xpra icon
Bug tracker and wiki

This bug tracker and wiki are being discontinued
please use https://github.com/Xpra-org/xpra instead.


Changes between Version 2 and Version 3 of Ticket #1951, comment 1


Ignore:
Timestamp:
09/10/18 04:46:12 (3 years ago)
Author:
Antoine Martin
Comment:

Legend:

Unmodified
Added
Removed
Modified
  • Ticket #1951, comment 1

    v2 v3  
    1 New heuristics mostly done in r20366 - see commit message for details.
    2 To revert to the previous behaviour, just use {{{--video-scaling=VALUE}}} instead of {{{video-scaling=auto}}}.
     1New heuristics mostly done in r20366 - see commit message for details. Updates in r20367 + r20368.
    32
    4 This new code targets ''1080p at 25fps'', more for "video" content, more for high quality or low speed, and less if quality is low or speed is high.
    5 The details of the calculations can be shown with "-d scaling".
     3Rather than choosing the downscaling using the old opaque heuristics, the new code will choose the downscaling required to fit within a target budget. This target can come from:
     4* the current bandwidth-limit value (#417) if we have one (roughly converted into a number of uncompressed pixels per second), the lower bound used here can be changed using {{{XPRA_SCALING_MIN_PPS=1000000}}} (ie: 1 megapixels per second), defaults to 240p25: 25*320*240
     5* the default target, configurable using {{{XPRA_SCALING_PPS_TARGET=16000000}}} (ie: 16 megapixels per second) defaults to 1080p25: 25*1920*1080
     6* some of the details of the calculations can be shown with "-d scaling"
     7* to revert to the old heuristics, just use {{{--video-scaling=VALUE}}} (with {{{VALUE=1}}}) instead of {{{video-scaling=auto}}}.
     8
     9Tweaks:
     10* we allow more downscaling for "video" content (by a factor of 2), since we care less about lossy pixel data with video content - we take into account the "video" content-type hint (#1952)
     11* we allow less downscaling for shadow servers (by a factor of 2): those sessions are using video encodings because of how we refresh via a timer and not because this is actually video content, a shadow session usually looks garbled with downscaling applied
     12* if the quality is above 80 (90 for video), no downscaling is used unless required by the video encoder's size limits
     13* higher quality or low speed settings raise the target, low quality or high speed lower it - roughly by a factor of ~4 each (r20369)
    614
    715Still TODO:
    8 * make target configurable? (XX MPixels/s?)
    9 * take into account fullscreen and maximized flags
     16* take into account fullscreen and maximized flags?
    1017* UI updates