computer-generated images for instructions

[rwb27]

It would be ideal if the instructions didn't require photos to be taken every time the parts change, and if we could have diagrams rather than photos (less cluttered). This could be done by rendering pictures from the STL files. It would mean the source gets much smaller (no need to store photos) and the documentation is easier to maintain. It would also mean the pictures in the instructions always correspond to the version of the microscope you have in front of you.

I started a very basic attempt at this in the illustrations folder in 3f1a841 but I think OpenSCAD is probably the wrong tool for that job. Much better would be a Python library that can render STL -> images, together with overlaid arrows/text. Blending together parts and cutting them away would also be nice, that might require some manipulation (which could be done in OpenSCAD if needed).

[rwb27]

I played about a bit with rendering the microscope - obviously (?) compiling it is slow, but rendering from an STL is pretty fast. However, if I try to (e.g.) cut the STL file to reveal the actuator, it fails. Probably what I need is a Python library that will do some basic solid geometry for me...

[rwb27]

I've just played with POV-Ray and stl2pov from stltools. The two povray files (one of which was generated by stl2pov) render the microscope body. It's fast, and I suspect it would be possible to have a bunch of POV-Ray scripts that can then render the images needed for the instructions. There's some description of how this works in the context of a POV-Ray based slicer.

Longer-term it might be nice to try embedding thingiview or a script using its underlying libraries to allow exploded-view renders live in the instructions.

[rwb27]

For now I think what needs to happen is:

• Write a script to convert all the STL files into povray files. This might be made simpler by re-rendering some of the STLs as "in place" rather than "print ready" (e.g. the illumination prints lying down, but can also be rendered in its correct orientation).
• Write POV-Ray layouts for each image required in the assembly process. Bonus points for using includes/preprocessing appropriately to minimise the effort per image.
• Write a script to render all the layout files and create the images
  I am not currently working on any of these things - if you would like to, please shout!

[rwb27]

By the way, it should be relatively painless to add arrows, etc. to the output:
http://www.f-lohmueller.de/pov_tut/a_geo/a_geo20e.htm

[rwb27]

The repository is moving to GitLab, and this issue has been migrated. I'll close all the issues here in due course, but am leaving notices on all the currently-open ones. If you head over to the other repository, this issue will be updated there.