KVCD: Confused by Telecide
 

Read somewhere at the forums abt telcide and decimate and fieldeinterlace.

Most of the dvd's i have when loaded with dvd2avi show
Fps - 29.97
Video Type - Ntsc
Interlaced

I ran the preview with dvd2avi it shows in video tpye ntsc 100%.

Should i save the project as forced film.
I have been saving the project normally and using the ma script and encoding to kvcd.

Do i need to add any lines to my script

Telecide() , Decimate() or Fielddeinterlace()

or just save as forced film and encode at 23.976 fps

look at my interlacing guide at the KVCD section!

1. Is your Movie a hollywoodfilm, means shot on Filmmaterial?
Thsi means it was originally digizized using 23.976 fps!

2. DON'T waste your time and do count on what the right window of DVD2AVI says! Instead, move using the scroll bar below to a moving scene ... do combs appear?? Yes?? Then its a telecined one and has to be generated as d2v WITHOUT forced film! Afterwrads in Avisynth a Telcide().Decimate() will gfet rid of that telecined Movie as it was telecined to 29.97 incl. combing. And by using that telecine/decimate combination in avisynth you will perform an INVERSE TELECINE which brings the stream back to its orig. progressive (not combed/interlaced) 23.976 state.

If NO combs appear in DVD2AVI on motion scenes ... do use "force film" and you will obtain the right 23.976 progressive stream on your HD

:)

I did some tests.

I save dvd2avi project as forced film and encoded to kvcd , the prediction went haywire and m1v file was 1.2gb.

I saved it regularly and tried including
Telecide()
Decimate()
GEt this error: "Telecide:you musy specift order parameter (0=bff, 1=tff)

I also treid just including FieldDeinterlace() in the script
This seemed to work fine , have to check after the full encode is over if file size is right.

I tried moving the scroll bar in dvd2avi no combs appear , in some scenes vertical lines appear.

No this is not a hollywood movie but a dvd of an Indian film.

kev23m,
Make sure you are using Decomb511.dll (you appear to be) and then use one of these:

Telecide(order=1)
Telecide(order=0)

Older versions of Decomb would allow Telecide() but new version does not.

I have Decomb510.dll not 511 , will give this a go though.

Thank You.



Edit : Working now with Telcide(order=1) , running some tests . Will let u know abt the results.

I encoded the film using Telecide() & Decimate.

The encode is done properly in size and sync , works with Fielddeinterlace() also , how do i know which is right.

Prolly gonna try forced film tonight.

Telecide/Decimate = Inverse Telecine!

Means a progressive FILM source at 23.976 was telecined to 29.97 :arrow: Frames have been added and a dynamical field shift was applied (combs now do appear) = no jerky picture on TV! :arrow: And Telecide/Decimate RESTORES these 29.976 to its original state of 23.976 progressive again by deleting these added frames (=decimate()) and by processing an inverse Field shift (=telecide()).

Fielddeinterlace = truly deinterlacing!

Everything depends on how your source was SHOT originally, means if the Movie orig. was shot on film (for cinema purposes!) then they did apply a telecine process as explained above (23.976>29.97).

BUT! That would mean you got a bad mastered DVD as most Film material (cinema movies) are just encoded at 23.976 PROGRESSIVE! and a pulldownflag was applied which tells the Standalone DVD player to perform such a telecining WHILE playback internally on-the-fly!
And thats why you got the option "force film", cause by this the pulldownflag will be ignored which causes a processed d2v file which serves the RIGHT and TRUE 23.976 fps at a progressive (=non interlaced) state

:)

Giving me good result with Telecide() & Decimate().

My cq has gone up by 6 and increased the max bitrate by 300.

What can i use in Avisynth to sharpen the image.

The script i am using now.

##
Mpeg2Source("D:\bn1\bn1.d2v")

Telecide(order=0)
Decimate()
GripCrop(352, 240, overscan=2, source_anamorphic=false)
GripSize(resizer="LanczosResize")
STMedianFilter(3, 3, 1, 1)
FluxSmooth(7, 7)
MergeChroma(blur(1.55))
MergeLuma(blur(0.1))

ScriptClip(" nf = YDifferenceToNext()" +chr(13)+ "unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) , 1, 1) ")


GripBorders()
#LetterBox(8,8,8,8)




## Functions ###

function fmin( int f1, int f2) {
return ( f1<f2 ) ? f1 : f2
}

##

Lanczos at 352x288 is already sharp enough!

Dont try to "compansate" these low resolutions by adding more sharpening!
This would end in glowing edges and "stairs" on diagonals to be recognised later on TV.

Better deleting this line:
And for shure in this case of MA deleting...
!!!
As this would cause double spatioTemporal Filtering and of cource at 352x288 details would be over-plained. (cause StmedianFilter and TemporalSoften will do the filtering work enough)

Thanks Incredible.