Re: Offscreen plotting in OpenGL

[Ben Abbott]

On Aug 12, 2011, at 11:58 AM, Jordi Gutiérrez Hermoso wrote:

> 2011/8/11 John W. Eaton <address@hidden>:
>> On 11-Aug-2011, Jordi Gutiérrez Hermoso wrote:
>> 
>> | So there's one outstanding wishlist item for making the OpenGL
>> | backend default for 3.6, and that's offscreen plotting.
>> |
>> | I asked around, and apparently it's now easier to do this than it
>> | was 3 years ago, and framebuffer objects are the way to go. It
>> | doesn't sound terribly complicated from my initial perusal of the
>> | subject. Should we "officially" (whatever that means) add this to
>> | the 3.6 release goals?
>> 
>> It's definitely needed.
>> 
>> Does this have anything to do with printing? As I recall, printing
>> also requires a display? Does that make printing somehow tied to
>> screen (or framebuffer) resolution?
> 
> AIUI, no, it doesn't require a display (I don't know if it requires a
> running X server anyways), and whatever it is that works for gl2ps can
> also take a framebuffer object. What's a high-level explanation of
> what gl2ps does? It's been a while since I've touched the OpenGL code.
> 
> As an aside, I showed a Windows user the fltk interface yesterday in
> IRC for Tatsuro's 3.4.2 binary, and he said he was "very impressed".
> This, with the profiler, and an "experimental" GUI (say, disabled by
> default, so as to not have a buggy GUI soil our reputation), could
> make some very interesting user-visible improvements for 3.6.
> 
> - Jordi G. H.

From the gl2ps manual (http://www.geuz.org/gl2ps/gl2ps.pdf)

GL2PS works by capturing the contents of the OpenGL feedback buffer. As
such, all the OpenGL operations applied in the pipeline after the creation of
the feedback buffer will be ignored or have to be duplicated by GL2PS (e.g.
font/image rendering, polygon oset or line stippling|see sections 2.2, 2.3, 2.5
and 2.6).

Are framebuffers and feedback buffers the same?

Ben