Suggested scripts

This forum is for questions and discussion of all the aspects of handling your footage. If you have questions about capturing/ripping footage, AviSynth, or compression/encoding/converting, look here.

Postby batmanMG » Tue Jan 23, 2007 8:10 pm

i need new eyes... or a better attunement to detail... i can't tell the diff between the ones above... but thats not the point

been working on a place promised in our early days video for 2 years now and am a cluts with the scripts... anyone know the best quality to get...? i've tried alot but much of it blurs up the backgrounds too much wich are some of the most beautiful parts
William Blake says "Babies are evil fiends hidden in clouds." Do you really want a baby hater to be in the chatholic church? Vote and excommunicate William Bake.
the confused have hearts too wonder where this came from.
User avatar
batmanMG
 
Joined: 28 Jan 2005

Postby Krisqo » Tue Jan 23, 2007 8:16 pm

Go gentle on the scripting for Place Promised. I only used MFToon and a light deen filter.
Image
User avatar
Krisqo
Cooking Oil
 
Joined: 07 Jul 2003
Location: Moderating the Adobe Forums
Status: W.O.A (Waiting on Aion)

Postby Autraya » Tue Jan 23, 2007 10:33 pm

batmanMG wrote:i need new eyes... or a better attunement to detail... i can't tell the diff between the ones above... but thats not the point

yeah sorry probably a bad example, and originally I didn't use fastlinedarken on the TV series like in the shots above , I was just trying to show the difference between warpsharp/fastlinedarken in a different order.
new banzors in the making :p
User avatar
Autraya
Zero Punctuation
 
Joined: 11 Mar 2003
Location: Terra Australis
Status: old

Postby Sereenie » Wed Jan 24, 2007 12:33 am

I'll jump in because I too need a new set of eyes and some advice. I've been fiddling with this since Friday and I seem to be going nowhere fast.

Well, that's not absolutely true. I'm... somewhat happy with the current results, considering the source. The episodes are coming out great, but the OPs look like crap. Here is where I stand now (take note that the before shots all have cpu=6, which is why they don't look quite *that* bad).

Example 1
Example 2
Example 3
Example 4

I went a bit overboard with the saturation, but considering how washed out the source is, I can't say it really hurts. There also seems to be slight (green? yellow?) tinge, especially visible in the moon and in various whites (the castle at the beginning especially, but I don,t think I have a screenshot for that), but I can't manage to get rid of it. I tried coloryuv(gain) for both the u and the v, individually and together, up to 15 each, but it didn't seem to solve the problem. Plus, it made my blacks non-black. I tried compensating by playing with brightness and contrast, but that didn't work either.

There's also sharpening artefacts, but taking out the sharpening makes everything look soft. And I can't quite see myself smoothing things even more. I already think I may have overdone it a bit, though it mostly looks nice.

But my biggest problem is that the OPs seem to have been interlaced with a different method than the episodes, because no matter what I try, I always end up with either some interlacing left or with blends, which doesn't happen anywhere else (well, maybe in the EDs, but I haven't ripped those).

See how crappy it looks!

Actually, that one almost looks nice. I've tried various methods, and while IVTC and such gave me somewhat decent results, there were still interlacing remnants and blending left over here and there. Sangnom+Unblend at least smooth the blending so it fits in better with the rest of the footage at regular speed. Still, I'd rather have no blending at al. Plus, the script is now five times slower than it was when I was just IVTCing.

Here's my current script

mpeg2source("path",ipp=true,cpu=6)
SangNom(order=1,aa=32)
Unblend()
Decimate(cycle=5, mode=2)
Crop(8, 0, -8, -0)
degrainmedian().
tweak(bright=-20,sat=1.6,cont=1.2)
fft3dfilter(sigma=5,plane=4,bw=48,bh=48,bt=4,ow=24,oh=24)
mftoonlite()
LimitedSharpen(ss_x=2.0, ss_y=2.0,Smode=3)
Lanczosresize(640,480)


I realise I can resize straight in Limitedsharpen, but I don't know if that would help me gain speed. Oh, and I used Degrainmedian because fft3d3's author recommended to do so (yes! I RTFM!), so I gave it a try. I stuck the tweaks in-between thinking it couldn't hurt to re-smooth after, but I'm not too sure about that anymore. I don't need a Beforeafter, I need an Afterafter to compare the results of various combinations and help me make up my mind! ^_^;

So, does anyone have any recommendations for me? Has any one else tackled successfully with uncut season 1 of Sailor Moon? Any help will be much appreciated, because I'm running out of ideas!

Thanks!

