How to reverse telecine 1080i29.97 to 1080p23.976

[MrVideo]

If this has already been answered somewhere here, please point me to that thread.

I want to correctly reverse telecine program material that was sourced from 1080p23.976 to 1080i29.97 for broadcast. I was using the following script:

AVCsource("f:\filename.dga") [or DGDecode_mpeg2source("e:\Reaper-HD-209-090428.d2v") depending on the source being H.264 orMPEG2]
tfm()
saved=last
cropbottom(300)
tdecimate(clip2=saved)

I didn't discover a problem until last night when I was working on some stuff and because of a problem, remembered that I now have some H.264 source material that has a cadence leader. This leader is there to make sure that the 23.976p material correctly converted to 1080i29.97. It makes for a great test for going the other way. The material is marked with an A for the odd/A field and a B for the even/B field.

What I discovered after running the above script is that the even fields are thrown away, effectively reducing the vertical resolution in half. I know this because the resulting video, while at 23.976, only contains the A field.

When working with MPEG-2 material, the odd/A field is thrown away, unless I set the following: tfm(order=1)

The cadence video has four basic frames: A, B, C & D (makes sense). These get converted to interlaced video as follows (normal 2:3 pulldown):

source: |A |B |C |D |
result: |AA|BB|BC|CD|DD|
field: |OE|OE|OE|OE|OE|

What needs to happen is video frames 1, 2 and 5 need to take the two fields and combine them into 3 new progressive frames. That is the easy part. The other part is the throw way the odd field of video frame 3 and the even field of video frame 4. That leaves one with the "C" source fields, which are now combine in the remaining 4th progressive frame.

What does it takes to do the above?

Thanks.

[wonkey_monkey]

How about:


assumeframebased # just in case
doubleweave
selectevery(10,0,2,5,7)


As long as the cadence holds up throughout, I think that should work.

David

[MrVideo]

As long as the cadence holds up throughout, I think that should work.

I can't guarantee cadence at all. The file contains a leader of 30 seconds of video bars, then 30 seconds of black, then the edited program, where cadence can definately be lost at each commercial break.

Does your code replace:

tfm()
saved=last
cropbottom(300)
tdecimate(clip2=saved)

[manono]

What's wrong with a simple:

TFM()
TDecimate()
CropBottom(300)#if necessary

or:

Telecide(Guide=1)#the Decomb IVTC
Decimate()
CropBottom(300)#if necessary

[MrVideo]

What's wrong with a simple:

TFM()
TDecimate()
CropBottom(300)#if necessary

or:

Telecide(Guide=1)#the Decomb IVTC
Decimate()
CropBottom(300)#if necessary

I do not know, since I am not a savy avisynth user. The code I have I got from someone else and do know know what is going on. Your example removes the (clip2=saved) option. Like I said, I have no clue what is going on, which is why I've asked for help.

[MrVideo]

How about:


assumeframebased # just in case
doubleweave
selectevery(10,0,2,5,7)


Sorry to say, but it made mincemeat out of the video.

[rebkell]

What's wrong with a simple:

TFM()
TDecimate()
CropBottom(300)#if necessary

or:

Telecide(Guide=1)#the Decomb IVTC
Decimate()
CropBottom(300)#if necessary


I actually gave him the script. The CropBottom(300) and saved=last are used so that tdecimate only compares the upper part of the frame in it's decision on which frame to drop.(it tends to get confused and makes a lot of bad decisions when the animated logos are dancing all over the bottom of the screen) The clip2=saved tells it to output the original frame after it decides which frame to drop.

[MrVideo]

TFM()
TDecimate()
CropBottom(300)#if necessary

No different than my original source. The B/even field is dropped.

Telecide(Guide=1)#the Decomb IVTC
Decimate()
CropBottom(300)#if necessary

Still dropped the B/even frame, but also had glitches.

[manono]

The CropBottom(300) and saved=last are used so that tdecimate only compares the upper part of the frame in it's decision on which frame to drop.
Oh, OK, although there's no way anyone would have known that. Anyway, TFM (and Decomb, both) has its own setting to take that into account:
y0/y1 -

These define an exclusion band which excludes the lines between y0 and y1 from
being included in the field matching decision. An exclusion band can be used
to ignore subtitles, a logo, or other things that may interfer with the matching.
y0 sets the starting scan line and y1 sets the ending line, all lines in between
y0 and y1 (including y0 and y1) will be ignored. Set y0 equal to y1 to disable.

