We like to keep things short and sweet, so here is how you extract frames from a video in Blender 2.9.
Load up Blender
- Click on the ‘editor type’ button, just below the blender logo and ‘File’ button
- Change the viewport to the Video Sequencer
- Click Add to add the Video Sequence
- Make sure that the clip length is set to the length of the video – use the ‘end’ panel
- Click on the Camera Render settings (looks like a DSLR camera on the right of Blender) and then Color Management
- Change the Color Management from Filmic to Standard. Filmic adds a soft overlay to it which is not what you want because you want the original looking image.
- make sure the Resolution in the ‘Output panel’ (this looks like a printer halfway through a print) is the same as the video.
- To find out the resolution, we use VLC. Open VLC and add your video file through the ‘Media’ menu
- Click tools -> Codec Information
- in the panel that comes up, you will see the video Resolution.
- Make sure the FrameRate is the same. Some video record at weird 29.3 fps, or 27.2 etc (you get the idea) – we normally round this up to the closest normal frame rate.
- Set your output location (the default is your temp folder). make sure you have enough space on your hard drive as the multiple images will quickly take a lot of space on your hard drive
- We use JPG to save space as we don’t need PNG with the alpha channel as not many video types have alpha channels.
- Once you are ready, render out the animation
Reasons why? Well a lot of programs prefer working with JPG or single image files due to not needing to convert the video file as its being read in. We use this technique with our Reallusion 3D Facial MoCap set up. We record our sequences on our static cameras, load the video footage into Blender, extract the frames, and then we use that copy for inputting into our Faceware Motion Live plugin for Reallusion iClone and then to Unreal Engine 4 – pretty cool ain’t it.