S.
User avatar
Sereenie
 
Joined: 28 Jun 2003
Location: The City Fun Forgot

Postby jasper-isis » Wed Jan 24, 2007 12:04 pm

I think two of the reason that the AMVWiki compilation didn't work out were (1) the focus on postprocessing and (2) the fact that each script went by its submitter's name, thus making it awkward for other people to come in take full advantage of the wiki by adding their own improvements. Postprocessing scripts are better passed along on a case-by-case basis, with the more advanced avisynth user and the person seeking advice having a conversation.

We would probably be better off with a compilation of simply deinterlacing scripts. There's less disagreement about what makes good deinterlacing vs. what makes good image quality, and properly restored footage is always a good thing. We could make a listing of shows (noted with NTSC or PAL), and under each would be a user-submitted script (or two). Scripts can gain reliability by being checked and affirmed by other people (this can be denoted by *'s or something).

Yes, we'll still have the people-blindly-copying-scripts-without-knowing-wtf-they're-doing problem, but they'll at least be copying "harmless" stuff that people generally agree upon.



Krisqo wrote:Go gentle on the scripting for Place Promised. I only used MFToon and a light deen filter.

I don't know if I would consider mftoon "gentle" by any means... :sweat:


trythil wrote:Stuff like
Code: Select all
crop(8,0,-8,0)
lanczosresize(720,480)

illustrates one of my problems with it: people contributing scripts without noting exactly what they were doing. (Yes, I was guilty of this too.)
My objection to that script segment is that it screws with the aspect ratio by clipping off the side black bars and horizontally stretching the image to fit the original image size.


In defense of that...

Some people like to keep a 720x480 lossless copy of their video (deliberately in the wrong AR) for future MPEG-2 mastering purposes, etc. While I now agree that there are better ways of fixing blurry black edges, I've never actually noticed the upsizing (after cropping) degrading image quality in any way. The video, of course, would be resized to the correct AR for internet distribution.
Image
User avatar
jasper-isis
P. Y. T.
 
Joined: 13 Aug 2002
Status: catching all the lights

Postby MousePotato » Wed Jan 24, 2007 2:54 pm

rotoscoping > filters!
User avatar
MousePotato
Boochsack whore
 
Joined: 14 May 2004
Location: Highlands Ranch, CO

Postby Sereenie » Wed Jan 24, 2007 3:14 pm

Well, sure! But if there is anything else worth trying first, I'd rather give that a shot!

S.
User avatar
Sereenie
 
Joined: 28 Jun 2003
Location: The City Fun Forgot

Postby Krisqo » Wed Jan 24, 2007 5:59 pm

jasper-isis wrote:
Krisqo wrote:Go gentle on the scripting for Place Promised. I only used MFToon and a light deen filter.

I don't know if I would consider mftoon "gentle" by any means... :sweat:



I meant that only two filters were needed. MFToon did most of the hard work. It's gentle compared to some I've used.

For example (yes... you may use this without credit) My script for Cardcaptor Sakura DVD1

Code: Select all
LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\TIVTC.dll")
mpeg2source("C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v",cpu=4,upconv=true)
TFM(d2v="C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v", PP=0)
Tcomb()
TDecimate(mode=1)
ConvertToYV12()
deen("w2d",7,5,9)
WarpSharp()
FastLineDarken()
SDDerainbow()
tweak(sat=1.4, cont=1.0, bright=-10)
ColorYUV(gain_u=1)
ConvertToRGB32()
Crop(8,60,-8,-60)
BiCubicResize(640,360)
AssumeFPS(24)


I had to do the colorspace change because I couldn't filter anything without using the ConvertToYV12(). Also, the PAR is set to so not to altar the PAR of the two movies which I was using for my video as well.
Image
User avatar
Krisqo
Cooking Oil
 
Joined: 07 Jul 2003
Location: Moderating the Adobe Forums
Status: W.O.A (Waiting on Aion)

Postby kikai_saigono » Wed Jan 24, 2007 6:27 pm

Krisqo wrote:


For example (yes... you may use this without credit) My script for Cardcaptor Sakura DVD1


*strangles*
Youtuber wrote:What is an MEP exactly? I looked it up but I don't think it has anything to do with being a Member of the European Parliament.
User avatar
kikai_saigono
 
Joined: 09 Jul 2005
Status: jailbait D:

Postby Krisqo » Wed Jan 24, 2007 6:46 pm

XD

:P
Image
User avatar
Krisqo
Cooking Oil
 
Joined: 07 Jul 2003
Location: Moderating the Adobe Forums
Status: W.O.A (Waiting on Aion)

Postby MousePotato » Thu Jan 25, 2007 11:11 am

has anyone got any tips for scripting Samurai Champloo? So far this is what im using:

LanczosResize(960,480)
Crop(120,0,-120,-0)
assumefps(24)

it seems pretty clean by it's self but let me know if you've found ways to clean it up. (if it really even needs it at all)

thx
User avatar
MousePotato
Boochsack whore
 
Joined: 14 May 2004
Location: Highlands Ranch, CO

Postby Keeper of Hellfire » Thu Jan 25, 2007 12:06 pm

Krisqo wrote:
Code: Select all
...
mpeg2source("C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v",cpu=4,upconv=true)
TFM(d2v="C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v", PP=0)
Tcomb()
TDecimate(mode=1)
ConvertToYV12()
.....