Default: y0 - 0 (int)
y1 - 0 (int)
No different than my original source. The B/even field is dropped.
I'm not sure I accept that argument of yours that half the resolution is lost based on the evidence you give. If the frame is static (I'm assuming these letters aren't jumping all around), even if it is deinterlaced (which I'm also not sure I buy), because the default deinterlacer is motion adaptive you get back full resolution. But there are many others that know the inner workings of TIVTC better than I.

[rebkell]

Oh, OK, although there's no way anyone would have known that. Anyway, TFM (and Decomb, both) has its own setting to take that into account:


It's not a matching problem, it's the decision that tdecimate makes on which frames to drop. The animated logos are probably running at 30FPS and tdecimate has to drop one frame and with all the action going on with the logos, it makes a lot of bad decisions(which I fully understand and can't fault it for)

[MrVideo]

I'm not sure I accept that argument of yours that half the resolution is lost based on the evidence you give. If the frame is static (I'm assuming these letters aren't jumping all around), even if it is deinterlaced (which I'm also not sure I buy), because the default deinterlacer is motion adaptive you get back full resolution. But there are many others that know the inner workings of TIVTC better than I.

As mentioned in my original posting, the cadence test countdown video has each field marked with a letter. The odd field has the letter A and the even field the letter B. If the two fields were combined, you'd see both letters, which is what you see if you do a frame grab and look at it with Photoshop.

The code that I've been using only results in the final having the fields with the letter A. The B is completely missing from the result, therefore that field is being dropped. When working with MPEG-2 video, the A frame was being dropped by default, leaving only the B field in the resulting video. If I add "order=1" to tfm, then the B field is dropped.

Just how is Y0/Y1 used with tfm? tfm(y0=780,y1=1079) ??? where the first scan line is 0, or is the first scan line 1 (different programmers have different ideas on 0 or 1 for a start)?

[Guest]

@MrVideo

Please post a link to the unprocessed test countdown video that we can use to replicate your scenario.

[MrVideo]

@MrVideo

Please post a link to the unprocessed test countdown video that we can use to replicate your scenario.

Done. Here is the zip (http://vidiot.com/Cadence-Leader-H264.zip) file.

[Guest]

What happens with post=0 for Telecide()?

[Guest]

Sorry, but you are going to have to be more explicit as to exactly what you want me to run as a script. This works just fine for me:

dgmultisource("E:\tmp\Cadence-Leader-H264.dgi")
assumetff()
telecide(post=0)
decimate()

[MrVideo]

This works just fine for me:

dgmultisource("E:\tmp\Cadence-Leader-H264.dgi")

What program produces .dgi files? I use the DGAVIndex program to produce .dga files.

I'm off to see Avatar at the local minIMAX, so I'll test things out when I get back.

[buzzqw]

DGI files are produced by DGDECNV (lastest is 200b8)

very very handy if you have nvidia (vp2+) cards

BHH

[laserfan]

I think MrVideo that you need simply to add the last 3 lines to your own Source avisynth entry.

[Guest]

Yes, what laserfan said:

avcsource("Cadence-Leader-H264.dga")
assumetff()
telecide(post=0)
decimate()

[MrVideo]

I think MrVideo that you need simply to add the last 3 lines to your own Source avisynth entry.

Didn't know for sure if the VP2 support was needed to make this work. Back from the movie, so will go test.

[Guest]

Use avcsource() if you don't have the NV tools.

[MrVideo]

Posting removed as I screwed up and ran the wrong AVS script. The next posting is the correct run of the script.

[MrVideo]

I did. No joy. Sorry about the delay in responding, as the system was busy doing feed captures and when it is busy doing that, you can't load it down doing other intensive stuff.

In any event, same problem, but like one of the other suggestions, there are glitches. I even did enough to go into the program material, in which I discovered that about every second, there would be a duplicated frame, which gave it a jerking motion.

I've grabbed some full frame image stills to show the problem.

Image 1 (http://vidiot.com/images/Cadence-H264-Test-01.png)
This is the first frame after processing that starts at the beginning of the countdown. The A & B fields are there.

Image 2 (http://vidiot.com/images/Cadence-H264-Test-02.png)
Image 3 (http://vidiot.com/images/Cadence-H264-Test-03.png)
These are the next two frames. Look at the breakup of the fields.

