====== Lab tests 2012-05-11 ======


===== Init test (without Torleif) =====

  * Case I (27ms + 27ms):
    * conduct-clap  + conduct-sing (ta-ta-ta father jacob)
    * Vivianne: more behind than last trail (which was 27 + 0ms)
    * Leif Arne: No change noticed
  * Case II (48 + 27, aspect and resolution change in video viewed by conductor)
    * conduct-clap  + conduct-sing (ta-ta-ta father jacob)
    * Vivianne: not much difference
    * leif arne & daniel: Less syncronized, more difficult two follow during slow-down
  * Case III (48 +48, aspect+resolution change both directions):
    * conduct-clap 
    * clap (no conduct)
      * echo (reverberation) noticable and disturbing
    * clap (no conduct, one-way audio)
    * conduct-sing (ta-ta-ta father jacob, one-way audio)
    * ?
  * Case V (48 + 70)
    * Better image, one way audio
    * conduct-clap 
    * clap-clap
      * vivanne can hear delay
    * conduct-sing (ta-ta-ta father jacob, one-way audio)
    * clap-clap with two-way audio

===== Tests with Torleif, Vivanne, Daniel, Leif Arne =====
  * Case 0 (in same room)
    * conduct-clap
    * conduct-sing (ta-ta)
    * clap-clap
    * vivianne: conduct-clap less presise than clap-clap

  * Case I (27+27)
    * conduct-clap
    * conduct-sing (ta-ta)
    * clap-clap
    * Thorleif: conductor is smaller, feels more distant
    * Vivianne: no noticable delay (compared to in-rom)
    * Leif Arne: camera infront no longer an issue (compare to previous test day)

  * Case III (48+48)
    * conduct-clap
      * camear adjustment needed (due to aspect ratio bug)
    * conduct-sing (ta-ta)
    * clap-clap
    * Vivianne: conducting works fine, clapping has delay, on speedup your are behind, on slowdown your are in sync

  * Case V (48 + 70)
    * conduct-clap
      * 30 fps: All noticed, smudged movements
      * Retail with 60 fps
    * conduct-sing (ta-ta)
    * clap-clap
      * Vivianne: 
        * bigger camera make poorer contact with middle person
        * delay is managable, simpilar to what is common with large choir (large choir may even be worse)

  * Case "Screen size"
    * Recorded conductor
    * Vivianne: When conductor is approx real size he seems too close
      * Combination of too close and recorded conductor is unatural
    * Torleif: No problem with enlarged sizes
    * Leif Arne: Keystone problems
    * Vivanne: 75% of real height seems to be a minimum, smaller give you a "look-for-him" feeling
    * Torleif: Size is easy to get used to, close to real size is preferrable though
    * Vivianne: Conductors gaze not in camera (but moving around in virtual choir) makes a more distant feeling.

  * Case VI (with gstreamer video in one pipeline)
   * To be tested next time

  * Case VII (with gstreamer video in both the pipe-line)
    * delay: ~100ms to ~220ms (unknown distribution/shape of variation)
    * conduct-clap
    * conduct-sing (ta-ta)
    * clap-clap
    * conduct-sing (ta-ta)
    * Video quality is poorer, video smeard out
    * Torleif: Worse than Case V
    * Vivianne: Not good. Hard to clap, delay seems much longer than case V. Variance in delay not really noticable. 

  * Case VIII ( choir sees 3D view of conductor )
    * Not to be tested yet due to challenges with technical equipment.
 

=====   New tests with choir of 8 =====
  * Mai 25, preparations from 9:00, choir show 13:00