possible to frameserve from Virtualdubmod to HCenc?

[manono]

I researched but couldn't figure what the last 4 entries in BicubicResize do (specifically "0,0,704, 288").

It's a way to combine the crop with the resize. No cropping is done there so those last 4 numbers can be omitted.

When encoding, when does it really matter to set HCenc to 16:9 or 4:3?

Yes, a lot. That script is for a 4:3 DVD. If you encode it as 16:9 the resulting aspect ratio will be badly messed up. And I wouldn't use that script even if I were encoding for 4:3. You've allowed for 2 blocks overscan (which is default, for some reason) which I don't do. If encoding for 16:9 I'd do it like this, which is a variation of the script FitCD gives for 16:9 encoding and no addition of overscan bars:


LanczosResize(720,360)
AddBorders(0,60,0,60)

[manono]

I was stating I still couldn't detect interlacing/blending for myself, and the questions that followed were so I could get a mental checklist to deduce it when I can't see the difference.

You have to look in the right places. Your first sample was useless. Even the second one wasn't all that good. but I'm enclosing a bobbed pic with the blending showing. It's near the end, bobbed frame 1110. Also, it would help to get rid of the crushed blacks and brighten it up a bit, as well as crop away the garbage. Here's the script I used to find that blend, although with a better lit scene with the right kind of movement it should be easy:

Yadif(Order=1,Mode=1)
Crop(6,0,-14,-12)
Lanczos4Resize(720,480)
Tweak(Sat=0,Bright=10,Cont=1.2,Coring=False)
YLevelsS(0,2.0,255,0,255)#an AviSynth Function

[spiritgumm]

Just a couple last-minute questions about these two videos.

Yadif(Order=1,Mode=1)
Crop(6,0,-14,-12)
Lanczos4Resize(720,480)
Tweak(Sat=0,Bright=10,Cont=1.2,Coring=False)
YLevelsS(0,2.0,255,0,255)#an AviSynth Function
SRestore(frate=23.976)

I did a sample of the "Bus stop" script (and your brightness settings are more apt than mine).
After cropping, there's no added borders, so there seems to be some image loss at edges when I compared the script video vs. original video. Is there anything wrong with adding in borders to replace cropped edges?
I included the SRestore(frate=23.976) from your previous script. If that is doing the unblending, what does Yadif do? I see it's a deinterlacer, but it's set here only to re-order the fields top first, right? Is this necessary for unblending or another reason?

Speaking of top fields first, my PAL widescreen download has bottom field first, according to Info(). Does that need to be addressed? I burned a sample and it looked fine. As you wrote, the encoder aspect ratio does make a difference (depending on dvd player settings). Not sure why my previous resized attempt didn't seem effected by dvd player output (doesn't matter anyway).

For my future reference, are the FitCd settings you have in the FitCd jpeg for this situation (an unusual widescreen ratio), and Output "anamorphic" should always be checked on?
I was curious if Fitcd would register an anamorphic source as such. I opened an anamorphic source but the only box that got checked automatically was "ITU-BT.6014" on the Input side. Is "anamorphic" something you have to eyeball rather than rely on software detection?

[manono]

Is there anything wrong with adding in borders to replace cropped edges?

Do it any way you like. It's your video. Me, I prefer to have as few borders as possible, and the dimensions of the video after cropping determines if there will be any and how big they are. Maybe it should have been cropped and resized:

Crop(6,0,-12,-12)
Lanczos4Resize(720,480)

If the crop values are the same all the way through the video, that's pretty accurate. But FitCD can help with adding borders, if that's what you want, when you add in 1 or 2 blocks overscan.

...but it's set here only to re-order the fields top first, right? Is this necessary for unblending or another reason?

No reordering. The sample was TFF. Maybe the whole thing is BFF, but I can only go on the sample you provide. You set the field order to what it is. What does Yadif do? It bobs the video, a necessary step before the unblending.

