[texhax] Request for advice on figures for on-line and printable book.

[Thomas Vaughan]

On Wed, Mar 9, 2011 at 07:33, Andy Farnell <padawan12 at obiwannabe.co.uk> wrote:
>
> Time spent devising a great system will save
> hundreds of hours once you start writing and
> typesetting.

I agree. That's what I'm trying to start on now.

> So you want LaTeX for static text but will need
> to feed data (images) into that system and take
> output (text formatting) too, maybe beginning with
> a customisable front end like Lyx or a set of emacs
> scripts will be useful to make a "central command"
> system for other scripts that will run.

Well, I used to be an emacs guy, but now I'm a vim guy. :^)

My plan is to write straight LaTeX text files for the most part.

For the figures, however, it's starting to look as though I'll need to
write a tool that reads in figure-description text files written
according to a language that I design and spits out both
   (a) a text file with instructions that sketch can process for the
static figure and
   (b) a text file with javascript code that calls WebGL to implement
the dynamic figure.

In this model, clicking on a link in the PDF file opens a browser
window that points to the WebGL application for the dynamic figure. So
one could view the PDF file in xpdf, evince, or acroread and click to
bring up any desired dynamic figure, or one could print the PDF to
paper.

The idea of autogenerating javascript seems a bit daunting right now,
but I want to learn javascript and WebGL anyway.

If all of this ends up looking too hard, I'll probably just punt and
use canned animation for each dynamic figure via the asymptote
package.

> Your choice of dynamic content is indeed vast.
> The tool choice will depend on whether your focus
> is mainly 2D or 3D.

I think that most of my figures will be 2D, but my system definitely
needs to be able to handle 3D. That's why I'm thinking that I should
not rely on gnuplot, whose 3D support is weak in its lack of
perspective.

> I would be tempted to look to
> the future and aim for HTML5 with vector graphics
> and the ability to run the formula/simulations locally
> on browser.

I agree. That's what I'd like to do.

> OTOH, systems like GNU Octave, Processing and
> WebGL will give you nicer 3D.

I'm thinking that WebGL goes along with the whole HTML5 thing.

I've used Octave some. I'm not sure how it fits in here.

> If you can get something looking great in any
> 3D modeller you can almost certainly export and
> render using a dedicated renderer like POVRAY
> and then obtain high quality images in postscript.
> But those won't be precisely the same images running
> on the real-time browser version. Some compromise
> between high quality and real-time may be needed.

I don't mind if the dynamic figures look somewhat different from the
printed ones.

However, I want the source to be the same for both. And I'd prefer to
write the source in a text file rather than to build a model via a
GUI.

-- 
Thomas E. Vaughan