Replicated Attenuation 1

 

We apply the EMOTE Scale operation to the upper body of a query marching clip. This operation is repeated over all matches to the query marching clip. The EMOTE Scale operation applied onto the matches is the same intensity as that applied onto the query. The EMOTE Scale operation has the effect of attenuating the motion. This is done by interpolating each joint's quaternion towards a its corresponding smoothed quaternion motion. In this example, the smoothed motion is smoothed over 25 keyframes. The interpolation intensity is 0.8 where 0 is the original motion and 1 is the smoothed motion.

Matching parameters: DTW, rotational parameterization, full-body area, no global position, no global rotation.

 

Unedited Query and Matches

 

Video Description

(in order of appearance)


Full-size video

  1. The query
  2. The  best n-matches, from best to worst.

 

Edited Query and Matches

 

Video Description

(in order of appearance)


Full-size video

  1. The edited query
  2. The edited best n-matches, from best to worst.

 

Edited Query and Matches

 

Video Description

(in order of appearance)


Full-size video

  1. The edited query in context (i.e. with previous 50 and following 50 frames of animation).
  2. The edited best n-matches in context  in context (i.e. with previous 50 and following 50 frames of animation).

 

 

Contextualised Replicated Attenuation 2

 

This is an example of contextualised editing.

Same example as above except that the intensity of the EMOTE Scale is controlled by the match strength. Therefore, the more similar the match is to the query, the more it will be attenuated.

 

Contextualised Attenuation Example 2

 

Video Description

(in order of appearance)


Full-size video

  1. The query
  2. The  best n-matches, from best to worst.
  3. Long pause
  4. The edited query
  5. The edited best n-matches, from best to worst.
  6. Long pause
  7. The edited query in context (i.e. with previous 50 and following 50 frames of animation).
  8. The edited best n-matches in context  in context (i.e. with previous 50 and following 50 frames of animation).