After Effects
Put the depth MP4 under your footage and hide it. Choose it as the Blur Map in Camera Lens Blur for depth of field, use it with Displacement Map for 2.5D parallax, or as a luma matte for fog and haze.
VIDEO TO DEPTH MAP
Upload a clip and get a frame-consistent depth map video for depth of field, fog, 2.5D parallax, compositing or ControlNet. It runs in the cloud with Video Depth Anything, so there is nothing to install and no GPU to own.
1. Upload your video
2. Colormap
Grayscale is the working format: brighter means closer. The color maps are for viewing.
3. Output
Upload a video to see the price.
Your uploads are not shown in a public gallery.
Free accounts receive a moving site watermark. Any purchase unlocks the clean original.
A depth map stores distance from the camera in every pixel. In Grayscale, white is nearest and black is farthest.
Upload an MP4 or MOV clip of up to 60 seconds on this page. It runs Video Depth Anything (Large), which keeps depth consistent from frame to frame instead of flickering, and returns an MP4 depth map video at the source frame rate, up to 1080p. In Grayscale, brighter means closer to the camera; Turbo, Inferno, Magma and Viridis are color views of the same depth. There is nothing to install and no GPU needed. It costs 3 credits per second of video, and failed jobs are refunded.
| Model | Video Depth Anything Large (CVPR 2025), run on fal |
|---|---|
| Input | MP4 or MOV, 1–60 seconds, up to 100 MB and 2,400 frames |
| Output | MP4 (H.264) depth map video at the source frame rate |
| Resolution | Source (up to 1080p), 1080p, 720p or 480p |
| Colormaps | Grayscale (brighter = closer), Turbo, Inferno, Magma, Viridis |
| Side by side | Optional: the original and the depth map in one frame, for checking |
| Price | 3 credits per second, rounded up to the next whole second; failed jobs are refunded |
| Free accounts | Can generate with their credits; results carry a moving tryh3.com watermark until the first purchase |
Three steps, all in the browser. The depth estimation runs on our servers.
Choose an MP4 or MOV clip of 1–60 seconds and up to 100 MB. The button shows the price before you generate: 3 credits per second.
Keep Grayscale for After Effects, Blender, Resolve, Nuke or ControlNet. Source keeps your resolution up to 1080p. Turn on side by side if you want to compare the result with the original.
Download the depth map as an MP4 when it is ready. It is also saved in Creations, so you can leave the page while it runs.
The result is an ordinary MP4. Import it next to your footage, keep the two in sync, and use its brightness wherever a tool asks for depth.
Put the depth MP4 under your footage and hide it. Choose it as the Blur Map in Camera Lens Blur for depth of field, use it with Displacement Map for 2.5D parallax, or as a luma matte for fog and haze.
Load the depth video as a movie texture on a subdivided plane and drive a Displace modifier for 2.5D parallax. For the compositor’s Defocus node, invert it first: Blender treats larger values as farther away.
Use the depth MP4 as a mask or blur map in Fusion for depth of field, fog or relighting. It also works in the free version of Resolve, which does not include the Studio-only Depth Map effect.
Copy the depth into a channel and drive ZDefocus. Set the depth math to match a map where brighter means closer.
Load the Grayscale MP4 with a video loader node and use it as the depth control video for a depth ControlNet, Wan or AnimateDiff. It keeps the layout and camera motion while you restyle the shot. Depth ControlNets expect near = white, which is what Grayscale gives you.
Bring the clip in with a Movie File In TOP to displace a point cloud or mask depth-based effects. In Premiere Pro it works as a track matte for depth-limited grading.
Running a single-image depth model such as Depth Anything V2 or MiDaS on every frame estimates each frame on its own. The depth scale shifts from one frame to the next, so the map flickers, and a blur or fog effect driven by it pulses.
Video Depth Anything (CVPR 2025) estimates depth across neighbouring frames together, so a wall keeps the same shade while the camera moves. That is what makes the result usable as a depth pass for compositing or as a control video for AI video generation.
You pay only for the length of your clip: 3 credits per second, rounded up to the next whole second. Resolution and colormap do not change the price.
| Clip length | Credits |
|---|---|
| 5 seconds | 15 |
| 10 seconds | 30 |
| 30 seconds | 90 |
| 60 seconds | 180 |
Online and in-browser tools that turn a video into a depth map, as listed on their own pages.
| Tool | Where it runs | Limits | Price |
|---|---|---|---|
| TryH3 (this page) | Cloud · Video Depth Anything Large | 1–60 s, 2,400 frames, 100 MB, up to 1080p | 3 credits per second |
| WaveSpeedAI | Cloud · Depth Anything | Up to 10 minutes, output up to 30 fps | $0.005 per second, 3-second minimum |
| Artificial Studio | Cloud | Not listed | 8 credits per second |
| YouArt | Cloud | 100 MB per clip; MP4, MOV, WebM | 8 credits per second |
| Depth Video Studio | In your browser · Depth Anything V2 Small | 60 s and 200 MB; desktop Chrome or Edge | Free |
A video depth map is a video in which each pixel’s brightness shows how far that point is from the camera, with one depth frame for every frame of the original. Editors use it for depth of field, fog, relighting and 2.5D parallax, and AI video tools use it as a control input that keeps the scene’s layout.
Yes, this page. Upload an MP4 or MOV of up to 60 seconds, pick a colormap and click generate. It runs in the cloud, so you need no software, no GPU and no command line, and you download the result as an MP4.
On this page: upload the clip, choose Grayscale, generate and download. Doing it yourself means extracting the frames, running a depth model such as Video Depth Anything on a GPU and putting the frames back into a video; this page does those steps for you.
Free accounts can use the credits on their balance; their results carry a moving tryh3.com watermark until the first purchase. Each second of video costs 3 credits, and failed jobs are refunded.
It is built not to. Video Depth Anything estimates depth across neighbouring frames, so surfaces keep the same shade as the camera moves, unlike running a single-image depth model on each frame.
Use Grayscale for any software or AI model that reads depth: white is near, black is far. Turbo, Inferno, Magma and Viridis are for looking at the result and showing it to others.
Yes. The result is a standard MP4. In After Effects use it as the Blur Map in Camera Lens Blur or with Displacement Map; in Blender drive a Displace modifier or invert it for the Defocus node; in Resolve use it as a mask or blur map in Fusion.
Yes. The Grayscale output uses the near = white convention that depth ControlNets expect. Load the MP4 with a video loader node and use it as the depth control video for ControlNet, Wan or AnimateDiff.
MP4 and MOV files from 1 to 60 seconds and up to 100 MB. Clips can have at most 2,400 frames, which is 80 seconds at 30 fps or 40 seconds at 60 fps. Trim longer videos first.
It depends on the length and resolution of the clip. You can leave the page while it runs; the finished depth map is saved in Creations.
No. AI depth from ordinary video is relative: it shows what is nearer and farther, not distances in meters. For measured depth you need a depth camera or a calibrated setup.
Yes, for footage you have the rights to use. Free-account results carry a watermark; any purchase unlocks the watermark-free version.
Upload a clip of up to 60 seconds and get a frame-consistent depth pass for 3 credits per second.
Upload a video