I had to do the colorspace change because I couldn't filter anything without using the ConvertToYV12().
Looks like you're not knowing what you're doing. If you remove the "upconv=true" from mpeg2source(...), you don't need the ConvertToYV12(), because your source is in YV12 and gets converted to YUY2 by this parameter. So you're doing a conversion YV12->YUY2->YV12.

And if you remove the "PP=0" from the TFM(...) command after it has fixed your DV2, you don't need the TComb(), because TFM perfoms the deinterlacing.

I'm not familiar with all of the filters that you are using, but probably you're doing some more nonsense.
User avatar
Keeper of Hellfire
 
Joined: 09 Jan 2005
Location: Germany

Postby Keeper of Hellfire » Thu Jan 25, 2007 1:19 pm

Keeper of Hellfire wrote:And if you remove the "PP=0" from the TFM(...) command after it has fixed your DV2, you don't need the TComb(), because TFM perfoms the deinterlacing.
Oops. My fault. Judging from it's name I was assuming it's an deinterlacer, but it isn't. It's Dedot and DeRainbow. So when are you doing the deinterlacing at all? Probably a proper IVTC shortens your filter chain. And as I understand it, the TComb() should be in front of TFM(...).
User avatar
Keeper of Hellfire
 
Joined: 09 Jan 2005
Location: Germany

Postby batmanMG » Thu Jan 25, 2007 4:06 pm

well i've reformatted my hard drive and am back to square one with my place promised source... sadly i can't really remember what im doing (its been over 2 years since i started this project and i've only used virtual dub and dvd decripter twice) oh yeah the point

last time i went through all the steps and everything and came out with a 30 gb (well around there) file for my footage... this time however i've been reading A&E's guides like mad but keep ending up a quarter way through the final save with a current file size of 18 gb (unless im reading that huge kb number wrong)... have i done something wrong? or is there a part of the script im missing? also becuase im trying to get the best i can out of this... any AS tips on getting better quality and smaller file size? huffyuv is the way to go right?
William Blake says "Babies are evil fiends hidden in clouds." Do you really want a baby hater to be in the chatholic church? Vote and excommunicate William Bake.
the confused have hearts too wonder where this came from.
User avatar
batmanMG
 
Joined: 28 Jan 2005

Postby Krisqo » Fri Jan 26, 2007 3:01 am

Keeper of Hellfire wrote:
Krisqo wrote:
Code: Select all
...
mpeg2source("C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v",cpu=4,upconv=true)
TFM(d2v="C:\AMV_SOURCES\CARDCAPTOR1\sakura1-FIXED.d2v", PP=0)
Tcomb()
TDecimate(mode=1)
ConvertToYV12()
.....


I had to do the colorspace change because I couldn't filter anything without using the ConvertToYV12().
Looks like you're not knowing what you're doing. If you remove the "upconv=true" from mpeg2source(...), you don't need the ConvertToYV12(), because your source is in YV12 and gets converted to YUY2 by this parameter. So you're doing a conversion YV12->YUY2->YV12.

And if you remove the "PP=0" from the TFM(...) command after it has fixed your DV2, you don't need the TComb(), because TFM perfoms the deinterlacing.

I'm not familiar with all of the filters that you are using, but probably you're doing some more nonsense.


The problem is that the script would not load without the Upconv=true. I know that does a colorspace change. I'm not THAT dumb. But I could not use anything like Deen without adding ConvertToYV12. I don't understand it, so I do what the damn thing says. I'm still a noob at this stuff.

I never understood how the order of the filters affects the footage except that you should filter noise before smoothing or you will be smoothing the noise.

TFM does the interlacing and TDecimate does what it says.

As for the rest... It's not nonsense. I came up with a decent script to fix up CCS. I don't have a great before after shot, but here is one that shows some of the changes for comparison purposes.

Image
Image
User avatar
Krisqo
Cooking Oil
 
Joined: 07 Jul 2003
Location: Moderating the Adobe Forums
Status: W.O.A (Waiting on Aion)

PreviousNext

Return to Footage Help

Who is online

Users browsing this forum: No registered users and 0 guests