Image 4 (http://vidiot.com/images/Cadence-H264-Test-04.png)
Image 5 (http://vidiot.com/images/Cadence-H264-Test-05.png)
These two images show massive glitched program content.

This shouldn't make a difference, but the video is being handed off to the x264 encoder for the final output.[/QUOTE]

[MrVideo]

This time I tried:

doubleweave()
pulldown(0,3)

I got the exact same results as the telecide(post=0).

Something is weird here.

[MrVideo]

Because neuron2 said that he had no problem with my source, I wondered if something is wrong with how the H.264 file was being handled, i.e., a bad H.264 codec.

I noticed that the H.264 frame grabs that I got out of VideoReDo appeared to be blended. I chalked it up to how VRD displayed the video. But, I remembered that if I had VRD re-encode to MPEG-2, the blending was gone.

So, I had VRD re-encode the test video to MPEG-2 and I then ran that through the script.

That worked.

Now to try and determine what the issue is with the H.264 handling, as I certainly to not want to spend hours converting to MPEG-2, only to turn around and convert to H.264 23.976.

Would getting a new Nvidia card with VP2, or later, will make a difference? I suspect that it would be faster, considering that the CPU would be involved less.

I also have to process 720p video, would this script be correct:

AVCsource("F:\filename.dga")
saved=last
cropbottom(240)
tdecimate(clip2=saved,cycler=3)

Thanks in advance. This has been a real hairpuller.

[Guest]

Now to try and determine what the issue is with the H.264 handling You have PAFF streams and DGAVCDec is known to have issues with some PAFF streams.

Would getting a new Nvidia card with VP2, or later, will make a difference? I suspect that it would be faster, considering that the CPU would be involved less. Yes, you will get correct decoding for all PAFF streams. You will also duplicate my results.

I also have to process 720p video, would this script be correct There's no way to say without seeing the source video.

[MrVideo]

You have PAFF streams and DGAVCDec is known to have issues with some PAFF streams.

I went googling for a definition of PAFF, but all I found was stuff that deals with it. What is PAFF? When I run the file through MediaInfo, it says nothing about PAFF, only that it is TFF.

Yes, you will get correct decoding for all PAFF streams. You will also duplicate my results.

Thanks, I'm looking at a new PNY card (VP4).

There's no way to say without seeing the source video.

OK, here is a 30 second sample of the ABC network feed (http://vidiot.com/ABC-H264-Sample.zip).

Do you think that all DVB-S2 H.264 feeds will be PAFF? I suppose that I can just try the AVS script on a sample and if I get garbage, it ust be :)

Thanks again for all your help and understanding.

[Guest]

I'm downloading your clip. Meanwhile, I'll answer the other stuff.

I went googling for a definition of PAFF, but all I found was stuff that deals with it. What is PAFF? When I run the file through MediaInfo, it says nothing about PAFF, only that it is TFF. It means Picture Adaptive Field/Frame. That means that the stream can contain a mixture of frame-coded and field-coded pictures. Some PAFF streams are not properly handled by the version of libavcodec I used in DGAVCDec.

Do you think that all DVB-S2 H.264 feeds will be PAFF? Not necessarily, but as PAFF is becoming quite common one should have a decoding solution that supports it properly. Your PNY card plus DGDecNV will serve you well in that regard. You can also use that environment to determine whether the stream is PAFF and to perform other useful analyses.

[MrVideo]

I'm downloading your clip.

Great, thanks.

I used the following on a MPEG-2 HD 1080i source and actually ended up cutting off the bottom 300 lines.

assumetff()
telecide(post=0)
cropbottom(300)
decimate()

What does one put into the script to keep the bottom 300 lines from being referenced during the reverse telecine processing, but not be thrown away?

Those damn snipes are a royal pain.

Since you were answering questions, any comment on my understanding, or not understanding of, the math given with the Wiki manual entry for pulldown()? While it seems I don't have to use that function, it would be nice to understand how those math formulas were derived.

[Guest]

What does one put into the script to keep the bottom 300 lines from being referenced during the reverse telecine processing, but not be thrown away? Manono already told you:

http://forum.doom9.org/showthread.php?p=1357951#post1357951

So, for Telecide() to ignore lines 200 through 499:

Telecide(...,y0=200,y1=499)

Since you were answering questions, any comment on my understanding, or not understanding of, the math given with the Wiki manual entry for pulldown()? While it seems I don't have to use that function, it would be nice to understand how those math formulas were derived. I'll move that into another thread as it is off-topic here.

