Et voilà, la nouvelle version est sorti. les nouveautés sont dessous mais en anglais, lol! Si quelqu'un pouvait m'aider à comprendre les nouveautés, se serait sympa ??
Quelqu'un l'a déjà tester ??
4/10/2002: VirtualDub 1.4.10 released
I was thinking of an April 1st release, but I figured you'd enjoy this more. The 1.4.10 release of VirtualDub is the "third-party release," because it mainly consists of the following changes:
Support for DivX Pro 5.0.1 B-frame encoding
I promised DivXNetworks that I would fast-track a new release with 5.0.1 B-frame support; well, this is fast for me. In exchange, I asked (demanded) that they fix some of the glaring bugs in their codec that you users have been complaining to me about; it looks like the bugs have indeed been fixed, including the black screen bug, the 24-bit scanline alignment bug, the output bit depth bug, and the ICCompressQuery() bug. Woohoo! B-frame support is somewhat of a hack in VirtualDub and involves advancing the video stream at the request of the codec; the audio interleaving will likely be a little screwy for the first sixteen frames or so. Also, the support won't work in capture mode, but I'll try to fix that later (you probably won't want to try B-frames on the fly anyway).
The XviD team is also working on similar support in their codec, but sadly, they did not provide me with a working implementation in time. Those of you with compilers can attempt to enable the code in VideoSourceCompressor.cpp once the new XviD codec is ready.
If any other codec makers are experiencing issues with VirtualDub, please don't hesitate to drop me an email so we can work out the compatibility problems.
Fix for incompatibility with files from Video Live
At the request of the author of Video Live, I've fixed a bug in VirtualDub that caused "can't initialize timer" errors when attempting to process files from that application. The cause was a bunch of signed MulDiv() calls on dwRate/dwScale fractions. Oops.
Fix for crash when producing large files from Avisynth
Two bugs here: VirtualDub was passing through the dwSampleSize value from Avisynth even when it didn't match the frames generated after compression and/or filtering, and the AVI2 output handler was crashing when the output file exceeded 2GB and no audio stream was being processed. It took me a long time to figure out what was going on here, and this bug may have been responsible for some applications rejecting output from VirtualDub.
All new disassembler
This one's for me as well as developers of third-party codecs who may get VirtualDub dumps from users -- I was tired of receiving bugs with SSE2 instructions that were totally unreadable. The new disassembler is data driven and can disassemble Pentium 4 and Athlon instructions, as well as perform symbolic lookups on some arguments. I was tempted to use bison to create the disassembler, but that would probably have been too much. This change also means that VirtualDub has a new debug file format; those recompiling will need to download the 1.1 release of the VirtualDub build tools in order to generate the new format.
Hint: "Privilege Violation" errors from SSE code mean you are attempting to access unaligned data (#GP fault).
I'm curious what experiences others have had writing x86 disassemblers; personally, I've found it an absolute pain in the (null). 3DNow! instructions are probably the most annoying, since you can't determine the instruction until after you have parsed the mod-r/m bytes. Also, it appears that the Pentium III throws #UD if you use more than ten lock prefixes in a single instruction.
Now leave me alone so I can finish V1.5!!! ^_^;
Build 13870 (Version 1.4.10):
[features added]
* Capture mode can now apply filters against I420/IYUV
video.
* Capture: "Start on click" added.
* Capture: File autoincrement added.
* Crash disassembler has now been totally rewritten;
now supports all IA-32 instructions through Pentium 4
and Athlon.
* Special support added for DivX 5.x B-frame mode.
NOTE: Interleaving may be a little funny.
[bugs fixed]
* Volume adjustments were active during input preview.
* dwSampleSize was being copied from source video
streams even if it was incorrect.
* AVI2 output code was crashing if the source video
stream dwSampleSize!=0 and no audio was attached.
* Confirmation dialog for "Reset video stream" didn't
have a cancel button. ^^;;
* Several places were using signed division for frame
rate fractions -- fixes "can't initialize timer"
errors.
---------------
L'informatique, c'est la science du peut être ! | www.planeterenault.com | http://forum.planeterenault.com
