Is it possible to expect pst-solides3d (in the future) to generate gradient mesh?

I have been using pst-solides3d to generate surfaces and then importing them to Adobe Illustrator to turn them into a gradient mesh. The procedure can be time consuming if I want to be very precise, and even there are only 4 vertices, I would have to manually adjusting their positions and all 8 handles, and then pick up the color from the original surface, quite laborious. See my post here

My question is: Is it possible for pst-solides3d to implement an automation of the above procedure?

This is not really an answer to your question, so you may want to wait for other answers which your questions about pst-solides3d.

This answer is to build up the knowledge base and to address your last sentence in the linked question https://graphicdesign.stackexchange.com/questions/42022/transforming-3rd-party-discrete-gradient-mesh-to-ais-smooth-gradient-mesh, namely

Alternatively, if there is a better way to do this, I would also like to know it. Thank you.

From what I understand, you want

• vector graphics

This specific problem can be solved by means of pgfplots. I have used a Möbius strip since I have neither your data files nor the parametrization of your surface:

1. bilinear patches. First-order resolution of the patch boundary, bilinear color interpolation: \documentclass{standalone}

\usepackage{pgfplots}

\pgfplotsset{compat=1.9}

\usepgfplotslibrary{patchplots}

\begin{document}

\begin{tikzpicture}
\begin{axis}
surf,
samples=25, samples y=25,
variable=r,   domain=-1:1,
variable y=a, domain y=0:2*pi,
patch type=bilinear,
]
({cos(a) * (1+r/2 * cos(a/2))},
{sin(a) * (1+r/2 * cos(a/2))},
{r/2 * sin(a/2)});
\end{axis}
\end{tikzpicture}
\end{document}
1. bicubic patches. Third order resolution of the patch boundary (and less patches), bilinear color gradient: \documentclass{standalone}

\usepackage{pgfplots}

\pgfplotsset{compat=1.9}

\usepgfplotslibrary{patchplots}

\begin{document}

\begin{tikzpicture}
\begin{axis}
surf,
samples=10, samples y=10,
variable=r,   domain=-1:1,
variable y=a, domain y=0:2*pi,
patch type sampling,
patch type=bicubic,
]
({cos(a) * (1+r/2 * cos(a/2))},
{sin(a) * (1+r/2 * cos(a/2))},
{r/2 * sin(a/2)});
\end{axis}
\end{tikzpicture}
\end{document}

Both examples need the patchplots library, and both have a parameterized surface depending on r and a. The parameterization is from wikipedia. Note that trig format plots=rad switches the default config from degrees to radians.

Bicubic patches are quite difficult to generate: they have 16 points for each patch. The key patch type sampling simplifies this considerably -- if the input data is sampled from a function anyway. If you have table input, you will have to provide the patches on your own.

The colors are taken from a colormap: the smallest z coordinate is mapped to the first color of the colormap, the largest z coordinate is mapped to the last color of the colormap. Everything in-between is mapped linearly into the colormap. Further reading: keys point meta (choose which scalar value to use) and colormap (which colormap), also related are "Surface plots with explicit color".

Related question: Creating Bezier surfaces using procedural graphics

• Thanks for your comprehensive answer. Im wondering if I could use a different colormap (other than the height function)? – Troy Woo Nov 16 '14 at 14:00
• Sure. In this case, the simplest case would be some math expression using point meta=<expression>. The actual colors can be configured using colormap (details in the manual pgfplots.sourceforge.net) – Christian Feuersänger Nov 16 '14 at 14:05
• Thank you. I have also another question, is it possible to remove the internal lines and only leave the boundary line of the surface? I fear not, since it involves Boolean operations. – Troy Woo Nov 16 '14 at 14:09
• This is impossible by means of pgfplots. You may consider using mesh/interior colormap which allows to choose a different colormap for the "other side" of the patches – Christian Feuersänger Nov 16 '14 at 15:36