Rev 1085 | Details | Compare with Previous | Last modification | View Log | RSS feed
| Rev | Author | Line No. | Line |
|---|---|---|---|
| 1085 | pj | 1 | CHANGES |
| 2 | ---------- |
||
| 3 | |||
| 4 | January 9, 1996 to July 17, 1996 |
||
| 5 | - cleaned up some code which gave warnings. |
||
| 6 | - altered some code to be compatible with Sun CC compiler. |
||
| 7 | However, support in the makefile for non-ansi C compilers |
||
| 8 | has been dropped (this is a stupid thing to support). |
||
| 9 | |||
| 10 | December 20, 1995 to January 9, 1996: |
||
| 11 | |||
| 12 | verified on HHI #22, TCEH #23 bitstreams. |
||
| 13 | |||
| 14 | 1. new arguments format. Got to be so many argument fields that |
||
| 15 | a new more consistent format was devised. |
||
| 16 | |||
| 17 | 2. Frame_Store_Flag (-f) option now controls lower layer prediciton |
||
| 18 | picture format (field or framewise) |
||
| 19 | |||
| 20 | 3. getpic() structural changes |
||
| 21 | |||
| 22 | |||
| 23 | Since December 18, 1995: |
||
| 24 | |||
| 25 | 1. added special case for current B pictures subsframe.c which |
||
| 26 | loads entire reference frame for buffer substitution. |
||
| 27 | |||
| 28 | 2. fixed -l omission (-lXext) in Makefile that drives Tristan nuts everytime. |
||
| 29 | |||
| 30 | |||
| 31 | Since December 14, 1995: |
||
| 32 | |||
| 33 | 1. organized frame buffer substitution routines into subspic.c |
||
| 34 | 2. added "big file" -b mode for Tristan ;-) |
||
| 35 | |||
| 36 | |||
| 37 | Since July 4, 1994: |
||
| 38 | |||
| 39 | 1. Concatenated elementary sequences within same bitstream |
||
| 40 | |||
| 41 | Decode can now handle concatenated elementary video sequences of |
||
| 42 | arbitrary parameters. |
||
| 43 | |||
| 44 | 2. TRACE and VERBOSE #ifdef flags |
||
| 45 | |||
| 46 | 3. More disciplined naming convention |
||
| 47 | |||
| 48 | normative variables and bitstream elements defined in 13818 are |
||
| 49 | verbatim, lower case. Implementation specific routines and variables |
||
| 50 | are capitolized. |
||
| 51 | |||
| 52 | 4. Spatial scalability corrections |
||
| 53 | - see Carsten's document (spatscal.doc) |
||
| 54 | |||
| 55 | 5. D-pictures (picture_coding_type==D_TYPE) |
||
| 56 | |||
| 57 | Only two small changes were necessary to accomodate D-pictures: |
||
| 58 | |||
| 59 | a. in Decode_MPEG1_Intra_Block() added line which termines |
||
| 60 | subroutine after DC coefficient has been processed. |
||
| 61 | |||
| 62 | b. in picture_data(), added line which parses marker bit. |
||
| 63 | |||
| 64 | |||
| 65 | 6. forced decoder to display frame progressively (regardless of whether |
||
| 66 | the picture is frame structured or field structured) when -f flag |
||
| 67 | is invoked in the command line arguements. |
||
| 68 | |||
| 69 | also: progressive_frame now decides whether a frame is to be displayed |
||
| 70 | as a frame picture to two field pictures, rather than the older convention |
||
| 71 | of testing progressive_sequence. |
||
| 72 | |||
| 73 | 7. Adapted systems parser from Stefan's mpeg2play to mpeg2decode. |
||
| 74 | The major changes are: |
||
| 75 | |||
| 76 | mpeg2dec.c: |
||
| 77 | - fseek() called twice |
||
| 78 | |||
| 79 | gethdr.c, getpic.c: |
||
| 80 | instances of Flush_Bits(par,32) changed to Flush_Bits32(par) |
||
| 81 | |||
| 82 | gethdr.c |
||
| 83 | Get_Bits(par,32) changed to Get_32_Bits(par) |
||
| 84 | |||
| 85 | global.h |
||
| 86 | added rdmax, sysstream, and bfr[] to parameters struct. |
||
| 87 | |||
| 88 | 8. Restructuring of getpic.c: |
||
| 89 | |||
| 90 | a. moved picture pointer rotation into Update_Picture_Buffers() |
||
| 91 | b. moved picture output logic into Output_Current_Frame() to |
||
| 92 | in anticipation of 3:2 pulldown |
||
| 93 | |||
| 94 | |||
| 95 |