GIFforge

Create polished GIFs from screen recordings directly inside VS Code.

GIFforge converts your screen captures into clean, lightweight GIFs using AI-assisted frame selection and automatic encoding.

Features

• Host-side FFmpeg screen recording (no webview capture dependency)
• Recording mode selection: Full Screen or Custom Region
• Custom Region controls: X, Y, Width, Height
• Live recording duration timer
• Stop Recording and Cancel Recording controls
• Import an existing video (.mp4, .webm, .mov, .avi, .mkv)
• AI-assisted frame selection (with automatic fallback mode)
• Automatic palette-based GIF encoding
• In-panel status updates and GIF preview
• Download/Save generated GIF to any location
• Default output at output/demo.gif in your workspace

Requirements

GIFforge runs locally on your machine.

• ffmpeg is required
• python is optional

If Python is available and core/ai/selector.py exists in your workspace, GIFforge uses AI frame ranking. Otherwise, it automatically switches to fallback frame selection.

Quick Start

1. Open a workspace folder in VS Code.
2. Run command: GIFforge: Start.
3. Choose recording mode:
   • Full Screen for whole desktop capture
   • Custom Region and set X, Y, Width, Height
4. Click Record Screen, then Stop Recording when done.
5. Or use Import Video for existing clips.
6. Click Generate GIF.
7. Optional: Click Download GIF to save anywhere.

Commands

• GIFforge: Start - Open GIFforge panel
• GIFforge: Record Screen - Open panel focused on recording workflow
• GIFforge: Import Video - Open import flow directly

Recording Notes

• Windows: Full and region capture supported via FFmpeg gdigrab.
• Linux (X11): Full and region capture supported via FFmpeg x11grab.
• macOS: Full capture supported via FFmpeg avfoundation; region mode not supported in this release.

How It Works

1. Capture input (record or import)
2. Prepare input video (temp/recording.mp4)
3. Extract frames with FFmpeg
4. Select frames (AI or fallback)
5. Encode optimized GIF
6. Preview and optionally download

Troubleshooting

Recording does not start

• Confirm ffmpeg is installed and available in PATH.
• On Linux/macOS, adjust FFmpeg capture device if needed.
• If host capture is unavailable, use Import Video.

Custom region mode fails

• Verify X and Y are >= 0.
• Verify Width and Height are > 0.
• On macOS, switch to Full Screen mode.

GIF generation fails

• Confirm ffmpeg is installed and available in PATH.
• Try shorter clips (5-20 seconds).
• Ensure workspace is writable (temp/ and output/).

AI selector not used

• Install Python and verify python --version.
• Ensure core/ai/selector.py is present in workspace.

Release Notes

1.0.10

• Added Full Screen / Custom Region recording mode switch
• Added host-side region capture (X, Y, Width, Height) with strict validation
• Preserved import pipeline and GIF generation flow

1.0.9

• Shifted recording to host-side FFmpeg process (reliable fix for webview permission failures)
• Kept import workflow fully intact
• Retained timer and cancel recording UX in panel

1.0.8

• Added live recording timer during screen capture
• Added Cancel Recording flow that safely discards in-progress capture
• Improved recording-state UX while preserving import/generate behavior

1.0.7

• Added real-time in-panel screen recording
• Added Stop Recording workflow and recording save pipeline
• Added Download GIF action to save generated output anywhere
• Kept import-video workflow fully compatible
• Improved panel state handling for recording/processing/done paths

1.0.6

• Fixed non-responsive panel buttons
• Reworked webview-to-extension message bridge
• Improved pipeline reliability and error handling
• Added fallback frame selection when AI path is unavailable

Feedback

If you hit any issue, please share OS details, logs, and sample input video in repository issues.