2D to 3D SBS Converter
Video or photo → side-by-side 3D, with AI depth. Runs in your browser: no upload, no signup, no watermark.
or click to browse
How it works
Video: every frame gets its own depth, then the left and right views are rebuilt with gap filling, so the result plays on any headset.
Photo: one depth pass → SBS in about a second. Nothing ever leaves your device.
About 80 seconds for a 30-second clip on a modern laptop. Real timings & headset playback settings →
About · Free 2D to 3D SBS Converter
How does a file become 3D?
DepthSnap estimates depth for your file inside the browser tab. For a photo that is a single pass: every pixel gets a depth value, then the left and right views are rebuilt into a side-by-side pair plus a depth map. For a video the same thing happens frame by frame, and the frames are stitched back into a normal video file — all on your device.
Which devices work?
The SBS output plays on Meta Quest 2/3, Apple Vision Pro, Pico 4, XREAL and VITURE glasses, 3D TVs, projectors and any glasses-free 3D display — anaglyph too. No re-encoding needed.
Why local processing?
100% on-device: files are never uploaded; no limits, watermarks or accounts. The AI model downloads once, is cached, and works offline afterwards.
FAQ
Getting started
What file types can I drop in?
Video: MP4, MOV, WebM, MKV. Photos: JPG, PNG, WebP. If your browser can decode it, the converter can read it — the file type is detected automatically.
Do I have to pick video or photo mode?
No. Drop the file into the single box and the right pipeline is picked for you: a clip is converted frame by frame, a photo takes one depth pass and comes back in about a second.
Is there a size or length limit?
No quota from us — nothing is uploaded. Big files (roughly over 150 MB or 10 minutes) are written straight to a file you pick, so the output is not limited by the memory of the tab; the tab has to stay open while it runs. Past 15 minutes the audio track is left out (decoding a whole soundtrack in the tab is what makes big files crash) — the picture is complete, and you can mux the original audio back in any editor.
How long does a video take?
On an M2-class laptop the default settings take roughly 80 seconds for a 30-second clip; refreshing depth on every frame roughly doubles that. The page shows an estimate before you start, and “every 2nd frame” halves the work with no visible difference on slow camera moves.
Does it work on a phone or inside a headset browser?
It loads on phones and in the Quest/Pico browser, but conversion needs a desktop-class GPU to be quick. Phones work for photos; for a video, convert on a computer and copy the file across.
The result
Why are there two images side by side?
That is the SBS format: the left half is for your left eye and the right half for your right eye. Every VR headset and 3D display understands it.
How do I watch the result on my headset?
Copy or download the file, open it in a player that supports SBS (Skybox, DeoVR, 4XVR on Quest; Pico Video; the Files app on Vision Pro), and set the format to “Side by side” if it does not switch by itself.
It plays as one wide, flat video. What went wrong?
The player did not recognise the file as stereoscopic and is showing both eyes at once. Open the 3D or format menu in the player and choose SBS / side-by-side; the file itself is correct.
Why does the 3D look stretched or doubled?
Stretched means the player is treating an SBS file as a normal widescreen video, or the wrong eye order is selected. Doubled edges usually mean the 3D strength is too high for that footage — drop it one step.
How strong should the 3D be?
Start with soft. It is the most comfortable setting and it is what most people keep; raise it if the scene feels flat, lower it if edges start to ghost or your eyes feel strained.
Can I see the depth without a headset?
Yes. After conversion switch the preview to Wiggle to see the parallax move, or Red / cyan if you own anaglyph glasses. Disparity wipe sweeps a divider so you can see the actual pixel offset. All three run on any normal screen.
Is the audio kept?
Yes, audio is carried through for video files (the mp4 gets an AAC track). Turn “keep audio” off if you want a silent result. For videos longer than about 15 minutes the audio is left out on purpose — see the note on large files above.
What format is the download?
An .mp4 with H.264 video and AAC audio — the same thing a phone or camera records, so it opens in QuickTime, VLC, Windows Media Player, a TV, or straight in a headset player. It is also tagged as side-by-side 3D, so players that understand the tag switch on their own. If a browser cannot encode H.264 we still keep the .mp4 container and use AV1 or VP9 inside it; only the rare browser that cannot produce an mp4 at all gets a .webm. The line under the result always says which one you got.
Can it handle a feature-length film?
Technically yes, and that is what the disk-streaming path is for: a 2-hour file no longer has to fit in the tab's memory. The real limit is time. On an M2-class laptop a 60-second clip takes about 4.5 minutes, so a two-hour film is roughly five hours — leave it running with the tab in the foreground; it shows an estimate first and can be cancelled at any point. Long clips decode sequentially instead of jumping around, which is what makes them finish at all. For anything you only want to preview, convert a short excerpt.
Privacy and limits
Are my files uploaded?
No. The depth model and the stereo rebuild both run inside this browser tab, and the video is encoded by the browser itself. Your file never leaves the device.
Is it really free?
Yes — no signup, no watermark, no usage counter, no upsell. The compute is your own machine, so there is nothing to bill.
Does it work offline?
After the first visit the model is cached by the browser, so a second conversion normally works with no network. A hard refresh, private window or cleared cache will download it again.
What happens if I close the tab mid-conversion?
The job stops and nothing is saved. Your original file is untouched; just open the page and drop it again.