Jump to content

PTMFitter and color spaces


Biancoand

Recommended Posts

Hello everybody,

 

I was reading Malzbender-Gelb-Wolters 2001 article "polynomial texture maps", and at page 3 I found this phrase: " We prefer this simple but redundat representation over color spaces such as LUV and YCbCr due to the low cost of evalutation although we have implemented the method in these color space as well", this is also confirmed by Malzbender HP Labs Technical Report of 2001 that reports as possible format also PTM_FORMAT_LUM (color space CrYCb).

I didn't found the possibility to select this color space in the PTMFitter anyway, and also in the rtiviewer there is no function to load this kind of file.

 

This question arises since the PTM paper describes what they mean by LRGB but not how to compute the values for the R G and B components ( this  http://users.csc.calpoly.edu/~zwood/teaching/csc570/final06/jrickwal/ maybe descirbes the problem in a more articulated way ). Could it be possible to calculate a LRGB components using a different colorspace like LUV or YCbCr, extracting the luminance component and use the other two as cromacity? Another point: LRGB is known in astrophotograpy as: http://www.astro-imaging.com/Tutorial/LRGB.html, and is the combination of 4 different photos taken whit different filters. Maybe this could be the source of misinterpretation?

 

 

Link to comment
Share on other sites

  • 1 month later...

Hi,

As the chromaticity of a pixel is fairly constant under varying light source direction, the unscaled RGB values of the pixel can be computed by separating RGB components from the luminance component.

This question arises since the PTM paper describes what they mean by LRGB but not how to compute the values for the R G and B components

 

As you can read in Robust Luminance and Chromaticity for Matte Regression in Polynomial Texture Mapping or Specularity and Shadow Interpolation via Robust Polynomial Texture Maps, the chromaticity χ is defined as RGB colour ρ independent of intensity and is estimated as the median of inlier values, for k = 1..3 ( resp. R, G and B ) :

27yr6t5_th.png

with ω = 0 for outliers (specularities and shadows ) and where let each RGB image acquired is denoted ρi.

 

You can identify these outliers thanks to many ways which vary in complexity, some of which are described in the previous papers. A simple way consists in defining a binary weight matrix W where the entry is zero if the intensity of luminance is below a predefined threshold τ, in order to reduce the influence of shadows.

 

 Could it be possible to calculate a LRGB components using a different colorspace like LUV or YCbCr, extracting the luminance component and use the other two as chromacity?

I dont know if the others color spaces (e.g.YCbCr) are used to calculate the chromaticity ( i.e. an unscaled RGB image) in the different existing fitters. 

But I suppose the principe stays the same. Instead of computing the RGB image reconstructed as the product of the luminance L (approximated by HSH or PTM technique) and the chromaticity χ, you must adapt the reconstruction function to the corresponding model of color space, the paradigm has changed. 

 

The principal difficulty is to prevent the confusion between luma, luminance, chominance, chromaticity...

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...