1 --- !ditz.rubyforge.org,2008-03-06/issue
2 title: Provide a means of offering rendering including frames just prior to, or clustered around, the current frame.
4 avgf provides rendering from first frames, but a generalized technique would be useful,
5 allowing for rendering from recent frames, or frames clustered around a given frame
6 (e.g., for facilitating time-salient visp rendering).
10 reporter: David Hilvert <dhilvert@auricle.dyndns.org>
13 creation_time: 2009-03-21 18:14:31.509749 Z
16 id: fe7d4be60f4124201844a23c0d70a87792494f9a
18 - - 2009-03-21 18:14:32.286613 Z
19 - David Hilvert <dhilvert@auricle.dyndns.org>
22 - - 2009-03-21 18:17:55.990551 Z
23 - David Hilvert <dhilvert@auricle.dyndns.org>
26 Note that the saturation argument appears to provide for this to some degree
27 (e.g., allowing for avgf and medianf), but that further facilities may
28 be required for things like analogous avgl and medianl, or avga
29 (avg about) and mediana (median about).
30 - - 2009-03-21 18:22:04.913045 Z
31 - David Hilvert <dhilvert@auricle.dyndns.org>
34 Note that it might be useful to make AVG/MEDIAN (etc.) independent from FIRST/LAST,
35 perhaps extending the latter group to include an ABOUT (or some such), but that the
36 details of implementation, and interaction with saturation values are not immediately
38 - - 2009-03-21 18:42:08.233573 Z
39 - David Hilvert <dhilvert@auricle.dyndns.org>
42 Consider that, if invariant concerns are made independent, one of the
43 concerns that might be worthwhile to make independent might be the
44 distinction between weight saturation and frame count saturation.
45 - - 2009-03-21 21:55:40.103253 Z
46 - David Hilvert <dhilvert@auricle.dyndns.org>
49 Consider that the correct approach to modifying the saturation behavior is
50 probably to separate saturation into two kinds, one that takes weight within
51 frames into account, and one that does not; the latter would naturally carry
52 over to Irani-Peleg renderers from an incremental reference, whereas the former
53 would not. Implementation in ALE could proceed with separate invariant types
54 avgf:<x> and avgff:<x>, or some such, in order to simplify the concept
56 - - 2009-03-22 11:58:03.886135 Z
57 - David Hilvert <dhilvert@auricle.dyndns.org>
60 Note that weighted saturation might just as well be carried over from incremental
61 to Irani-Peleg renderers, both for the sake of simplicity and because there seems
62 to be no obviously better approach (aside from allowing the client to set the values
63 independently, which can be done implicitly by not propagating settings from
64 the incremental renderer to an existing renderer using this as a reference).