What software is best for creating 3D graphics in scientific papers?

[ilvreth]

Hi to all.

I want your opinion about which software is the most suitable for rendering 3D graphics in order to include these as figures in manuscripts, papers etc.

I have seen many papers such as in Nature Physics, Nature Photonics etc that some 3D figures are really awesome. Especially, i have seen 3D vectors and trajectories with axes shadows projections and awsome lighting, 3D graphics of microcavities with multiple layers, 3D representations of optical pulses, 3D models of the experimental setup etc.

Could you tell me which programs are used in papers for 3D graphics?

I have used extensively the 3D graphics rendered by mathemata but they seem not suitable for papers or theses.

I have seen another one solution which is called PovRay. Is it used for papers?
It seems very difficult but if this is my last chance, then i will learn it from scratch despite of the difficulty.

 

[gsal]

Have you thought of contacting the authors of the papers whose 3D graphics you like?

 

[Dr Transport]

Matlab has the ability to export out figures into a large number of formats for inclusion into documents, I uawe .wmf all the time for powerpoint and .eps for LaTeX documents.

 

[f95toli]

Note that much of the graphics you see in articles in Nature (and Science) were NOT generated/drawn by the authors.; they have professionall illustrators who create the graphics and I suspect they use professional (=expensive) software (3D Studio, Maya etc).

This is one reason why you rarely see graphics this nice in other journals where the authors themselves have to create the graphics.

 

[Solkar]

I have used extensively the 3D graphics rendered by mathemata but they seem not suitable for papers or theses. [...]
I have seen another one solution which is called PovRay. Is it used for papers?

3D Plotting with Mathematica, Maple, Matlab, Maxima & gnuplot et al. simply has a different focus than rendering with POV-Ray, RenderMan or plain OpenGL.

The core difference is the quality needed for lighting and shading; an artist modelling Botticelli's Venus for 3D would have different display requirements than a mathematician plotting e^(1/z).

Of course, what a full-scale renderer can do with a given mesh is superset of what any plotting interfaces could do, but using the POV-Ray scene description language needs training and practice; let alone using RSL (Renderman Shading Language) or GLSL efficiently.

 

[ilvreth]

Thanks for the reply! I ended up doing all my 3D graphics with PovRay. Is pretty awesome.

 

[Solkar]

For converting scientific data to pov:

Likely the most straight-forward (numerical) geometry asset format is Wavefront obj;
it's quite easy to generate such files from whatever language¹ or CAS, and easy to parse.

For quickly generating pov "mesh2" objects from that, "osgconv", part of the OpenSceneGraph tookit, is well-suited; it's as easy as

osgconv /tmp/foo.obj /tmp/foo.pov

---

In addition - there is an awesome reference for POV-Ray textures
http://texlib.povray.org/
you may find useful.

Happy hacking!

Solkar¹C/C++ users should keep in mind that vertex numbering in .obj starts at 1, not at 0.