EDIT: Regarding the ABC stream you uploaded. It is 3:2 *frame* pulldown, so you can easily return it to film rate with:

TFM()
TDecimate(cycle=5,cycleR=3)

[MrVideo]

Manono already told you:

Yes and no. He mentioned y0/y1 but gave no specifics as to how it is used. There was no reference as to which function that it is used with. I asked if the first scan line is referenced as 0, or 1 (different programmers have different as to what is the start). I never got a response to that query.

So, for Telecide() to ignore lines 200 through 499:

Telecide(...,y0=200,y1=499)

Thanks, now I know where it is used.

Regarding the ABC stream you uploaded. It is 3:2 *frame* pulldown

I would have really been surprised if you said it was anything else.

TFM()
TDecimate(cycle=5,cycleR=3)

Thanks. Where does the y0/y1 fit in here, since telecide isn't used? I did a mediawiki search for y0/y1 and y0, but came up with nothing.

Happy New Year1

[Guest]

Yes and no. He mentioned y0/y1 but gave no specifics as to how it is used. There was no reference as to which function that it is used with. You have to learn to read product documentation if you want to be able to write scripts *yourself*.
I asked if the first scan line is referenced as 0, or 1 (different programmers have different as to what is the start). I never got a response to that query. It's documented but I'll make it easy for you. It's 0-based.

Where does the y0/y1 fit in here, since telecide isn't used? I did a mediawiki search for y0/y1 and y0, but came up with nothing. The TFM() documentation gives you the parameters. Hint: they are the same as Telecide's.

[MrVideo]

TFM()
TDecimate(cycle=5,cycleR=3)

Worked great. Even with a leader of 30 seconds of video bars and 30 seconds of black, it was correct when the program material started.

Thanks again.

BTW, the PNY Gt240 card order has been placed. I should have it in my hands by next Friday.

I'll get the software ordered as well, so it will be licensed and ready to go when the card gets here.

[MrVideo]

You have to learn to read product documentation if you want to be able to write scripts *yourself*.

Under normal circumstances, that is indeed what I do. If I could find a user manual for AVIsynth and its filters (preferably in PDF format), then I would. A user manual not only describes all of the features, but builds on them, starting out with the simple things in the early chapters and then builds on more complex things, in later chapters. The command line reference would normally be a separate section.

What I have found is here: http://avisynth.org/mediawiki/Main_Page

While it appears to fill the requirements that I mentioned above, I can't find it in PDF, i.e., in a book. A book that I can print and parse through. A search of the PDF would have found the y0/y1 references (though I don't know why the wiki didn't).

Yes, on-line documentation is the wave of the future, but have you tried flipping through the pages of an on-line manual? I'm not trying to be funny here. It is extremely difficult. There's no index to look at in the back of the book, as there is no back of the book.

A book, hopefully, would have a section devoted to the telecine process and how one can remove the 2:3 telecine process. A search of the wiki for "reverse telecine" yields nothing.

If we all read and understood what is in the wiki, there wouldn't be a need for the forum, as we all would know the answers. But, we don't. Even after reading the wiki for hours, days even, I still might not have come up with the things you said to use. I certainly would have been pulling my hair out as to why stuff that should have worked, didn't with the PAFF H.264 files. That isn't in the wiki. That is why I use forums like this one. The experts, and authors, are here that know AVIsynth inside and out.

That said, I went looking for tfm. The wiki search got me to here: http://avisynth.org/mediawiki/TFM There is indeed a nice long list of parameters for tfm. None of which are linked. No usage examples either. Ultimately I find tfm used under tivtc, here: http://avisynth.org/mediawiki/TIVTC None of the examples are what you provided for getting 720p to reverse telecine. Plus, only a very small number of the long list of paramerters is used in some examples. There is no reference section at all for the parameters.

Next I searched for telecide. That leads to the external filters page, where telecide is not in the list, but is mentioned by others. I click on a link and get: You have followed a link to a page that does not exist yet.

The conditionalfilter page uses it in an example, but still no reference page. Same goes for the VFR page and the rest of the 7 items returned in the search.

If I sound frustrated, you are correct. I have no clue as to where this stuff is referenced.

[manono]

Where did you get the TIVTC package? Didn't yours come with documentation? The y0/y1 stuff I gave earlier was a direct quotation from the included TFM - README.txt. Most of the filters include docs for their use, some with better docs than others. neuron2's are probably the best in the business, but tritical's (creator of TIVTC and many others) are pretty good also. When in doubt, search the WarpEnterprises link for the filters:

http://avisynth.org/warpenterprises/

If you download TIVTC from there (or tritical''s site) you'll find a whole host of docs that thoroughly explain its use.

