Ok, I did a lot of testing with different VGA cards, and different SoftDVD programs...here's what I found […]
Show full quote
Ok, I did a lot of testing with different VGA cards, and different SoftDVD programs...here's what I found
ATI-Rage128Pro support (iDCT + HWMC + subpic-alpha)
1) Cinemaster 1999 or later (including ATI DVD 3.1 or later)
(VideoDecoder.PerformanceClass = 0x04)
http://www.ravisent.com
2) WinDVD 2.2 or later
**does not de-bob interlaced video correctly, displays
half-frame only
http://www.intervideoinc.com
3) MGI SoftDVD MAX 4.0
**does not de-bob interlaced video correctly, displays
half-frame only
http://www.mgisoft.com
4) Mediamatics DVD Express 5.0.94
(*OEM only, IBM Thinkpad)
(must manually enable 'MCAM')
All above players, except MGI, also support RageXL/Mobility iDCT. I just got a RadeonLE, so I'll see which of these players support RadeonLE in iDCT mode. Obviously, Cinemaster2000 supports Radeon. Medimatics and WinDVD probably also support Radeon.
---------------
The following players support ATI-Rage128Pro in 'MC12' mode (HWMC only...pseudo-compatible with ATI Rage Pro.)
1) Cinemaster 1999 or later (couldn't test earlier version)
Through registry hacking, user can force 'MC12'
(VideoDecoder.PerformanceClass = 0x03)
http://www.ravisent.com
2) WinDVD 2.2 or later (couldn't test earlier version)
***not directly...through 'bug' on my system
***WinDVD will automatically attempt to 'MCAM' (Rage128)
3) Mediamatics DVD Express 5.0.94 (couldn't test earlier)
(must manually enable 'MC12')
4) PowerDVD 3.00 (1016)
**does not de-bob interlaced video correctly, displays
half-frame only
...buggy MC12 support, motion-comp artifacts
5) **PowerDVD 2.55 0113 or 0202
**does not de-bob interlaced video correctly, displays
half-frame only
...buggy MC12 support, hangs on most interlaced videos
Build 0620 does not support Rage128-Pro.
This leads me to believe, PowerDVD 'reverse engineered' ATI's HWMC interface.
...
I'm doing a 'low-end PC-DVD roundup.' As soon as I'm done with testing, I'll post some results...
I tested these cards :
(1) S3 Savage3D 8MB AGP SGRAM (64-bit mem)
(2) S3 Savage4 16MB AGP SDRAM (64-bit mem)
(3) SiS 305 16MB AGP SDRAM (32-bit mem interface, yuck)
(4) ATI Rage Pro 8MB AGP SDRAM (64-bit mem)
(5) ATI Rage Pro Turbo 8MB AGP SGRAM (64-bit mem)
(6) ATI RageXL 8MB AGP SDRAM (64-bit mem)
(7) ATI Rage128 Pro 8MB AGP SDRAM (64-bit mem)
(8) Trident 'Blade3D' 9880 8MB AGP SDRAM (64-bit mem)
(9) NVidia Geforce2/MX 32MB AGP SDRAM (128-bit SDR mem)
(10) ATI Radeon LE 32MB DDR SDRAM (128-bit DDR)
Each card with these programs :
(1) PowerDVD 2.55
(2) PowerDVD 3.0
(3) Cinemaster2000 Build 6605 (ElsaDVD)
(4) WinDVD 2.3 trial
(5) MGI SoftDVD Max 4.0.28
As you can see, that's 50 combinations, and it's taking a while...
So far, of the tested cards, Rage128-Pro has most performace-acceleration. S3 Savage3D/4 and Rage128/Pro are neck to neck in image-quality. (Savage has better downscaling below shrink factor of 0.5. Rage128 has better upscaling above zoom factor of 1.0.)
Strangely enough, Geforce2/MX has worst performance-acceleration, especially with PowerDVD.
Trident Blade3D has worst image quality (no subpic-alpha, and mediocre overlay-zoom quality.)
Rage-Pro and RageXL's overlay controller require minimum X-stretch factor of 2.0X (or drops adjacent pixels.) Probably due to RAMDAC and/or memory bandwidth constraint.
RageXL, despite iDCT support, is marginally faster than the other HWMC cards.
S3 Savage3D and NVidia Geforce2/MX both 'cheat' when constructing Directdraw overlays -- this means both devices allocate excessive video-memory. The Savage3D allocates exactly 2.0X 'expected amount' (16bpp 720x480 YUV overlay = 675KB. Savage3D allocates 1350KB.) Geforce2/MX allocates between 2-3X 'expected amount.'
SiS DVD-overlay performs hardware subpicture-alpha blending through overlay, but the subpicture overlay is filtered poorly (zoom/shrink looks worse than background DVD-video.)
S3 Savage3D/4, NVidia Geforce2/MX *probably* do NOT have hardware overlay blending. Instead, the drivers use BITBLT engine to combine subpicture data with decoded video-frame. Savage3D/4 both have 'perfect' overlay scaling (i.e. seamless blending with background, at all zoom/shrink factors.) Nvidia Geforce's subpicture-blending shows evidence of artifical sharpen-filter (edges of subtitle text are brighter than surrounding pixels?!?)
ATI Radeon uses 3D-core to perform BITBLT stretch/shrink. (This is obvious when playing AVI/MPEG at high-resolution, the screen wipes along a diagonal line. The diagonal line divides the upper screen exactly into 1 triangle.) Radeon's stretch-blt quality is *inferior* to Rage128, Savage3D.