Speaking of top fields first, my PAL widescreen download has bottom field first, according to Info()

What widescreen download? The 704x288 AVI? It's progressive. It has no field order. AviSynth assumes BFF in the absence of anything definite.

For my future reference, are the FitCd settings you have in the FitCd jpeg for this situation (an unusual widescreen ratio), and Output "anamorphic" should always be checked on?

You check the anamorphic box when you intend to encode for 16:9. You wouldn't check it, for example, if using FitCD to generate a script for Bus Stop. That's a 4:3 movie. Or for a 640x480 AVI.

Is "anamorphic" something you have to eyeball rather than rely on software detection?

If it's an MPEG source you can find out which it is by running the preview in DGIndex. If it's an AVI source, choose however you want to encode. But it wouldn't be too bright to encode a 1.33:1 AVI as 16:9. The only time you have to decide is when it's widescreen, and then you can decide if you want widescreen 4:3 or 16:9.

[spiritgumm]

Well, some good/bad news about these two videos.
Inre to widescreen 704x288 AVI, video was fine, but I couldn't get Besweet to convert mp3-ac3 directly or in separate steps(mp3-wav-ac3), so I used Belight in steps. While that effected a framerate change, it didn't change the pitch. Didn't notice until after muxing & burning dvd. I corrected by converting mp3-wav-ac3, then used Besweet to do the framerate change ac3-wav, then wav-ac3 with Aftengui (Aftengui doesn't change volume level).
My question about this: rather than go thru all the above steps, how can I use Avisynth (perhaps via Virtualdub or Winff?) to output a converted framerate audio? It definitely produces a good pitch-altered audio when I've previewed with AssumeFPS(23.976, true).

The Bus stop video, while improved, has a blinking effect which I don't think is on original. Black and grey alternate in the top quarter of the image (actually the top half, but really noticeable in top quarter). Very distracting on TV, but only faintly noticeable on a monitor so I don't know if you'll see what I mean in attached sample. Rather than repeatedly tweak the script and burn samples to get an idea what's happening, I thought I'd ask you first about what might be the cause & cure.
busstop blink.m2v

[manono]

I don't handle audio in scripts, so maybe someone else can help.

The blinking is in the source, not the field blended conversion, but whatever 16mm film source or whatever they used to create the master for the crummy broadcast or whatever it's from. I don't know of any way to get rid of that blinking. Maybe there are ways, but I don't know them because I haven't had to do anything like what needs to be done before. That source is pretty bad, as you well know. If it bothers you you could always just reencode it as interlaced, the way you got it.

Sorry, not much help this time.

[spiritgumm]

Inre to "Bus stop" I omitted the Yadif/SRestore from the script, and while the blinking is gone, there are the horizontal lines. Tried some other deinterlacers just to see what happens, but nothing gets rid of the lines. Perhaps I need a less effective deblender - that is, some combination that gives a compromise so the black-grays don't seem to be alternating and the horizontal lines are blurred?
Would changing the black level lessen the grays?
Just wondering, can one deduce if "Bus Stop" was made using a dvd recorder or a PC (assuming it was connected to a vcr)? I was looking at frames from a dvd made in my Panny recorder (from vcr source) and they was pretty blurry.

Sorry, not much help this time.

I'm pretty happy with your help so far!

[poisondeathray]

I find it hard to believe that omitting yadif/srestore would eliminate the "blinking". The "blinking" is caused by luminance changes. There must be some other explanation, like using an "auto" filter like auto levels. I'm inclined to believe it was in the source, and whatever levels changes you made just made it more obvious

Can you post the an original sample corresponding to that section , and the script you used

[spiritgumm]

Original video (not quite identical timeframe):
Busstop original.m2v

The script below (after running thru HCenc with 3:2pulldown) is what I used. The sample in previous post was from that video (blinking background).

Yadif(Order=1,Mode=1)
crop(0,0,0,-12)
addborders(0,0,0,12)
Lanczos4Resize(720,480)
Tweak(Sat=0,Bright=10,Cont=1.2,Coring=False)
YLevelsS(0,2.0,255,0,255)
SRestore(Frate=23.976)

Then I eliminated Yadif and SRestore, so only brightening/cropping. I thought that's what manono meant by encoding it as interlaced (the way I got it), and I wanted to see what it would look like. Actually I think I encoded that with 3:2 Pulldown - would that have any bearing on this?