Earlier I mentioned that Telecide was part of the Decomb package that also includes Decimate and Field Deinterlace. You can download Decomb from WarpEnterprises or from neuron2's own site:

http://neuron2.net/

[MrVideo]

Didn't yours come with documentation?

I've now gone digging into the installation (version 2.58), and sure enough, there are separate text files.

I do all of my main web browsing on my Solaris 10 x86 server. So, I went to the avisynth.org web page in order to find documentation. There is documention there. Every other program that I've downloaded from its page, the documentation is part of that site. I did not expect anything distributed with avisynth to be drastically different from the software's main site. To have t broken up into html and text files, IMHO, is not very user friendly. Makes it extremely difficult to search. It certainly doesn't make for a user guide of any sort.

Why isn't all of the documention that is included in the package placed on line? Having the tfm text file content as part of the wiki documention (along with all of the other docs) would be ideal for all users.

That is why I didn't see it, as I was under the wrong impression that I was looking at the documention,

avisynth, along with its plugins/filters/etc. is a very complicted and intricate piece of software. Having documentation spread out like makes it hard learn. Maybe someone needs to write a avisynth user and reference guide book.

Remember, you are on the inside looking out. I'm an avisynth nube on the outside looking in.

The y0/y1 stuff I gave earlier was a direct quotation from the included TFM - README.txt. Most of the filters include docs for their use, some with better docs than others. neuron2's are probably the best in the business, but tritical's (creator of TIVTC and many others) are pretty good also. When in doubt, search the WarpEnterprises link for the filters.

I believe I got the distribution from sourceforge. So, all of what I have installed came from the AviSynth_081227.exe file that I have.

You may think that I'm an idiot, so be it. But, I hope you can understand the logic I used for looking for the documentation.

So, when are you and neuron2 going to write the user and reference guide and get O'Reilly books to publish it?

[laserfan]

Why isn't all of the documention that is included in the package placed on line?
Then it has to be maintained in multiple places. I consider it lucky that the documentation exists at all, and that it is as good as it is (though DG's is superior to most/all of course!). :)

Your best bet I think for all things Avisynth is to start with the docs that come packaged with each filter. If a filter comes with an installer then it can be found in the docs directory where the plugins are located (within your Avisynth program directory).

Oh, and maybe you meant "publish" (as in hard-copy book) as a joke, but really, given how dynamic some of these tools tend to be, and the frequent changes (that we enjoy) with them, a book would be obsolete before the ink dried on it...

[MrVideo]

Oh, and maybe you meant "publish" (as in hard-copy book) as a joke, but really, given how dynamic some of these tools tend to be, and the frequent changes (that we enjoy) with them, a book would be obsolete before the ink dried on it...

I did mean publish as in soft-cover (O'Reilly books does softcovers). A lot of books are obsolete before the ink even hits the paper. A prime example is subversion. It is constantly under development, but it has a book that is way out of date. It makes for a great starting point for learning the software, which is the point.

A book on avisynth, and its many filters, would be a great place to learn a lot of tricks. When documentation is piecemeal, and for the most part, a reference, you don't get a picture of how it all fits together. It also can't be searched, easily.

I'm old school, in that I prefer a book in which to thumb through.

As for frequent changes, what frequent changes? I'm gathering I need to keep an eye on warpenterprises. But even that page lists an old avisynth release and admits that it might not be up-to-date.

As for DG's docs, I did an initial look and that does indeed start off slow and easy. I'll have print out those pages.

[Wilbert]

I did mean publish as in soft-cover (O'Reilly books does softcovers). A lot of books are obsolete before the ink even hits the paper. A prime example is subversion. It is constantly under development, but it has a book that is way out of date. It makes for a great starting point for learning the software, which is the point.

A book on avisynth, and its many filters, would be a great place to learn a lot of tricks. When documentation is piecemeal, and for the most part, a reference, you don't get a picture of how it all fits together. It also can't be searched, easily.

You mean something like this:
http://www.wilbertdijkhof.com/AviSynth-Documentation_20100101.pdf

Yes, i know it's not perfect yet (any external plugin descriptions are missing (which ones should be included?), tocs is a mess, some lines are too long, some pics are missing, structuring, etc ....

