Aller au contenu

Splitter Node

By Kriss Norde

The Splitter splits a combined stereo stream into two independent streams, Left and Right. Compatible with the Side-by-Side, Top-Bottom, Stacked and Interlaced formats. The first node of any stereo pipeline that starts from a single unified source file.

Overview

The Splitter is the reverse of the Combiner: it takes an image in which both views are packed (SBS, TB, etc.) and exposes two separate outputs, Left and Right. You can then process each view independently and recombine them with a Combiner at the output.

It detects the format automatically in some cases (based on the image ratio), but in general you have to set the format manually.

Where to find this node

  • Effects Library: Tools > Stereo > Splitter
  • Quick add: right-click in the Node Editor → Add Tool > Stereo > Splitter

Splitter vs Combiner

Splitter
Splits SBS/TB/etc. into two streams. The first step of the pipeline.
Combiner
Assembles two streams into SBS/TB/etc. The last step of the pipeline.

Canonical pipeline: MediaIn (SBS) → Splitter → [L and R comp] → Combiner → MediaOut (SBS).

Inputs

  • Input (orange) - the combined stereo stream.

Outputs: Left and Right (two separate outputs).

Setup

Standard stereo pipeline:
MediaIn (SBS) → Splitter ─┬── Left → Color Corrector
                          │            ↓
                          │       (rest of the L processing)
                          │
                          └── Right → Color Corrector
                                         ↓
                                    (rest of the R processing)
                                         ↓
                          [merged back in a Combiner]
                                         ↓ MediaOut (SBS)

Format conversion:
MediaIn (TB) → Splitter (Format = Top-Bottom)
            ─┬── Left ─┐
             │         ├── Combiner (SBS)
             └── Right ─┘
                        → MediaOut (SBS)

Multi-format input:
MediaIn (Interlaced) → Splitter (Format = Interlaced)
                    ─┬── Left
                     └── Right
                          → separate processing

Controls tab

Format (default: Side-by-Side)
The format of the input stream. SBS, TB, Stacked, Interlaced. A mismatch = an incorrect split (views offset/swapped/mixed up).
Aspect Compensate (default: On)
Restores the correct aspect ratio after the split. SBS halves the X axis, TB halves the Y axis → this option stretches it back to recover the original ratio.
Swap Eyes (default: Off)
Swaps the Left and Right outputs. Useful if the source uses the reverse convention (right first).
Resize Filter (default: Linear)
The filter used for the re-stretch when Aspect Compensate is on. Linear = standard, Cubic = higher quality.

Use cases

  • Standard stereo pipeline - the first node after MediaIn to start per-view compositing work.
  • Format conversion - Splitter (TB) → Combiner (SBS) = convert a TB master to broadcast SBS.
  • Asymmetric L vs R processing - for example, one camera has a flaw the other does not. The Splitter lets you fix only the affected view.
  • Color matching between the two eyes - Splitter to expose L and R, then a dedicated Color Corrector on each, before the Combiner.
  • Analyzing / debugging a single view - show only Left or Right through the Splitter to check a detail without stereo interference.
  • Delivery format conversion - a pipeline that takes broadcast SBS as input and produces Stacked archives through Splitter → Combiner (Stacked).

Tips

  • Format = critical - check the actual format of the source file (read the metadata, or test visually). A mismatch = mixed-up views and completely wrong compositing.
  • Aspect Compensate depending on the use - turn it on to process each view with the correct ratio. Turn it off if you deliberately keep the "squashed" source ratio.
  • Swap Eyes when in doubt - if after the Splitter the 3D effect looks odd (inverted depth), try Swap Eyes.
  • Check with an anaglyph - Splitter → Anaglyph → viewer lets you quickly confirm that the split and the convergence are correct.
  • Performance - the Splitter turns 1 stream into 2 → all downstream processing is done twice. Cache the heavy nodes on each branch to save time.
  • Imperfect format auto-detection - Fusion can try to detect it automatically, but you often have to specify it. Always check the result.
  • For dual-camera masters (two separate files), no Splitter is needed - connect two MediaIn nodes (one per eye) directly to the rest of the pipeline.