[poisondeathray]

As manono said above, the flickering is in the source. You can see this for yourself, use assumetff().separatefields() and you will see each 2nd field has different luminance, that explains why the flickering occurs primarily at the top of the bob-deinterlaced script

If you single rate deinterlace yadif(mode=0, order=1), you will notice, the flickering improved , but that's because you throw out every 2nd field. You can't use a restore function by single rate deinterlacing, and will have to accept the blends then

You have to make some decisions / trade offs to which you think is more acceptable. IMO, the flickering is less acceptable (and difficult to fix), and the blends aren't as bad (at least in this section).

You might have other sections that need to be treated differently. In that case, you could apply different filters to different sections

[spiritgumm]

Hi poisondeathray,
earlier I experimented with the Yadif order but kept Srestore in. The video had a sort-of slow motion quality.

At your suggestion (I think you were suggesting this) I omitted SRestore and changing Yadif order. The video became 59.94 fps and not dvd compliant. Was I suppose to substitute something for SRestore?

I wonder what would cause this alternating field luminance? And if there's a way to change luma for every other frame? Seems like something Avisynth could be set to do.

[poisondeathray]

earlier I experimented with the Yadif order but kept Srestore in. The video looked like slow motion.

do you mean the mode, not order? mode=0 is single rate, so it should be 1/2 the motion samples. wrong field order would give you jittery forward and backward. The field order should always be kept constant regardless of how you choose to approach this

At your suggestion (I think you were suggesting this) I omitted SRestore and changing Yadif order. The video became 59.94 fps and not dvd compliant. Was I suppose to substitute something for SRestore?

I think you mean the mode again, because the order refers to the field order (1=TFF or 0=BFF), the mode refers to bob or single rate (0=single, 1=bob)

1) If you bob deinterlace to 59.94, you will see the blinking and this will still occur after using any deblender such as srestore because all of them require bob deinterlacing first, is that what you want? Using the restore functions will return to 23.976 film rate and you could encode as progressive

2) If you don't and only single rate to 30p you get 1/2 the motion samples, but less blinking and keep blends - is that what you want? 30p is compliant for DVD

3) The 3rd option is to bob-deinterlace, then do your filters (i.e color correction etc..), then re-interlace and encode as interlaced, but you should would see the blinking still (and blends) - I see no benefit in this, except you have smoother motion (keep 60 fields per second)

I'm not suggesting anything. These are the trade offs you have to make. IMO there is not good way to approach this, each has significant problems. Also if other sections are different (in regards to blinking etc...), this might not even apply or you might apply filters in sections

You could also try some of the deflicker plugins like deflicker in avisynth , or deflick in vdub (msu plugin) - these try to control and even out the luminance changes , but will not work very well in your case, because it's only the top half that flickers (from 1/2 the fields being affected)

[spiritgumm]

I meant "mode" in both usages.

I thought you might have been making a suggestion here inre Yadif and Srestore -

If you single rate deinterlace yadif(mode=0, order=1), you will notice, the flickering improved , but that's because you throw out every 2nd field. You can't use a restore function by single rate deinterlacing, and will have to accept the blends then

As for what I want, I'd like to minimize defects. I can't be more specific since don't know the abilities and limits of the filters. One thing that's becoming clear, a crappy video is a good learning experience.

[poisondeathray]

I thought you might have been making a suggestion here inre Yadif and Srestore -

If you single rate deinterlace yadif(mode=0, order=1), you will notice, the flickering improved , but that's because you throw out every 2nd field. You can't use a restore function by single rate deinterlacing, and will have to accept the blends then

As for what I want, I'd like to minimize defects. I can't be more specific since don't know the abilities and limits of the filters. One thing that's becoming clear, a crappy video is a good learning experience.

