heya, if you have any show-off result images, i'd be interested..
also, if you need laplacian pyramid code, i think i have a working implementation of a local laplacian now (it's monochrome though). cheers, jo On Wed, Aug 10, 2016 at 10:50 PM, Wolfgang Mader <wolfgang_ma...@brain-frog.de> wrote: > I finally have something to show. > > You can grab the code under > https://github.com/wmader/darktable > the branch in called color2gray. > > The iop is designed to work with multiple operators. Right now, there are two > implemented. The first "lightness" simple keeps L and sets a = b = 0. > > The second one is called "apparent grayscale" and is more interesting. This > method uses a model for the Helmholtz-Kohlrausch (HK) Effect. This effect > describes the difference in the perceived brightness of an stimulus depending > on its color. The model is described in [1,2]. Basically, the chromaticity of > a color is measured as an angle with respect to a gray-point. This angle is > the input for the HK-Effect model. There is one free parameter, the ambient > luminance, which the paper says should be 20. As there is no sound reasoning > for this value, I implemented a slider for it. The effect, however, is minor. > > I used a test image from [2] to test my implementation against the result of > their published gimp-plugin. As they use d65 and we use d50, I patch relevant > functions it dt to use d65 for testing. Then, I get matching results. > > While I implemented the HK model, [2] takes this one step further by enhancing > the contrast for areas in the grayscale image where the respective contrast in > the color image is larger. I had not yet time to implement this, and for > family reasons, I will not have the time in the near future. Fortunately, > Maximilian, also on this list, approached me, as he is also interested in the > topic. He will take up the task of implementing the chromatic contrast, which > is based on Laplacian pyramids. > > So, check out the code, give it a try. If you have any questions, please let > me know, I try to answer in a timely manner. And as I already mentioned, this > is my first contribution. Therefore, if the code does not live up to your > expectations, let me know, and I try to fix it. > > Best, > Wolfgang > > P.s.: If anyone is interested in the test images, I am happy to send. I > assume, however, that this list does not accept attachments. > > [1] Simple Estimation Methods for the Helmholtz–Kohlrausch Effect, Y. > Nayatani, CIE Journal, 5:2, 1986 > > [2] Apparent Greyscale: A Simple and Fast Conversion to Perceptually Accurate > Images and Video, K. Smith, P. Landes, J. Thollot, K. Myszkowski, > EUROGRAPHICS, 27:2, 2008 > > > > On Friday, July 1, 2016 8:18:19 PM CEST Wolfgang Mader wrote: >> Hi, >> >> thanks for asking. I defensed my Ph.D. theses last week and overestimated >> the free time I would have while preparing the defense. Therefore, I will >> only find time for darktable starting with the coming week. However, I come >> back to the list once there are results such that I do not have make >> promises I can but live up to. >> >> Sorry, >> Wolfgang >> >> On June 30, 2016 3:49:19 PM GMT+02:00, Moritz Moeller > <virtualr...@gmail.com> wrote: >> >Hey Wolfang, >> > >> >On 09/05/16 11:46, Wolfgang Mader wrote: >> >> Thanks for the literature. I will have a look at is at soon as >> > >> >possible, and >> > >> >> will come back to you, once there is something do discuss. >> > >> >I'll have some B&W images to develop soon. Anything I could test >> >already? :) >> > >> >.mm >> > >> >___________________________________________________________________________ >> >darktable developer mailing list >> >to unsubscribe send a mail to >> >darktable-dev+unsubscr...@lists.darktable.org > ___________________________________________________________________________ darktable developer mailing list to unsubscribe send a mail to darktable-dev+unsubscr...@lists.darktable.org
