“Sleep Tight” Notic Nastic Live Clip (1 min. 14 sec.)
(Watch the Flash Video above or download (“Save Link As…”) the .mov.)
1: Selecting “the good stuff”

Click the image to see a larger version.
I load my whole footage into the Blender Video Sequence Editor (VSE) and start selecting “the good stuff”: with the mouse over the Image Preview window (see screenshot 3: Selecting “the really good stuff”) I use Space for start/stop and Right/Left Arrow for going back and forth. In the timeline (= Sequence): K for Cut and Shift S for snap editing (shortens/extends clip to the playhead when start/end is selected). The “good stuff” gets moved upwards a row (= Channel) with G for Grab and Y for Y-Axis. I keep the uncut original inside a (muted) Metastrip should I need it later since I am working without timecode. (You can use the Stamp render option for overlaying time information.) Then I make a back-up of the .blend.
2: Selected vs. original footage

Click the image to see a larger version.
I move the selected clips together using Ctrl for snapping clip to clip. The white numbers on the right show the last frame of “the good stuff” (top), the music track (middle) and the not used clips etc. that are now also inside the muted Metastrip (bottom).
3: Selecting “the really good stuff”

Click the image to see a larger version.
Same procedure as above (1: Selecting “the good stuff”) with the mouse over the Image Preview window (top right) and Space for start/stop, Right/Left Arrow for going back and forth… Since I easily get realtime playback using DV footage (no effects applied) with Ubuntu 8.04 and Blender 2.48a on my PC (current Core 2 Duo processor, 2 GB of RAM and a reasonably fast 250 GB harddrive) this workflow allows me to find my in and out points while looking at moving images the way the audience will see it… Frame 6 067 is now the last one after “the really good stuff” is moved together (that’s down from 20 520 from the previous step). I use the Blender Text Editor for taking notes. Again I make a back-up of the .blend.
4: Sorting the clips

Click the image to see a larger version.
I add a Text Editor window left of my Sequence and name/tag my tracks. I adjust the Sequence window with (Number Pad) Home and then use Middle Mouse Button and Ctrl to fit the Channels to my tagged tracks. Then I move “the really good stuff” clips in the appropriate track (for moving clips up or down without moving them in time I once again use G and Y, for selecting multiple clips I use B for Border Select…).
5: Editing the video

Click the image to see a larger version.
I use Markers for the points where the singing starts. Then I move the best of the previously sorted clips to roughly where I need them and work on the details again using Shift S for snap editing and also G for Grab to shorten/extend a clip’s start/end… The basic structure for the edited video: details/close-ups at the beginning, the artists performing in the middle and a longer zoom out close to the end. I move the unused clips into Metastrips on the left.
As before I use (Number Pad) Home for seeing everything that’s in the timeline. But since there are the unused clips on the left now I then select the music track and (Number Pad) Dot/Del (normally used for “zooming in” on a single clip) for focusing the Sequence window to the area that I actually want to see when editing…
I export the edited clip as a PNG Sequence (= series of .png images). (F10 (pressed multiple times) for switching between the Sequencer buttons when editing and the Render buttons when exporting.)
The annotations in the screenshot above show what’s important for exporting/rendering.
6: Through the Compositing Nodes

Click the image to see a larger version.
I load the previously exported PNG Sequence into the Compositing Nodes and let it run through a series of filters: DV video artefacts get smoothed out and my own graininess/structure and artefacts get added. The particular filters used here may be experimental and only make sense for this clip, but a couple of combinations can be useful for other projects (values would need to be adjusted): Gauss-Darken for a toon look, Screen-Overlay-Mix for optimising an image, Sharpen-Soften for focus related compositing tasks (high values for Sharpen may introduce artefacts).
The yellow and blue annotations in the screenshot above show what’s important for processing a series of PNGs with the Compositing Nodes: in this set-up navigation is possible by moving the Sequence playhead (the green line at frame 90) for previewing different parts (individual frames) of the Nodes processed video. I export the Nodes filtered PNGs once more as a Sequence of PNGs.
7: The final look

Click the image to see a larger version.
I load the Nodes processed PNGs into my VSE Sequence (Space to Add Image Sequence), add and finetune a Glow (with clip selected Space to Add Glow), make a Metastrip of both (with clips selected and M) and use Color Balance (with clip (Metastrip) selected: Filter tab > Use Color Balance) for creating the basis of the yellowish (but at this point rather dark) look. I then add a one step Strobe (Filter tab > Strobe: 2.00). With Shift D I duplicate the Metastrip, move it up one Channel and also move it one frame out of sync (see screenshot). The Glow for this FX 1 track gets removed (Tab to open/close a Metastrip) and in the Edit tab I select Blend Mode Add and Blend: 33.00. (31.01.09: And Use Color Balance – as can be seen in the screeshot above – gets also unselected for this FX 1 track.) I also duplicate the FX 1 track, move it up one Channel and also move this FX 2 track out of sync for yet another frame. There is one last step not shown in the screenshot: a final, subtle Glow is applied to all strips.
I once more export the clips as a Sequence of PNGs, import them one last time and change the export settings in the Render Buttons > Format tab to FFMpeg, make my selections for the video and audio format/codec and make sure that Multiplex audio is selected in the Audio tab before exporting the finished video clip with sound.
8: Credits

Click the image to see a larger version.
This is actually done somewhere before finishing the editing (5: Editing the video) and using my own 2D Titles Preset .blend/tutorial. I adjust the text to the right size first. In order to have the Center where I need it for my title animation: Object > Convert Object Type… > Mesh. Then: Object > Transform > Center New.
Tutorials etc.
Download my Video Editing Preset .blend (Quick Start tutorial inside) for all the basics you need to get started with the Blender Video Sequence Editor.
Download my 2D Title Presets .blend (tutorial inside) for making “crystal clear” titles with Blender.
Download Blender for all major platforms.
Thanks to Paul for hosting these .blends on the Final BUG server!
See Using VSE (and Previous/Next pages) from the Blender Wiki for the official Blender Video Sequence Editor documentation.
See Using Nodes (Blender Wiki) for getting started with the Compositing Nodes.
Check out Ubuntu for getting started with GNU/Linux.
Notes for editors new to Ubuntu
While many to most things in Ubuntu will “just work” this is not the case for most things related to FireWire. Either plan some time for testing/research or get help from an experienced Linux user. It may be best not to rely on external FireWire drives for anything other than copying media to your internal drive as long as you are new to Ubuntu.
For help with all things related to Ubuntu and multimedia see also the Ubuntu Studio pages.
(26.04.09:) Ubuntu 9.04 users who experience problems with opening Blender 2.48a, exporting video with sound or audio playback in Blender see the fourth comment for solutions!