But you can't have your cake and eat it too! There are only rocks and hard places here. You have to make some tradeoffs

You can't single rate deinterlace and use a deblending function (deblending functions require bob-deinterlacing first) - so you're stuck with "flickering" if you choose to use a deblending function

If that sample is representative of the whole movie, I would say the flickering is the worst artifact (most annoying), and I could live with blends

[manono]

If that sample is representative of the whole movie, I would say the flickering is the worst artifact (most annoying), and I could live with blends

Although I hate to say it, because I spend a good part of my time removing blends from videos, I think I agree - in this case it might be better to live with the blending. The quality of the video is so horrible to begin with that the blending isn't even the worst part about it.

Actually I think I encoded that with 3:2 Pulldown - would that have any bearing on this?

That'll only screw up the final framerate. If encoding at 29.97fps (as when deinterlacing it), you don't apply pulldown during the encoding or afterwards. And be sure to set the correct framerate in CCE.

[spiritgumm]

I tried Antiflicker, but not knowing the best combination of parameters and merely experimenting, I wasn't able to improve the picture.
I use HCenc, and I wasn't sure if I enabled 3:2 Pulldown. Trying it again, turns out it wouldn't even allow the option with the Busstop clip "as is," so it wasn't a real issue. So with only added brightening/cropping filters, the output was still blended, and the blinking was gone (or perhaps I should say minimized to the point where it's not distracting).
Now to study logo removal.

[spiritgumm]

Inre to logo removal, there's the first video I started this thread with (a B&W Pal movie, semitransparent logo), and also a color Pal movie with larger opaque logo.
I tried manono's XLogo info which generally works well on the B&W film. Also tried Logotools (NoLogoAuto, and NoLogo), and the logo is diminished but still visible. XLogo seems a better choice, even with the occasional wiggle (as everyone knows). Any other methods that might work well for this kind of logo? Here's one that sounded interesting, but there's no feedback for it:
https://forum.videohelp.com/threads/306371-easy-way-to-remove-alpha-logos-%28with-no-opaque-parts%29

The opaque color movie logo is tough, and the XLogo wiggle is more distracting. It doesnt sound like there are many good ways to remove opaque logo. I'd try Inpaint but it seems daunting at the moment.

[manono]

For the opaque ones Inpaint isn't so hard to use. Its main drawback is that it's so incredibly slow. Save a frame out and make the logo white and everything else black. I use PhotoFiltre for this, although just about any photo editor can be used. Then you need a bunch of filters in the script. The page on it (which I guess you discovered already) is good about saying what's needed:

http://avisynth.org/mediawiki/InpaintFunc

LoadCPlugin("C:\Path\To\AVSInPaint.dll")
LoadPlugin("C:\Path\To\ExInpaint.dll")
LoadPlugin("C:\Path\To\fft3dfilter.dll")
LoadPlugin("C:\Path\To\ttempsmooth.dll")
LoadPlugin("C:\Path\To\medianblur.dll")
Import("Import("D:\AviSynth Stuff\plugins\InPaintFunc.avs")\InPaintFunc.avs")

Note the 'C' plugin above. Then, in the script, something like this:

InpaintFunc(mask="logo1.bmp",loc="68,54,-600,-388",AR=4.0/3.0,mode="Inpaint",speed=10, ppmode=1,pp=100,radius=10.0,preblur=10.0)

Name your BMP with the white-on-black logo on it (mine's called Logo1.bmp). Either give the crop values for the logo's location, or, more easily, something like "TL" (for top left). Give the AR, the mode (InPaint for opaque ones) and the other stuff as required. If you don't do any blurring, you can leave off the 3 cleaning/blurring filters (FFT3DFilter, MedianBlur, TTempSmooth).

I highly recommend trimming off a part of the movie before the logo line to test it to make sure it's working:

Trim(20000,20500)
InpaintFunc...

For me it often takes about the length of the movie just to load it, and then 24 hours or so to run a pass. If you find that intimidating, just run the XLogo on it. I find that XLogo leaves behind more 'wiggling' as you called it.

