GIF Creator Learn

Docs

Redact sensitive data fast

Use manual blur, AI blur detection, preview mode, and region propagation together.

Overview

This is the core GIF Creator workflow: upload a recording, detect or draw blur regions, preview the result, and export a safe GIF.

  1. Open the project editor.
  2. Upload a video and wait for frames to appear in the frame strip.
  3. Select a frame that contains sensitive data.
  4. Draw a manual blue blur region, or run AI Detect.
  5. Resize the blur region with the handles until it fully covers the sensitive content.
  6. Use Forward to propagate that region through later frames.
  7. Toggle Preview to validate the blur result.
  8. Encode the GIF.
Editor showing manual and AI blur regions on the updated M365 Health frame
The updated M365 Health example combines one manual region with AI-generated email-address regions directly on the frame before export.
Blur region side panel with manual and AI regions
The right panel lists one manual region plus AI-generated regions. The blue pills, such as `email address`, are the PII classifications returned by the AI for each suggested blur region.

Why propagation matters

If the same password field, badge, or API key appears across many frames, region propagation saves time immediately. You draw one blue region, size it correctly, then copy it to the remaining timeline with a single action.

AI-assisted path

When the frame contains multiple sensitive elements, use the AI detect button to generate blur suggestions, then review the AI labels and token totals before exporting. In this walkthrough, the saved AI suggestions were generated with GPT-4.1, and labels such as email address identify the PII category the model assigned to each suggested region.

The AI prompt is designed to look for common sensitive categories, including:

  • full names, usernames, and display names
  • email addresses
  • phone numbers
  • IP addresses, MAC addresses, and hostnames tied to real machines
  • passwords, tokens, API keys, secrets, and credentials
  • financial data linked to a person
  • passport numbers, national IDs, and other government-issued identifiers
  • internal system names, device names, and organisation-specific identifiers
AI token usage in the editor
After a `GPT-4.1` detection run, the editor shows call counts plus input, output, and cumulative token totals for the project.

Before export

Use preview mode before encoding. It lets you verify that the blur coverage matches the final output rather than the editable overlay view.