I such a document useful?

[MrVideo]

You mean something like this?
I such a document useful?

Yes, it looks to be very useful. I'll have to print it at work :cool:

A suggestion would be to add Don's avisynth core html documentation to the front of this guide and call it an avisynth and external plugin manual.

[IanB]

@Wilbert,

I think MrVideo is trying to volunteer to help update and reorganise the documentation :D :D :D

[Leak]

A prime example is subversion. It is constantly under development, but it has a book that is way out of date.
At least the subversion book doesn't look outdated at all (http://code.google.com/p/svnbook/source/browse/trunk/src/en/book/) to me - unless you happen to be running experimental versions of SVN that are unreleased...

Also, expecting to find documentation of external third party plugins in the AviSynth documentation is a bit much, don't you think? Yes, some of them are partially documented there, but if you need documentation for those, it's up to the authors of the third party plugins to provide it and not the AviSynth maintainers...

np: Bibio - The Palm Of Your Wave (Bibio Remix) (The Apple And The Tooth)

[MrVideo]

I think MrVideo is trying to volunteer to help update and reorganise the documentation :D :D :D

Ya, just what you want, someone who is a noob writing/updating documentation. :rolleyes:

[MrVideo]

At least the subversion book doesn't look outdated at all (http://code.google.com/p/svnbook/source/browse/trunk/src/en/book/) to me - unless you happen to be running experimental versions of SVN that are unreleased...

Current subversion is 1.7 and my book is 1.2. AFAIK, there are subtle changes. In any event, some of the documentation for the filters goes back 2-3 years, without any changes. I wouldn't cal that constantly changing documentation :rolleyes:

Also, expecting to find documentation of external third party plugins in the AviSynth documentation is a bit much, don't you think? Yes, some of them are partially documented there, but if you need documentation for those, it's up to the authors of the third party plugins to provide it and not the AviSynth maintainers...

No, not really. If the external plugin is documented at all on the avisynth wiki, there is no reason that it can't include what is distributed with the filter. At least it could incude a note to consult te documentation with the filter for more info. I do not remember seeing that.

[Leak]

Current subversion is 1.7 and my book is 1.2.
Not that I want to drag this out, but current stable subversion is 1.6.6 (http://subversion.tigris.org/) and the book's nightlies are also at 1.6.x (http://svnbook.red-bean.com/nightly/en/index.html) - and since further minor revisions of 1.6 won't get new features there shouldn't be many updates to the docs for them...

(If you're talking about the dead-tree edition of the SVN book - well, you won't get updates for those of course; but if you want a PDF version of the nightlies version you can find it on their site next to the link I gave above...)

[MrVideo]

Not that I want to drag this out, but current stable subversion is 1.6.6 (http://subversion.tigris.org/)

Sorry, must have misread my subversion maillist mail. Don't have my laptop on, as I get that mail list through work.

You are quite correct, I can keep up-to-date with the PDF distribution and will be doing that in the next week, or so.

It would be nice if the avisynth documentation was as good as the subversion documentation. Wilbert's filter PDF is a big step in that direction.

But, the subject of documentation is outside of the scope of this thread.

[Wilbert]

No, not really. If the external plugin is documented at all on the avisynth wiki, there is no reason that it can't include what is distributed with the filter. At least it could incude a note to consult te documentation with the filter for more info. I do not remember seeing that.
Sure, there is a reason. Functionallity of plugins gets updated and docs need to be updated too. Also, there are hundreds of plugins. Good luck keeping the avisynth wiki up to date with respect to those. Your second suggetion is a good one though.

It would be nice if the avisynth documentation was as good as the subversion documentation. Wilbert's filter PDF is a big step in that direction.
Yes, it is. But the problem is volunteers. Although people are helping sometimes i can't even get this page: http://avisynth.org/mediawiki/External_filters finalized, and that's only a "small" part of things which need to be done.

[MrVideo]

This works just fine for me:

dgmultisource("E:\tmp\Cadence-Leader-H264.dgi")
assumetff()
telecide(post=0)
decimate()

New PNY GT240 card has been install and is up and running.

The license for the software has been installed.

The script modified with the above, as well as adding the necessary loadplugin line. If I copy the DLL the to the auto load area, I might forget to copy a new version, so best to just load it.

I ran my test file through my script (after minor modifications) and the encoded cadence, plus several seconds of program material, encoded beautifully.

Like all others before me, thanks for a great piece of software.