As for the translucent ones, I find that LogoTools usually does a slightly better job, but takes more time to set up and fool with the settings. XLogo is fairly easy (read the doc because it has a lot of settings which can also improve the logo removal), once you get the logo mask prepared correctly. Sometimes I have trouble with that - not enough black around the outside, or not all the black is really black.

XLogo's not mine. AlanHK turned me on to it:

https://forum.videohelp.com/threads/273109-Remove-an-opaque-logo-using-Xlogo-in-Avisynth

[spiritgumm]

Thanks for the scripting info on Inpaint. Given the extra processing time, I would probably only try it on something extremely important (to me). I'll give it a shot just to see how it looks, but for now I'd like to decide on one of the quicker filters.

As it turns out, my b&w Pal video logo has opaque and semitransparent elements, so no single filter will fix it. Logotools (Nologo) made the logo fainter (and seemed better than Delogo which I tried in Vdub). If NoLogoM might be better than NoLogoAuto, can you advise about the parameters? Would you recommend Nologo followed by XLogo?
In XLogo and other filters, I see settings for "number of passes." Is Avisynth making several passes before encoding?

My other logo removal project is a color Pal avi. The logo is opaque, so Logotools (NoLogo) only made it black.
XLogo fills in the logo alittle too wildly, and doesn't quit in static scenes. By contrast, MSU Logo Removal in Vdubmod filled in the logo more smoothly, changing only when the video changed (though both filters didn't always use appropriate fill-in during static scenes).
AlanHK sticks with the defaults in his XLogo guide, and while I've read the readme, I'm still hoping you can you recommend adjustments to make XLogo mimic MSU's behavior (or maybe MSU can be used in Avisynth?).
I'm attaching a still image showing the logo (top left corner), then MSU removal and XLogo removal videos.
murder logo.jpg
msu_murdr.m2v
xlogo murder.m2v
msu bad.jpg

[manono]

The MSU one doesn't look too bad. Maybe I'll play around with it some time.

If NoLogoM might be better than NoLogoAuto, can you advise about the parameters?

No, not really as the parameters used depend entirely on the logo. And to be honest, I spend entirely too much time already removing logos from my own projects (damn Indian DVDs and VCDs!). These things take a lot of effort and I have neither the time nor the energy to take on additional projects.

The logo is opaque, so Logotools (NoLogo) only made it black.

Yeah, LogoTools is no good on Opaque or even barely translucent logos.

AlanHK sticks with the defaults in his XLogo guide,

Yeah, and there's more that can be done to improve things. Not that AlanHK wouldn't approve of further tweaking the settings. His guide is a basic getting-started-with-XLogo type guide. But since I've never used the MSU one, I can't make any suggestions to make XLogo more like the MSU one.

If you're going to be playing around with a lot of these caps, and if they often have logos, then it's in your best interests to learn about the different methods to remove the logos, and which work best under which circumstances. Then pretty soon we'll all be learning from a guide you write.

[spiritgumm]

Then pretty soon we'll all be learning from a guide you write.

I think you meant " ." When it comes to software (to borrow a joke), I'm an idiot savant minus the savant.

I had uploaded the wrong XLogo clip in previous post (now reupped correct clip). The logo area is in constant flux, compared to the more concrete MSU clip. I also uploaded a still from a different MSU sample showing the flipside of this advantage (logo filled in too much). Unfortunately there aren't many setting to adjust in MSU, and it's Vdubmod, and I want to stay with Avisynth. I'll see if I can play with the XLogo settings to make them behave better.

[spiritgumm]

Haven't had any luck with tweaking XLogo... so if I want to use MSU Logo Remover, what are the drawbacks to using Vdubmod? If a loss in video quality, would it be noticeable when converting mpeg4 to mpeg2? I would be frameserving to HCenc.
To convert resolution of a 16:9 Pal mpeg4 to Ntsc mpeg2, I only need to add Resize and Border filters, right?
To properly change the framerate, do I use a filter, or use the option under the VIDEO dropdown menu?


