you mean by "color matching"?

Many DSC Nebula images originate from scientific images which are designed to present exactly the levels corresponding to the amount of radiation received in particular bandwidths. Changing their colors would damage this correspondance.

That said, I'm sure that most such Addons have been created without worrying about maintaining their scientific color accuracy.


Imaging a user is using a specific color profile for his screen but also for his graphic app, he will edit a file with this specific color setting which one can be complitely different for another user using another color profile...

In the edition industry there is norms to be sure the color of a printed document match the color of the digital document or vice versa. Those norms use some specific color profiles, so my question is if we could implement such control within Celestia.
Image

Avatar
selden
Developer
Posts: 10192
Joined: 04.09.2002
With us: 22 years 3 months
Location: NY, USA

Post #4by selden » 01.09.2006, 20:02

I think my initial reaction is a partial "No," for the reason stated above: the input colors, the colors in Celestia's surface texture image files, are not supposed to be artistically balanced. They're supposed to be accurate representations of scientific data.

("Supposed to" but probably aren't by the time they make it into most Addons, especially if they use jpeg or compressed dds :( )

For reflective objects, could one encode a color profile which says "Only do those color transforms required to represent lighting from sources that are different colors"?

For light emitting objects, could one encode a color profile which says "Only do those color transforms required to represent loss of intensity due to distance or obscuring materials"?

To put it another way, I am assuming a "scientific" color balance is maintained and used for all of the images provided for use with Celestia. I.e. the original images acquired from professional astronomical image repositories must be assumed to have an appropriate color balance. (Although, of course, most astronomical sites do not document the image color transforms used for the images on their oureach sites. You have to get the original FITS image files from their scientific database servers, which usually do include documentation about the profile of the frequency bandwidths represented in the image.)

To me, this means that the color balancing problem is an output issue, specific to each system, and not necessarily something that Celestia should be expected to handle.

Each individual display and printer requires its own color balance configuration to ensure that it accurately presents the input texture map image colors to the viewer.

However, this does require everyone's screens and printers to be calibrated with appropriate output color profiles, so that everyone sees the same representation of the colors that were provided. This suggests to me that a procedure or software package for calibrating screens and printers to create those output profiles would be very helpful.
Selden

Topic author
ElChristou
Developer
Posts: 3776
Joined: 04.02.2005
With us: 19 years 9 months

Post #5by ElChristou » 01.09.2006, 22:29

selden wrote:...To put it another way, I am assuming a "scientific" color balance is maintained and used for all of the images provided for use with Celestia. I.e. the original images acquired from professional astronomical image repositories must be assumed to have an appropriate color balance. (Although, of course, most astronomical sites do not document the image color transforms used for the images on their oureach sites. You have to get the original FITS image files from their scientific database servers, which usually do include documentation about the profile of the frequency bandwidths represented in the image.)

Yep, I don't know much about the original data, but I suppose the color profile do exit in a form or other. Now the point is to keep this right profile all along the edition. Just take Mars as example, each map has a different tone and there is no way to even know if this tone is the one the author was looking for...

selden wrote:...To me, this means that the color balancing problem is an output issue, specific to each system, and not necessarily something that Celestia should be expected to handle.


Right, but if celestia could "correct" the result within it's windows, no more problem IF the maps have a specific color profile as norm...
The question is, is this possible?
Image

Avatar
dirkpitt
Developer
Posts: 674
Joined: 24.10.2004
With us: 20 years 1 month

Post #6by dirkpitt » 03.09.2006, 03:14

Sorry for the late input. Just to clarify what ElChristou is saying,

Certain images that have been edited by consumer graphics programs (e.g., Photoshop) can end up containing embedded ICC color profiles.

Profiles cannot be embedded in DDS, or unprocessed data files straight from space probes. But color profiles can be embedded in JPG and PNG files and not taking profiles into account can result in inaccurate color rendering.

When JPG and PNG images are rendered in Celestia without taking into account the embedded profile, they may appear to have vastly incorrect colors depending on the type of monitor that's being used.

Case in point: The default setting for Photoshop is to embed the sRGB profile in saved images. This is perfectly fine as long as the monitor used by the creator/person editing the image is in fact sRGB. However, for example this can be a bad idea for the vast majority of Mac users:

sRGB gamma = 2.2 (dark)
Typical Mac gamma < 2.0 (bright)

If the monitor used to edit the image was sRGB, then if the same image is viewed on a Mac without color correction the image will be too bright. This is because the sRGB user has a darker monitor, so he/she made the image brighter to compensate.

The person who edited the image can however embed an sRGB profile. When this image + sRGB profile is viewed on a Mac, in a program that supports embedded profiles, the bright colors will be automatically darkened by the sRGB profile so that the resulting colors match what the sRGB person sees.

Celestia does not currently take into account ICC profiles. However, it wouldn't be too hard to add a platform-specific layer to color-correct JPG and PNG textures. There is one side-effect to consider, and that is textures with incorrect embedded profiles. Mac users for example may unintentionally create textures with embedded sRGB profiles - incorrect, but could very well happen because of Photoshop's default settings. Some false bug reports may result.

Topic author
ElChristou
Developer
Posts: 3776
Joined: 04.02.2005
With us: 19 years 9 months

Post #7by ElChristou » 04.09.2006, 11:58

dirkpitt wrote:...There is one side-effect to consider, and that is textures with incorrect embedded profiles.

It's why I suppose a "prefered" profile should be define...
sRGB is very common those days, and it's not only used on osX; it is becoming a norm in the global med quality graphic industry (games, screen default setting etc...)

dirkpitt wrote:...Mac users for example may unintentionally create textures with embedded sRGB profiles - incorrect, but could very well happen because of Photoshop's default settings. Some false bug reports may result.
Image

Avatar
dirkpitt
Developer
Posts: 674
Joined: 24.10.2004
With us: 20 years 1 month

Post #8by dirkpitt » 04.09.2006, 12:50

No need to choose one single profile if the viewer application (e.g., Photoshop) supports embedded color profiles.

But sRGB is apparently a good-enough cross-platform profile if one were to choose one, and I think even Mac users can just work in another color space like say, Adobe RGB and _remember_ to convert to an sRGB-profiled+embedded image for final distribution. The problem comes when no regard is made as to whether sRGB matches the actual monitor - this would be a user who say, sets the working space to sRGB on a Mac.


Return to “Ideas & News”