Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion. Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules. |
17th January 2003, 23:52 | #4 | Link |
Registered User
Join Date: Oct 2001
Location: Gainesville FL USA
Posts: 2,092
|
Marc -
By all means, release it. I actually took another crack at YV12 SimpleResize last night, some weeks after writing it. But it still seems to be terminally busted, so I can't release mine yet. But please don't call it SimpleResize since that would probably come back to haunt us forever. - Tom |
18th January 2003, 12:29 | #5 | Link |
XviD fan
Join Date: Jun 2002
Location: France
Posts: 907
|
no problem, tom, i called it FastBilinearResize.
i've done a bicubic version too, because my project is to make a hyperfast bicublin resampler, IMHO a very good trade-off between speed and quality. so it would be a 3 in 1. i'm gonna release some beta stuff today ^^. |
18th January 2003, 13:28 | #6 | Link | |
Registered User
Join Date: May 2002
Location: Czech rep.
Posts: 390
|
Quote:
Can you implement 2 in 1, please? Crop + FastBicubicResize in one? |
|
18th January 2003, 14:09 | #7 | Link |
Remember Rule One
Join Date: Oct 2001
Location: SWEDEN
Posts: 1,611
|
@MarcFD
Any more thoughts on a resizer that thakes YUY2 input and converts it to YV12 on output, like I suggested in this thread a while ago. The bicublin would be wonderful to have modified in this way. |
18th January 2003, 18:19 | #8 | Link |
XviD fan
Join Date: Jun 2002
Location: France
Posts: 907
|
okay, here's the stuff.
i tested it a lot. i hope you will like it hakko, this bicublin is unfiltered, i'm not sure that using YUY2->YV12 resising will lead to better quality. i'll try anyway (C coding is so fast ^^) (attached "BicublinResize beta 1.zip") |
18th January 2003, 20:49 | #11 | Link | |
Registered User
Join Date: May 2002
Location: Czech rep.
Posts: 390
|
Quote:
One month - one new filter! |
|
18th January 2003, 21:04 | #12 | Link |
XviD fan
Join Date: Jun 2002
Location: France
Posts: 907
|
thx Harry
well, in fact, i've coded much more than that. edeen & atc are fully working & optimised, but i still didn't released them. i'll do that this WE too. they would be my 2 last releases. so, does the stuff work ? i've spend many hours of debugging, i'd like to know how perform the last filter i'll probably write before a long time. |
18th January 2003, 22:56 | #13 | Link |
TotalEclipseOfTheBrain
Join Date: Sep 2002
Posts: 347
|
First tests seem ok. Hard to say about visual quality, I need to test more to say, but seems OK.
About compressibility, BicublinResize with default parameters gave me similar results than bicubic 0.6. "that's one of the last thing i'll do, so i take my time " -> are you too busy (your "bac") or what else ? Cheers, FuPP. |
19th January 2003, 11:21 | #16 | Link |
XviD fan
Join Date: Jun 2002
Location: France
Posts: 907
|
>"that's one of the last thing i'll do, so i take my time " -> are you >too busy (your "bac") or what else ?
no, i'm just tired of filter devel, i programmed a lot of stuff and i simply don't need anything more ^_^ >could you make the filter raise an error to the user if not fed a YV12 >source? yep, i'll do it, but it's only a beta, i want things to work first, i play after ^^ |
19th January 2003, 11:42 | #17 | Link |
Registered User
Join Date: Jun 2002
Posts: 250
|
Really nice filters.. I don't see any difference ( but didn't tested a lot ) between FastBilinear or Bilinear..
But FastBilinear is really faster.. ------- But there's a little bug with you filters. I saw this with "FastBicubic" & "Bicublin" ( not with "FastBilinear" ) When I try this script ( source = DivX 528@288 : Code:
AviSource("*.avi", false) #Crop(4,30,-4,-30) FastBicubicResize(640,288,1/3,1/3) #BicublinResize(640,288) If I un-comment the crop() command of the script, the line turn Green ( http://seewen.levillage.org/download...cubic_crop.jpg ) With this script + "Bicublin", the line turn White ( difficult to see, I didn't take picture ). ----------- But what is more strange, is that if I open this script ( crop() un-commented ), the Green line appear. Then if I comment the "FastBicubic" line ( with VdubMod editor ) and un-comment "Bicublin" the line turn white. But then, if I comment "Bicublin" and un-comment "FastBicubic", the green line doesn't appear... And if I comment "FastBicubic" and un-comment "Bicublin", the white line doesn't appear too ! Then if I just close the script. and re-open it, the line doesn't appear.. But if I close VdubMod and Re-open it, the green line come back.. And changing crop values changes color, or make line disapear ( like #Crop(2,30,-4,-30) or #Crop(4,30,-2,-30) for example ). Sorry, I cannot be more precise, as this problem is so special.. A+ P.S. Using avisynth 2.50 ( 10.1.2003 ) and the last VdubMod and Vdub ( bug appear in both ). |
19th January 2003, 13:03 | #18 | Link |
XviD fan
Join Date: Jun 2002
Location: France
Posts: 907
|
no, no, it's a normal problem.
it took my several hours to tweak the boders, my best advice is to oversize and to crop after resize. it's so fast anyway, and the problem only occurs with very specific input/output sizes |
19th January 2003, 18:57 | #20 | Link | |
what's this for?
Join Date: Apr 2002
Location: Berlin, Germany
Posts: 993
|
Hi Marc FD,
your filter is really amazing fast!! [DickModeOn] I get nearly 44fps in first pass and almost 51fps in second pass... (XP1600, 256 RAM, single 60GB_HD without raid, WinXP-SP1) [/DickModeOff] I've used the following simple script: Code:
LoadPlugin("D:\Tools\AVISynth_Plugins\Avisynth_2.5\MPEG2Dec3.dll") LoadPlugin("C:\Tools\AVISynth_Plugins\Avisynth_2.5\BicublinResize.dll") mpeg2source("D:\Movies\the 51st state\51.d2v") crop(2,8,716,553) BicublinResize(544,288,0.75,0.5) Quote:
I did test relatively a lot but the standard parameters Code:
BicublinResize(544,288,1/3,1/3) Code:
BicublinResize(544,288) my 2 cents and with kind regards - Sill
__________________
latest mix sets: http://soundcloud.com/sill |
|
Thread Tools | Search this Thread |
Display Modes | |
|
|