Your First Conversion: A Practical Walkthrough
A step-by-step guide to converting your first 2D video to stereo 3D with anelo. What to expect, what to check, and common first-timer mistakes.
This walks through a real conversion — start to finish — with the decisions you will face and the mistakes that are easy to make the first time.
Start with short footage
Your first conversion should be 30-60 seconds, not a feature film. Processing time scales linearly with duration, and you want fast iteration on settings before committing hours of GPU time. Pick a clip with clear depth — a person in front of a background, a landscape with foreground and distance, a tracking shot down a hallway. Flat content (whiteboard presentations, screen recordings) produces flat stereo. That is correct behavior, not a bug.
Preflight tells you what you have
When you load a file, the preflight stage runs automatically. It reports resolution, frame rate, codec, duration, and metadata flags. Pay attention to the resolution and frame rate — these determine which pipeline stages are relevant. If your source is already 4K, upscaling is unnecessary. If your source is already 60fps, interpolation may not add value.
Choose a preset or build a recipe
For your first conversion, use the "Standard 3D" preset. It runs: extraction, depth estimation, temporal smoothing, edge refinement, stereo synthesis, SBS packing, and assembly. This is the core pipeline without upscaling or interpolation — fewer stages, faster iteration, and the result tells you whether the depth estimation works well on your content before you add complexity.
If the standard preset produces good stereo, you can add upscaling and interpolation in subsequent runs with a custom recipe.
Preview before committing
Run a preview on a 5-second segment before processing the full clip. The preview uses the same pipeline at the same settings but processes only a short excerpt. Look for: correct depth ordering (near objects in front, far objects behind), absence of visible artifacts (edge bleeding, shimmer, black strips), and comfortable 3D effect (not too flat, not too deep).
If the depth looks wrong, try a different depth model. If the 3D effect is too strong, reduce the stereo baseline. If you see edge artifacts, the temporal smoothing stage may need a higher strength setting.
Common first-timer mistakes
Processing at maximum settings on the first run wastes time. Start with the default settings, evaluate the result, then adjust one parameter at a time. Changing three settings simultaneously makes it impossible to know which change helped.
Choosing too high a stereo baseline produces impressive-looking depth in the first 30 seconds and a headache by minute two. If you are viewing in a VR headset, start with a conservative baseline and increase it only if the effect feels too subtle.
Skipping the preview and processing a full 90-minute film on the first attempt. The preview exists specifically to prevent this. Use it.
Forgetting to check the output format. If you are viewing on a Quest headset, you want HSBS. If you are archiving, you want full SBS. If you are previewing on a monitor, you want anaglyph. Set the output format before processing, not after.
What to expect from the output
A good first conversion looks noticeably three-dimensional in stereo. Objects at different distances sit on different depth planes. Camera motion produces natural parallax. The effect is not as strong as native stereo photography — this is a reconstruction from monocular cues, not a true binocular capture. But for most content, the difference between flat and felt is immediate.
If the result looks flat, the content may genuinely lack depth cues (flat backgrounds, uniform lighting). If the result looks wrong (inverted depth, obvious artifacts), something in the pipeline configuration needs adjustment. The job report will tell you which stage produced warnings.