On Thu, Nov 21, 2002 at 07:31:01PM -0600, Billy Biggs wrote: > > Careful, some tvtime filters are based on showing data from old > fields, when motion isn't detected. You might just be seeing artifacts > from poor algorithms, but I kinda doubt it based on those weird errors > you're getting. How does the effect change when using greedy vs > greedy2frame vs weave vs linear? Oooops! Major oops. I don't for one second think this is a tvtime problem, but rather feel strongly that it's a hardware (mismatched with software (driver) perhaps?) problem. I only offered up tvtime as a comparison of software that uses both fields as opposed to xawtv which only uses one. The problem only happens when both fields are being acquired/used. > I don't think it's an issue with showing both fields at once, but > tvtime does keep, if it can get them, three buffers at all time while > the card is writing into the fourth one (which is why we like to have > you use gbuffers=4). These old buffers are used for our reference > frames. I will try adding gbuffers=4. I could try recording a bit with mp1e and see how well the "stangeness" is preserved through the compression. > > I also see errors such as: > > Nov 21 19:56:25 pc kernel: bttv0: SCERR @ 16dd6000,bits: HSYNC OFLOW FBUS FDSR SCERR* > > This does freak me out. I have not the foggiest what it means. Any ideas? b. -- Brian J. Murrell
Attachment:
pgpR9DN5mqCxM.pgp
Description: PGP signature
