6
$\begingroup$

I need to hide faces from videos I make. I've found tutorials on blurring, which are great, but the render times are too long to be practical.

So in place of that, is there a way to pixelate the face in using the Video Sequence Editor?

$\endgroup$
7
$\begingroup$

You can use a VES transform effect first to shrink the image...

Pixelate in VSE

Make sure to set interpolation to off.

Shrink VSE

Next and another Transform effect onto the first, and reverse the scale. Enlarging the image. Make sure that Interpolation is off again.

Enlarge VSE Pixelated

$\endgroup$
2
$\begingroup$

I added a pixelate function to the Blender Addon VSE_Transform_Tools using 3pointedit's method. You could download the addon and use the P key to apply pixelation, then set the severity of the pixelation with the tool that pops up.

For a stationary subject, I recommend using the crop tool that also comes with the addon to crop out a face.

enter image description here

For moving subjects, I recommend tracking, then applying a mask.

enter image description here

$\endgroup$
  • $\begingroup$ Hey thats a nice use! I haven't looked at the Transform tool in ages. Glad the crop is better these days $\endgroup$ – 3pointedit Mar 13 '18 at 20:47
0
$\begingroup$
  • The material texture is better suited to the task of quantization.
  • The animation of the 3D View is better suited to specifying location of video manipulation.
  • The VSE has limited effects compared to the compositor.

enter image description here

The image above was quite legible in the original and has been quantized or pixelated in Blender.

enter image description here

In the Blender screen capture above. Note the settings.

  • Interpolation false.
  • Filter FELINE
  • Probes 1
  • Filter Size 20.

Explore Probes and Filter to suit your taste.

In 3D view the material texture can sample images at reduced resolution to produce pixelation. You can create a moving object such as a sphere which might be positioned to cover a face and produce a "mask".

So two texture planes and one mask object are indicated, perhaps in different render layers.

In the Compositor phase of rendering, the mask image can then be used to compose the original video with pixelated video selectively with an alpha over node.

This explanation is a reduction. I cannot comment on speed characteristics of your machine.

$\endgroup$
  • $\begingroup$ To use this technique in the VSE you could add a plane with image mapped then add it's scene strip in your master timeline (using an alternate scene) if you stack the original complete media strip and the scene strip (starting at the same frame) you could meta strip them together. Turn the scene strip off for speed. Use a mask modifier to cut out a shape. $\endgroup$ – 3pointedit Aug 27 '15 at 6:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.