CinePaint in London, After GLLUG

After CinePaint presentation at GLLUG, what’s next?

By Robin Rowe

LONDON, UK ( 2012/12/10 – Gabrielle Pantera and I spoke at GLLUG this week and heard interesting feedback from the audience on the direction CinePaint may take. This was at the Mozilla Firefox space in London.

There were two presentations before us at GLLUG. Luke Leighton spoke about the Rhombus-Tech Project and the significant challenges getting Android SoC devices in compliance with the GPL. Every Android device is customized and Chinese chip makers generally consider themselves beyond the reach of the GPL. Leighton did report limited success, that one device had been open sourced with dramatic results in increasing its popularity. Hopefully, that precedent will encourage other chip makers to open source.

Andrew Lack spoke at GLLUG about his TinyBasic project for Raspberry Pi. Lack described the many improvements he’d made to an abandoned BASIC interpreter he’d revived and extended to support the Palo Alto BASIC dialect. While I was talking with Leighton before the talks, he pointed out that the graphics subsystem of the Raspberry Pi is not open source and because the GPU boots that device, it makes that entire system more closed than most people realize. Open hardware isn’t easy.

In discussing CinePaint at GLLUG, a challenge I noticed is that suggested improvements from potential users are often in the high-end, that it’s hard to keep the focus on moving CinePaint down market, to make it a general purpose system suitable for the mass market.

Some desired features that are on my short-list for implementation:

  1. GUI Tray. The tray will be a filmstrip-like window that lets you see frame sequences in order as though printed on motion picture film.
  2. Motion graphics. Simple motion graphics across multiple frames.
  3. Storyboarding. Option to turn the tray vertical and place annotations to the right of each frame.
  4. Stop-motion animation.
  5. Rapid design prototyping. Quickly create beer can and cereal box designs with wrapped textures. OBJ file support.

Not new features, but needed to make CinePaint easier to support:

  1. New build system based on make. Autotools and CMake are just too exotic to expect anyone but an expert to support, a barrier to new programmers joining CinePaint.
  2. Mac and Windows support. The build system has been the big obstacle to efficiently releasing CinePaint cross-platform.
  3. FLTK. That CinePaint uses GTK has always been a source of bugs. This C API isn’t very forgiving and is very hard to trace in the debugger. Safer C++ APIs such as Qt and FLTK are the most portable alternatives. I like FLTK because it’s much smaller. Significant work has already been done to bring CinePaint over to FLTK, but there’s more to do.

So, when will we see these wonderful features in CinePaint? The plan is a January release.

  1. FYI, copy of my response to the commentary at below…

    You’ve got your own agenda, so I won’t try to change your opinion of me. However, there are two things you may want to understand about CinePaint.

    Programmers do rush to join the CinePaint project. They quickly get frustrated that there’s nobody within the project to make timely changes to the build system, mainly because I’m not an autotools or CMake expert. Open source contributors expect the build system to just work and I think are justified in being turned off that CinePaint hasn’t been good at that. I’ve asked that no programmers join the project until I have a good solution to our build problem.

    Before this codebase was enhanced by Sony Pictures and renamed CinePaint, the original developers were paid by the film industry to write code to enhance a well-known open source project. The sponsors provided funding of limited duration to create code to be maintained by the open source community. This was a Google Summer of Code type of project before Google existed and on an impressive scale.

    Regarding your remarks that the original developers called their code a dead end, nobody anticipated that GIMP developers who’d been paid to enhance their software in 1999 would say in 2000 they didn’t like their own work and fork it, seemingly for the purpose of evangelizing against it. Ancient history. Perhaps time to drop as an argument?

  2. Pingback: Exotic CMake in the alien CG land | Getting Things Done

Leave a Reply