Convert between SBS, top-bottom and red/cyan
A stereo pair can be packed three ways and at two resolutions — and every player wants a different one. This page re-packs a stereo image in your browser, and works out from the pixels which half is the left eye, so the result is never inside-out.
Which layout for what
| Layout | Shape | Wanted by |
|---|---|---|
| Side by side (SBS) | Twice as wide as tall | Quest, Pico, Vision Pro, XREAL, VITURE, VR players, 3D TVs, YouTube |
| Top & bottom (TB / OU) | Twice as tall as wide | 3D Blu-ray rips, some phone and VR180 exports, a few older players |
| Red / cyan (anaglyph) | Normal photo shape | A plain laptop or TV screen with paper glasses — no headset |
Half versus full is a separate choice. Half keeps each eye at half the frame width so the file stays the size of a normal video; full gives each eye its native resolution, doubling the frame. Headsets usually resolve far less than a full-resolution eye in motion, so half is the safe default and full is for archiving or projection.
Why it detects the eyes instead of assuming
The layout does not say which half is the left eye — that is a convention, and files violate it. Getting it backwards does not look like a bug: the picture still reads as 3D, but depth is inverted, near things sink into the screen. Most people can't tell whether the file is wrong or their eyes are.
So the two halves are compared directly. Anything nearer than the screen is drawn further to the right in the left-eye image; the tool measures that offset with a greyscale block match and reports it, together with how clearly it beat the no-offset case (the confidence). If the scene has no depth — a flat wall, an all-background shot, two identical halves — there is genuinely no direction information, the confidence lands near zero, and the page says so rather than pretending. Swap eyes is always one click away.
What actually changes
- SBS ⇄ TB at full resolution — pixels are copied, not resampled. The picture is bit-for-bit the same; only the arrangement differs.
- Half resolution — each eye is resampled to half size. That is what most players expect by default and what keeps the file the size of a normal video.
- Red / cyan — the left eye keeps its red channel, the right eye keeps green and blue, and the two are combined. It is lossy by design: that is the format, not a shortcut. Use it to watch on a screen, not as your master file.
Limits, stated plainly
- Images only. Video needs the same work per frame — use the converter, or re-mux.
- Exactly half and half. The split is taken at 50%, so a file cropped unevenly comes out offset.
- Anaglyph back to stereo is not offered. Recovering two colour views from one red/cyan image is guesswork; the tool would rather not pretend it is a conversion.
- No alignment correction. If the two halves are also misaligned vertically or rotated, that stays as it is.
What if the preview looks inside-out?
Press Swap eyes. The measurement is a strong hint, not an oracle: content that sits entirely behind the screen inverts the sign, and a frame with no parallax at all carries no information. The status line tells you which case you got.
Can I convert a whole folder?
Not yet. One image at a time is what the browser gives us for now; the conversion itself takes a few milliseconds, so the limit is clicks, not compute.
Does this work with photos from a phone's 3D mode?
If the phone saved a side-by-side or top-and-bottom pair, yes. If it saved a single image plus a depth map in separate files, that is a different format and this page cannot rebuild it.
Related: About DepthSnap · TB to SBS with eye detection · Play SBS/TAB files in the browser · SBS vs top-bottom · Half vs full SBS · All tools
DepthSnap — free browser-based 2D to 3D SBS converter. No upload, no signup, no watermark.