HD > DVD Image issue, barberpole effect on resize

[sullen]

Okay, I have been doing HD > DVD conversions for a while now and am currently having a problem that I am not sure if or how to get some correction with. I am trying to convert Neil Young's performances from Conan O'Brien from 1080i to DVD spec. The stream is a true interlaced stream with no IVTC from what I have researched. The issue comes at the resizing in combination with the interlacing I believe. The problem is centered on his guitar strings and small lines in the picture, sometimes in very white areas during movement or panning. Just these specific areas are effected. It almost looks like a barber pole where there is black lines going down the length of the guitar strings as there is a pan or he moves the guitar. It also happens when the camera pans and there are straight white lines close together. I am using AVISynth to do the resize. I have tried LanczosResize, Lanczos4Resize, straight BilinearResize, and am now encoding a clip using ConvertToYUY2 > InterlacedResize. I know the Lanczos resizers sharpen. I had a little bit of success with the BilinearResize, but it didn't completely eliminate the issue. I want to keep the sharpeness but try to eliminate this effect. Thanks for any help.

PS I am playing back the DVD files on my Pioneer DVD player using the Component out to a LCD HDTV.

[BJMoose]

Is your recording a TS (transport stream) file recorded from a cable box? If so, try googling for a free software called HDTV2DVD. It uses the ffmpeg encoder and is pretty accurate. It will do the downcovert to DVD in one step. Native bitrate is 7800, but can be altered to best fit a dvd-r or a +DL disc. This will produce both an mpeg2 file that can be demuxed with VOBedit or it also creates a Video_TS folder ready for burning if you don't need menus etc. I don't use this last method, but I understand the author has also included chapter points to be placed at 10 minute intervals. (BTW, I use VideoReDo to edit and save the final TS file.)

Prior to using this, I was using the latest standalone version of MainConcept's Encoder to downconvert my m2v file produce by demuxing the ts file in VOBedit. I set the bitrate at 7500 and found that using upper field first worked best. Progressive gave too soft a picture and Lorer field first had stutter problems. Hope this helps.

[edDV]

An LCD monitor creates its own problems with interlace to progressive conversion. See if the same problem exists with a CRT (interlace and progressive modes).

Fine detail + motion is the worst case for scaling. Most algorithms resort to blur during pixel motion to hide alias artifacts (too much resolution for sample space).

[sullen]

Thanks for the replies. I haven't yet sold the old tube TV so I will drag it into the living room and check out the picture with the stuff I have encoded. I will also look into the alternate types of encoding as mentioned above.