Skip to main content

Video to Stills

BEGINNER
📖3 min read

Introduction​

Extracting still frames from video is a common task in visual investigation. You might need to turn a CCTV clip into a sequence of images for photogrammetry, or extract specific keyframes for analysis. This guide covers methods using Blender, FFmpeg, and VLC.

Blender VSE

Visual interface. Best for selecting specific ranges or cropping before export.

FFmpeg

Command line. Fastest for bulk extraction or converting entire videos.

VLC Player

Simple tool for grabbing individual snapshots manually.

Blender is excellent for this because it handles color management correctly and allows you to preview exactly what you are exporting.

  1. Open Blender and go to File > New > Video Editing.
  2. Drag and Drop your video file onto the timeline.
  3. Set Range:
    • Move the playhead to where you want to start. Press Home or set Start Frame.
    • Move to where you want to end. Press End or set End Frame.
  4. Output Settings (Properties Panel > Output):
    • Resolution: Ensure it matches your video source (usually automatic, but check %).
    • Frame Rate: Ensure it matches the video source.
    • Output Folder: Click the folder icon to choose where to save the images.
    • File Format: Choose PNG (lossless) or JPEG (smaller size).
  5. Render:
    • Go to Render > Render Animation (or press Ctrl + F12).
    • Blender will save each frame as a separate image file in your output folder.

Method 2: FFmpeg (Command Line)​

If you have FFmpeg installed, this is the fastest method.

Extract all frames:

ffmpeg -i input_video.mp4 output_folder/frame_%04d.png

Extract 1 frame every second (for photogrammetry):

ffmpeg -i input_video.mp4 -vf fps=1 output_folder/frame_%04d.png

Extract a specific range:

ffmpeg -ss 00:00:10 -t 00:00:05 -i input_video.mp4 output_folder/frame_%04d.png

(Starts at 10 seconds, extracts for 5 seconds)

Method 3: VLC Media Player​

Good for grabbing just one or two specific frames.

  1. Open video in VLC.
  2. Go to Video > Take Snapshot.
  3. The image is saved to your Pictures folder (usually).
  4. Tip: You can use the E key to step through the video frame by frame to find the perfect moment.

Summary​

  • Use Blender if you need visual control and want to export a specific sequence.
  • Use FFmpeg if you need to process large files quickly or automate the process.
  • Use VLC for quick, single-image snapshots.

Documentation for Methodology​

To ensure your work aligns with the Berkeley Protocol and the Guide for Judges, use the following points to document your extraction process.

Methodology: How to document Frame Extraction

In Your Method Section​

Guidance: Describe the purpose, the tool version, and the input data.

  • Investigative Objective: To extract individual frames from video evidence [Video ID] for detailed analysis/photogrammetry.
  • Software Version: Blender [Version] / FFmpeg [Version].
  • Input Data: Video file [Filename] (Hash: [First 6 digits]).

In Your Decision Log​

Guidance: Record the specific procedural parameters and integrity checks.

  • Specific Settings: Output Format (PNG/JPG), Sampling Rate (Every frame / 1 fps), Range ([Start Time] to [End Time]).
  • Assumptions: Assumed the video frame rate was constant.
  • Integrity Check: Verified that the extracted frame count matches the expected duration.
  • Date Performed: [YYYY-MM-DD].

Verification & Mitigation​

Guidance: How did you mitigate bias or verify accuracy?

  • Verification: Checked random extracted frames against the source video to ensure no frame skipping or corruption.
  • Bias Mitigation: Extracted all frames in the sequence rather than cherry-picking, to provide a complete dataset.

Example Methodology Statement​

"Frames were extracted from the source video [Hash: A1B2C3] using FFmpeg. The video was sampled at 1 frame per second to reduce redundancy for photogrammetry. Output images were saved as lossless PNGs to preserve pixel data."