I tried following these instructions to load MSU into Avisynth via Vdubmod:
http://forum.doom9.org/archive/index.php/t-95955.html
I adapted the script - 3 lines below - but get errors since I don't really know what I'm doing.
When I saved the Vdubmod Processing settings, the bottom line "Config" was: Config(10, "10.000000"). When I put the numbers into the _MSUDelogo(), got an error about the first quote mark, so I omitted it.

Global VirtualDubMod = "C:\Program Files\AviSynth 2.5\VirtualDubMod\plugins"

LoadVirtualdubPlugin(VirtualDubMod+"\MSU_LogoRemover.vdf", "_MSUDeLogo")

ConvertToRGB32._MSUDeLogo(10, 10.000000")

I get a script error with either/both of the last lines present. Any help?

[manono]

The main drawback is speed. You'll lose some when using VDub Filters. Most of the MSU filters can be used within AviSynth, if I remember correctly. This one doesn't have AviSynth instructions?

Anyway, when planning on doing some filtering in VDub, I think you're supposed to convert the colorspace to RGB in the script.

To convert resolution of a 16:9 Pal mpeg4 to Ntsc mpeg2, I only need to add Resize and Border filters, right?
To properly change the framerate, do I use a filter, or use the option under the VIDEO dropdown menu?

Yeah, if the source is progressive already, you only need resize (720x576 to 720x480). It may or may not be necessary to Crop/AddBorders or Letterbox, depending on whether or not you like to add "fresh black". Unless there's some additional cropping that has to be done to the sides. Then it might get a bit more complicated. FitCD says to do it slightly differently, but I wouldn't do it that way:

LanczosResize(704,480)
AddBorders(8,0,8,0)

I always just do a straight resize. You can change the framerate in the script also, as in AssumeFPS(23.976).

[spiritgumm]

you only need resize (720x576 to 720x480)

It's progressive but 624 x 352. I was going to follow Fitcd: LanczosResize(704,480), AddBorders(8,0,8,0).
Any reason I shouldn't do that? Isnt it just trying to maintain correct aspect ratio?

Now if use Vdubmod, where do I integrate a script? I know there's a script window I can open, but how does it work in concert with the rest of the program?

Inre to MSU in Avisynth, I only found that one reference (above) which I was hoping someone would elucidate.

[AlanHK]

Now if use Vdubmod, where do I integrate a script? I know there's a script window I can open, but how does it work in concert with the rest of the program?

You save your script in a text file with the .avs extension, then open that from VDubMod.
You can edit that in VDubMod using control E; using F5 will save changes and refresh the display.

This is very useful for making quick changes, Trims, etc, but for building complex scripts and adjusting parameters, AvsP is much more powerful.

Inre to MSU in Avisynth, I only found that one reference (above) which I was hoping someone would elucidate.

To use any VirtualDub filter in Avisynth you get the parameters from the vcf file, in a rather cryptic but useable form. Follow these instructions.

You can go on from that to write a wrapper function to make it more user friendly, as in files linked from the wiki page above.
If you make one, post it here, on Doom 9 and/or the wiki so others can benefit.

As for my "stick with the defaults" remarks in my Xlogo guide, its simply that I get a "good enough" result that way and don't want to invest the time to tweak further. As it is now I can make and apply a working filter in 2 minutes. That's enough for a video I will probably only watch once. I'm sure you can get better results if it's important to you. Also, while Xlogo and Nologo have a small cost in processing, better ones are much slower.

[spiritgumm]

Haven't done anything yet, but have a MSU question. Even if MSU is looking forwards and backwards fill in the logo, sometimes it doesn't do it ideally (see attached jpeg in previous post, from outputted video).
While previewing the video in Vdubmod, if I paused and restarted from a poorly filled spot, the logo area would change and look great. Is there a way to have MSU pause, reevaluate and reimplement it's process? Or is the only way for me to set time points manually (which would be too time consuming)?

