I am still working on selecting what patches are ready for inclusion in a pull request, and which still need to much work. Unfortunately there turned out to be still some problems in fairy-move-model that surfaced when I refactored Elven Chess to use it, and debugging those took more time than I had hoped.
One of the remaing issues in your patches is that you included fr-rhino2 and fr-cobra in the cazaux/timurid series, and that I think these pieces (although improvements of what we had) are still not up to standard. How about the following Rhino:
Is this better? I now extended tube.c with the possibility to distort the rings you define in more ways than squeezing them to ellipses, and tried to use that for creating a more plausible cross section for the snout. What I do is modulate the radius of the ring as we go around it from phi = 0 to 360 degrees by multiplying it with (1 + a_n*cos(n*phi)) for n = 1 - 5, with a_n some constants specified per ring that by default are 0. In lowest order the n=1 factor does the same thing as displacing, and n=2 the same as the elliptical distortion, which we already had. (For large values of a_n there is a difference, though: a_1 gives you a 'banana distortion', and a_2 an 'hour glass'.) For n = 3 and 4 you get triangular and square distortions, which I used here to flatten the top of the snout (a_4 = -0.15) and broaden the lower jaw (a_3 = +0.05).
I am still working on selecting what patches are ready for inclusion in a pull request, and which still need to much work. Unfortunately there turned out to be still some problems in fairy-move-model that surfaced when I refactored Elven Chess to use it, and debugging those took more time than I had hoped.
One of the remaing issues in your patches is that you included fr-rhino2 and fr-cobra in the cazaux/timurid series, and that I think these pieces (although improvements of what we had) are still not up to standard. How about the following Rhino:
Is this better? I now extended tube.c with the possibility to distort the rings you define in more ways than squeezing them to ellipses, and tried to use that for creating a more plausible cross section for the snout. What I do is modulate the radius of the ring as we go around it from phi = 0 to 360 degrees by multiplying it with (1 + a_n*cos(n*phi)) for n = 1 - 5, with a_n some constants specified per ring that by default are 0. In lowest order the n=1 factor does the same thing as displacing, and n=2 the same as the elliptical distortion, which we already had. (For large values of a_n there is a difference, though: a_1 gives you a 'banana distortion', and a_2 an 'hour glass'.) For n = 3 and 4 you get triangular and square distortions, which I used here to flatten the top of the snout (a_4 = -0.15) and broaden the lower jaw (a_3 = +0.05).