The new emerging NLE for GNU/Linux
Caution incomplete, more content to come…

In this section of the Lumiera website we’ll collect some documents and background informations about the history of the Lumiera project.

Roots of the Project

As with many open source projects, it is difficult to determine the absolute beginnings precisely. There was no “kick off” meeting with some people to decide “now we’ll start writing a video editing software”. Rather, several of us where using Cinelerra for editing tasks with rather professional quality level. It turned out that indeed Cinelerra is capable of delivering professional quality, but users of Cinelerra are constantly facing some more or less serious problems.

While not being part of the long-standing “Cinelerra-CV” community right from start, since 2005 we tried to remedy this situation by contributing improvements, especially addressing things perceived as major impediment in our everyday editing work. But, overall, the net result of these efforts was rather sobering

  • first of all, the basic constellation within the Cinelerra-CV project isn’t encouraging. Judging from past discussions, several forums and mailing list archives, the original upstream author(s) — generally only known under the pseudonym “Heroine Warrior Inc.” — seem(s) rather not to be interested in a more collaborative working style. Just, from time to time, a new upstream version will be “thrown over the brick-wall”, featuring some new and nifty additions, some bugfixes plus some more or less serious new quality problems.

    Fortunately, there was the Cinelerra-CV crew, which used to merge these new upstream versions into the community’s source control system and then, over the course of the next months, did a great job in getting the new version into a roughly usable shape for a larger variety of distributions and configurations. Some of these improvements where picked up by Heroine Warrior, some weren’t, without any real feedback.

    But given this overall situation, it is quite understandable that this long-term Cinelerra-CV crew was rather reluctant towards anything beyond isolated fixes and small improvements. Because any change, irrespective of the result, would just add to the increasing amount of differences to maintain and merge with each new upstream release.

  • even ignoring this problem of “mergeability”, Ichthyo’s attempts to rectify some of the most time-consuming oddities in the behaviour of frequently needed features (transitions, automation, effects, interlacing, proxy editing) quickly evolved into disproportionally expensive refactoring and reworking projects. At the same time, Cehteh concentrated on the frequent deadlocks and crashes. Being an experienced system programmer, he quickly spotted the reason — yet any attempt of fixing the problems would lead to “chainsaw surgery” all over the place. Seemingly, Cinelerra’s code base was just resilient towards change and evolution.

The strange year 2007

I recall 2007 as one of these pivotal years, which — while featuring rather few “news events” — still endow some kind of atmospheric change in several areas of life, both in small and on the large. For the Cinelerra-CV community, in spring 2007 it became evident that, from the long-standing core contributors, all but one had left, for one reason or the other (not related to Cinelerra). The development efforts for the CV-Version basically stalled. Several discussions on IRC gave the impression that the project as a whole might take another turn.

In May, Cehteh and Ichthyo for the first time had some longer conversations on IRC. Cehteh showed a “Cinelerra Guerilla Git repository” with the official CV-Version and some of his disruptive changes to rework the asset handling, memory management and thread locking. He wrote a Braindump page on his personal Wiki, proposing

  • to replace Cinelerra’s excessive use of threading by a thread pool with precisely scheduled, side-effect free frame calculation jobs, optimally using the I/O bandwidth

  • to introduce a more stringent module- and plugin system, and then to modularise the codebase step by step

  • to replace Cinelerra’s “background rendering” by a frame index and frame cache

  • to handle the render pipeline more flexibly, so that every component (Camera, Mask, Projector, Effects) can be inserted everywhere

  • to switch to a really open distributed project structure, allowing everyone to contribute freely. Let just the best merged version win!

Some days later, ill tempered and in a bad mood after yet another frustrating editing session with Cinelerra, spending most of the time just with re-adjusting already finished sound work messed up by side effects of further editing, Ichthyo wrote a rant, peppered with random insults and curses. Herman Robak encouraged him to rewrite that bluster into a polemic text which can be taken seriously. The resulting »Cinelerra Woes« page caused a surprisingly strong resonance among the Cinelerra-CV community and quickly led to the proposal for a gradual reworking of the existing code base to rectify and cure the problems pinpointed here. Ichthyo condensed that into a project plan with seven milestones, first to define a solid new tooling and project framework and build some trusted islands of new code. Starting from there, gradually then the whole source base could be reworked. On another page, he defined some crucial »best practices« to complement and support the desired open distributed nature of the new project.

So, suddenly we found ourselves right in the middle of an exciting new movement with much momentum, long debates and lots of new ideas — and, most exciting, it seemed doable right now and here. Yet we should note that we didn’t want to start a new video editor project — we didn´t want something different, we wanted a better Cinelerra. Throughout these discussions, it became custom to label it the-next-Cinelerra, or — since the current Version was 2.1 — just call it Cinelerra-3. After some weeks, Cehteh wrote a Project Manifest and sent a Project Announcement to the contributors, the known Cinelerra-CV developers and Cinelerra’s original Author (“Heroine Warrior” alias “Jack Crossfire”).

more to be written as time permits…