@ Boulder
yes, the same kernel params of Bob(x,x) are used like in BicubicXXX(x,x).

But it would be interesting if someone enlightens me in these maths of those two values used in Bicubic to determine the interpolation! :)
As there are no explicit infos at the avisynth page.

Also I would like to know how to set rigth a convolution matrix in GeneralConvolution() as IF you/me would get some skills of these convolution-matrix then that filter gots VERY useful as so many nice things could be done (masking/blurring/swapping/shifting/sharpening) with that.
Avisynth Docs do point on this explanation regarding to Convolution matrixes:
http://www.gamedev.net/reference/pro...proc/page2.asp
*wow*!!

@ Viking

If u use FitCD, KEEP IN MIND to crop at MOD2 in a BOBBED! (temprarely de-weaved to full fieldrate) state and resize to MOD 16. As by this the YV12 stream is temporarely not in an interlaced state.

Those b and c values are really a mystery..I also wonder why soft bicubic a=b=1/3 has been visually most pleasing in the test the docs mention. One probably has to dig up some research papers if he's to find out the real information. Everyone seems to have copied the same sentence about blurring and ringing constants :wink:

I wanted to define a definitive pattern for me to capture, therefore I have many VHS here, and I am looking for the best configuration before converting all, then it would like an opinion, for me to reach a final conclusion...

then, considering that I have a capture card Prolink PixelView PlayTV Pro (bt878), using BTWinCap WDM v5.8 opensource driver, Huffyuv v2.1.1 CCE SP-Patch v0.2.5 codec ( Click here to see the configuration, please ), YUY2 colour format, VirtualVCR 2.69 software to capture, 29.970 FPS (reminding that I live in the Brazil, and the system here is NTSC(PAL_M)), VHS source (signal Video Composite (RCA)).

Then, the best resolution (and correct, of course) and definitive for me to capture (and is the destiny dvd) is it 704x480, correct?

the script to codify interlaced correctly, is:

about the lines in red, is my doubt if it is correct for captures done of VHS (TFF) ?

(in case I will codify a musical show, i change "kenelbob order" to 0 and selectevery to 4,0,3 ? )

if I will capture to transform for XviD later (to leave FS and to look in PC) would I capture in which resolution? which the best script for deinterlace ?

thank you very much for all :wink:

I would use predict left or gradient in HuffYUV. It's much faster to decompress :wink:


If your input clip's field order is top field first, keep the script as it is. If the field order is bottom field first, do the changes you suggested. The field order doesn't depend on the source material itself, it depends on your capture card and drivers.

Also remember to convert the colorspace before reinterlacing (put the convert line before the SeparateFields() line).

See the thread which contains my guide to treating interlaced sources. There are some posts regarding different deinterlacers.

What comes to capture resolutions, I'm afraid I can't help you there. I capture happily with my non-scaling PVR-250 at 720x576 and do ITU compliant resizing.

You should also make sure that the source is truly interlaced and not telecined. If it's telecined, you can IVTC it and encode at 23.976fps as progressive.