What is 3D Parallax?

3D Parallax is a free tool that estimates depth from a single image and creates a pseudo-3D (parallax) MP4 video that moves with a sense of dimension. As the foreground subject and the background shift slightly out of sync, a flat photo gains depth and presence. No expertise or paid software is needed, so you can easily produce eye-catching "moving photos" for social media.

How to use

1. Tap "Select image" and pick one photo to bring to life. Photos with a clear distinction between near and far elements show the effect best.
2. On your first use, the in-browser depth model (about 26MB) and runtime load automatically. After that they are cached, so the wait is shorter.
3. The model estimates the image's depth (a depth map) inside your browser. This is the basis for moving near subjects more and distant ones less.
4. Adjust how the parallax looks using the provided parameters, such as motion strength and camera sway.
5. Check the motion in the preview, and fine-tune the settings if it feels too strong or too weak.
6. Export as MP4. The encoding also happens in your browser, and the finished video file is ready to download.
7. Use the completed MP4 directly in social posts, profiles, stories, and more.
8. To try another photo, just pick a new image and repeat the same steps.

Great for

• Eye-catching social posts

  Turning a still photo into a moving shot with depth makes people pause while scrolling, helping you earn more likes and views in the feed.

• Bringing travel and landscape photos to life

  Scenery like mountains or cityscapes with clear depth gains immersion from a touch of parallax, showing off your memories more dramatically.

• Animating profiles and avatars

  Gently animating a portrait creates a profile video more memorable than a still image, strengthening first impressions in intros and on social media.

• Showcasing product photos

  Adding dimensional motion to product shots makes them more appealing than flat catalog images and gives you ready material for short videos.

• Creating short-video material

  Mass-produce motion clips from a single photo for reels and shorts, quickly expanding the variety of footage you have to edit with.

• Presenting illustrations and artwork

  Adding parallax motion to illustrations or art gives even flat works a sense of depth, helping your portfolio or promotion stand out.

Privacy

Your image is processed by WASM inside your browser and is not sent to a server. Only on the first use does it fetch the depth model (about 26MB) and the inference runtime (WASM) from a CDN (Hugging Face / jsdelivr), but that is fetching static files, not sending your image anywhere. Both the estimation and the video encoding finish entirely on your device.

FAQ

Q. Is my image sent to a server?

A. No. The image is processed in your browser on your device and is not sent to a server. Depth estimation and video creation both finish locally.

Q. Why does it use the network on the first run?

A. To fetch the depth model (about 26MB) and the inference runtime from a CDN. That is downloading static files; your image is not sent externally.

Q. What kind of photo shows depth best?

A. Photos with a clear sense of distance between the foreground subject and the background. Compositions that separate subject from background, or near from far, show the parallax effect best.

Q. What format is the output video?

A. It is exported as MP4. You can use it directly in common video environments such as social media and messaging apps.

Q. How long does generation take?

A. It depends on your device's performance and the image size, but estimation and encoding run in your browser. The first run adds time for loading the model.

Q. Can I adjust the motion strength?

A. Yes. You can change the motion strength and camera sway with the provided parameters, and fine-tune to your taste while checking the preview.

Q. Is there any cost?

A. No. It is free to use, with no limit on the number of videos and no sign-up required.

Q. What technology is used for depth estimation?

A. It runs a model called Depth Anything V2 Small in the browser via transformers.js. Rather than calling a server-side AI, it performs inference on your device.