How to Blur Faces in DaVinci Resolve (And When to Automate It)

DaVinci Resolve is a professional video editor, and yes — you can blur faces in it. But doing it correctly on anything longer than a short clip is slow, manual work. This guide covers exactly how to do it in Resolve, where the manual approach runs out of steam, and how automated tools handle the same task without the frame-by-frame grind.

TL;DR

• DaVinci Resolve blurs faces using Power Windows + the built-in tracker in the Color page — it works, but every face needs its own mask and manual correction passes.
• Tracking drifts on fast motion, occlusions, and crowded frames — you fix those by hand, which adds up fast.
• For multi-face or longer clips, an automated tool like Medianonymizer detects and tracks all faces without manual masking and produces an irreversibly re-encoded export.
• Blur faces automatically →

The manual method in DaVinci Resolve

Step 1 — Set up your project and go to the Color page

Open your project in DaVinci Resolve and place the clip on your timeline. Switch to the Color page (the palette icon at the bottom). All the masking and blur tools live here.

Step 2 — Add a serial node for the blur

In the node graph, right-click your existing node (or the default node) and select Add Node → Add Serial. This isolates the blur effect on its own node so it doesn't affect color grading on other nodes.

Step 3 — Draw a Power Window around the face

In the new serial node, open the Power Window panel (the second icon in the toolbar on the left). Select the Circle or Polygon shape and draw it over the face in the first frame where it appears. Resize and position the shape to cover the face snugly.

Step 4 — Invert the qualifier and apply blur

With the Power Window active, click Invert so the effect applies inside the mask (not outside). Then go to the Blur panel, increase the Radius slider until the face is unrecognizable. A radius of 30–50 works for most footage.

Step 5 — Track the mask across frames

Open the Tracker panel and click Track Forward (the right-pointing arrow). DaVinci Resolve will attempt to follow the Power Window across subsequent frames by analyzing motion in the video. Watch the playback — when the mask drifts off the face, stop the tracker, correct the shape manually by setting a keyframe, then resume tracking from that point.

Repeat this correction loop until the mask covers the face through the end of its screen time. For each additional face in the clip, add another serial node and repeat the entire process.

Step 6 — Export with the blur baked in

Go to the Deliver page, choose your output format (H.264 or H.265 are typical), and render. The blur is only irreversible after you render — in the project file the node effect is non-destructive and could be removed by anyone with access to the project. Discard the project if the source footage is sensitive.

Where this approach breaks down

Scenario	Manual Resolve approach	Result
Single face, static interview	Draw once, track, a few corrections	Manageable — 20–40 min
Two faces moving in conversation	Two nodes, two correction passes	Time-consuming — 1–2 hours
Street footage, 5+ faces	One node per face, constant drift corrections	Very slow — multiple hours
Face turns away or is occluded	Tracker loses the shape entirely	Manual keyframe required every time
Audio contains spoken names	Not covered by blur tools at all	Separate manual step

Resolve's tracker is designed for grading work — isolating a sky, a skin tone, an object. It was not built for identity anonymization at scale. For a two-minute interview clip it is perfectly usable. For bodycam footage or a conference recording with ten speakers, the manual correction loop becomes the bottleneck.

The flickering problem

When the tracker loses a face and you don't catch it, you get exposed frames — the blur drops for a few frames and the face is visible. At 25 or 30 fps, three exposed frames take a tenth of a second of screen time, which is invisible during casual review but trivial to extract frame-by-frame. This is the same failure mode described in our guide to stable video anonymization.

When to use an automated tool instead

Manual masking in Resolve makes sense when:

• You have a short clip (under two minutes) with one or two faces.
• You need precise control over which specific region gets blurred.
• You are already doing color work in Resolve and want to keep everything in one tool.

Automated anonymization makes sense when:

• You have multiple faces, a longer clip, or batch files to process.
• You need the audio track handled in the same pass.
• You need a documented, auditable output — not just a rendered file but a record of what was redacted.
• Compliance requirements (GDPR, CCPA, HIPAA) mean you need irreversibility guaranteed, not assumed.

How Medianonymizer automates this

The Medianonymizer approach to blurring faces in video separates two concerns: AI locates every face across every frame using detection plus geometric tracking; then deterministic code applies the blur to those exact coordinates and re-encodes the pixel data.

The key differences from the Resolve workflow:

• No manual masks. Every face in every frame is found automatically — you don't draw a single Power Window.
• Tracking with interpolation. When a face is missed in a frame, geometric tracking holds the blur in place over the interpolated position. No exposed frames.
• Audio in the same pipeline. Spoken names, phone numbers and other PII are located with speech-to-text plus entity recognition, then beeped or silenced on the waveform — not a separate editing task.
• Irreversible by design. The output is a re-encoded video file. There is no project file with removable nodes. The original pixels are gone.

A practical checklist before you call a video anonymized

Whether you used Resolve or an automated tool, verify these before sharing the file:

• Every face is covered in every frame, including frames where the subject turns or is partially occluded.
• The blur is baked into the rendered file, not a removable layer or project node.
• The audio track has spoken names, numbers and other identifiers silenced or beeped.
• Container metadata (GPS, device ID, creation timestamp) is stripped from the file.
• You have done a spot-check — scrub through the hardest parts of the clip manually to verify no exposed frames slipped through.

Blur faces in your video now

If you have a clip that needs faces blurred — whether it's a single interview or an hour of surveillance footage — upload it and let the pipeline handle detection, tracking and audio in one pass. No account required to start.

Blur faces automatically →