Now thats a special one for phil. (and other shurely too).

I know that you used ffdshow for previewing avisynth functionalities on the fly. Ok, I got my own workout using Vdub for that, BUT .....

..I tested the last days (and before my vacations) the internal ffvfw (capturing via Vdub which uses vfw) /ffdshow filter options.

I got INCREDIBLE outputs! Compared to classical avisynth way.

The workout was, that I choosed in Vdub as capture codec the "ffvfw mpeg-4 codec". I do choose HuffYUV or MJPEG as encoding codec. Now I did play in "Image Processing" under NR using some of these optional filters like HQ3dn, akf mencoder or others and the result when capturing a grainy Source was MUCH better than when using classical avisynth ways.
The static dot-phenomenom on moving surfaces was much more less and the details where kept very very good.

Intention of this Thread:
Maybe we could test even if not capturing, to let handle ffdshow the avs files BUT not for avs filter checking but pre-frameserving using some of these implementated filters like deblocking or deringing or deinterlacing or denoising and compare.

I dont have any blocky test-mpeg4 sources here so you (phil) could maybe participate (or you already did tests on your own??)

The captures (using the internal smart! denoiser and the xsharpen) are WONDERFUL and do end up "like" a good broadcasted DVB Stream. Cracks would hit me now as I do call it like that, but its just called as that to understand.

Also the speed is "wow" as I can even use these NR and xsharpen when capturing on the fly! Also if I would EVEN use that CPU consumpting mpeg4/XVID codec for capturing.

Or someone with C++ experiences could get mencoder libavcodecs internal filter routines and do a avisyth dll from some ones of these?? SansGrip maybe

I'm only wondered about the fact that will claim to use directshowsource and not avisource or mpeg2source. With all potential problems inherent to that. But that won't prevent to do some tests of course.

No, for the reason exposed above. I have a lot os DS filters on my PC that are used when playing an avi and that I don't want to see in the render chain during an encoding project.
Doing a graph under graphedit is always possible but how will you explain that to noobs ?

That can be a good idea as wwe already saw with mencoder than filters of libavcodec are really good.

Well in case of a generic YV12 output on my machine avs is handled by XVID codec. SO why dont we try to let ffdshow/ffvfw mpeg4 codec doing that job, means all special avs functions will be done in the script and the NR filtering in the "frameserver" of ffdshow/ffvfw.

and in case of YUY2 output in Avisynth script it should also be possible to do a system routing to ffdshow/ffvfw

That is correct. In case of avs script that's right that the render chain does not contain other DS filters for the video than ffdshow.

I declared ffdshow as filter to use for any raw stream. That is, in other words, any time !
For that just go under the "codec" tab of ffdshow, set "disable" for all codecs (xvid, divx, ...) and set "raw video" to "all suported".

Ok! Ill do that .....