[AlanHK]

Haven't done anything yet, but have a MSU question. Even if MSU is looking forwards and backwards fill in the logo, sometimes it doesn't do it ideally (see attached jpeg in previous post, from outputted video).
While previewing the video in Vdubmod, if I paused and restarted from a poorly filled spot, the logo area would change and look great. Is there a way to have MSU pause, reevaluate and reimplement it's process? Or is the only way for me to set time points manually (which would be too time consuming)?

I'd guess that's what is mentioned on p 10 of the manual:

Use only blur is for scenes with complex motion and filter can not
remove logo with good quality.

Blur threshold is for changing when blur will be used. Lower
value means that blur will be used more often. It is useful when
video has scene changes that couldn’t be found with this filter –
then user decreases the Blur threshold and after scene changes
logo will be blurred.

So try reducing the Blur threshold.

[spiritgumm]

Already tried tweaking the MSU settings. It's not a matter of the blur threshold, blur-only, etc. You might have try it to see what I mean. For example, in the above MSU jpeg with blocky logo fill-in, if I were playing the video in Vdubmod and paused at that moment, then started playing, the fill-in would look perfect.

Side note, my mask wasn't working so great, but when I used the Repair mask (which Delogo generated from my mask), MSU worked much better.

UPDATE
To use MSU in Avisynth, I made a simple functioning script based on the "FAQ using virtualdub plugins" page, but am having a problem. The mask (bit depth 24) which works fine Vdubmod isn't covering the white parts of the logo. Does it have anything to do with converting to RGB32? I can upload a sample video if needed.

Used the VDubmod process settings in vcf file - Config(10, "10.000000").
I set loading plugin preroll to "50" because a lower preroll outputted very pixelated video at the start.

LoadVirtualdubPlugin("C:\Program Files\AviSynth 2.5\plugins\MSU_LogoRemover.vdf", "VD_MSU", 50 )
AVISource("C:\MurderShe.avi")
ConvertToRGB32()
VD_MSU(10, "10.000000")
ConvertToYV12()

[spiritgumm]

Well, I didn't get any replies about what i was doing wrong with my MSU plugin script, so I had to use Vdubmod. I used an Avisynth script to change the framerate, and opened that up in Vdubmod, and used the MSU and resize filter there. Avisynth to HC Usually takes about 2 hours, but this took 6 hours. Was this from using the MSU filter, or because I was using VDubmod? Did I save any time by doing the frame rate change in Avisynth?
Btw, results were pretty good, same as the sample I posted earlier. The audio was off slightly, so I had to tinker with that.

Now onto subtitles. Any advice about processing optional subs in Avisynth or Vdubmod?

What I did recently with .SRT (for widescreen PAL mpeg4 converted to Ntsc) was use "Subtitle framerate changer" and Txt2sup. Muxed everything together, but still need to adjust the subs color and location.
If using DVDSubEdit, and the movie is widescreen with some added top/bottom borders, where do you place the subtitles (with the editor window set for 16:9 widescreen)? So top line of text is over the image, and bottom text is in the black? Or should the bottom line be in the image too?

[spiritgumm]

Question: How to specify various times to apply two different logo remover filters? This is in regards to my first video in this thread.
I think you can set in/out times in Vdubmod, but I'm hoping using Avisynth would process it quicker. I'm using:
Nologo (lightens logo)
XLogo (blurs out logo area)
It appears that XLogo will output the same video whether Nologo was run before it or not (from previewing in AvsP). That suggests to me that the whole movie can be processed in NoLogo by default, with only XLogo requiring whatever clip times to be specified. But how do I do that? Here's my script for running both together:

A=MPEG2Source("C:\RedLight\VTS_01_1.d2v", cpu=0).ConvertToYUY2()
B=ImageSource("C:\RedLight\VTS_01_1 copy.bmp").ConvertToYUY2()
NoLogoAuto(a,b,30)
XLogo("LogoFinal1_x_32_y_24_2.bmp", X=32, Y=24, alpha=10)