Announcement

Collapse
No announcement yet.

need help with ADVC 300

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #16
    The "thick bands" is related to tearing and is usually a decoding latency problem. It's common when the player doesn't use buffering or overlay for display. You shouldn't need to compensate for it when encoding unless frame-by-frame is also showing the problem.

    Comment


    • #17
      right well solved it although not the quiqest of results.

      after capturing with sony vegas, put it through tmpgenc xpress which you can then select the output source at the frame rate you want and also select PROGRESSIVE SCAN this will sort the issue out.

      Comment


      • #18
        Encoding as progressive will deinterlace the video, resulting in an effective framerate of 12.5 fps (you're in PAL, right?)...

        If the deinterlacing does frame-blending or its motion detection is sub-par, you can end up with artifacts similar to your original problem, except now the problem is not in the display end, but actually in the video.

        It's one of those things that isn't necessarily bad, but you should be aware of the effect of what you're doing.

        Comment


        • #19
          Wikipedia's article on interlaced video seems to imply that interlacing is bad and progressive is good.

          While this is the case for the same sampling rate (where a progressive datastream would hence require double the data bandwidth), it doesn't touch a lot on the temporal aspect.

          The temporal aspect deals with the perception of motion.

          For two streams at the same datarate - 30i and 15p, the 30i clip will appear to have smoother motion, because even though each sample is only half a frame (one field), there are more "snapshots in time" compared to the 15p clip.

          For those in PAL land, you probably don't notice as much at the movies since your TV is 25 fps, but for us NTSC folks, we notice the difference in motion between a 24 fps film and regular 29.97 fps TV.

          Horizontal panning in movies seems to "skip" or "jump" in movies compared to the same horizontal pan in regular TV video.

          To get a better idea, imagine a flip-book - a pad of paper that you flip through the pages quickly and see something move.

          For the same data rate:
          Interlaced video has poorer pictures (because each is half resolution), but more pages. So when you flip through, the motion is more fluid.

          Progressive video has better pictures, but half as many pages. So when you flip through, the motion tends to "jump" more from picture to picture.

          Again, for the same data rate:
          Turning progressive video into interlaced video is easy. Turning interlaced video into progressive video, is significantly more difficult. Because each pair of fields that create a frame have been sampled ("snapshot-ed") at different times, sophisticated motion detection and frame blending alogrithms must be used. After all, if in one field my arm is to my side and in the next field my arm is raised, if we blend those two fields, it'll look like I have 3 arms - one solid and two "ghost" arms! And trust me, I only have two arms.

          Okay, that's my brain dump for today.

          Comment


          • #20
            Deinterlaced or interlaced

            Originally posted by GrassValley_BH View Post
            Wikipedia's article on interlaced video seems to imply that interlacing is bad and progressive is good.

            While this is the case for the same sampling rate (where a progressive datastream would hence require double the data bandwidth), it doesn't touch a lot on the temporal aspect.

            The temporal aspect deals with the perception of motion.

            For two streams at the same datarate - 30i and 15p, the 30i clip will appear to have smoother motion, because even though each sample is only half a frame (one field), there are more "snapshots in time" compared to the 15p clip.

            For those in PAL land, you probably don't notice as much at the movies since your TV is 25 fps, but for us NTSC folks, we notice the difference in motion between a 24 fps film and regular 29.97 fps TV.

            Horizontal panning in movies seems to "skip" or "jump" in movies compared to the same horizontal pan in regular TV video.

            To get a better idea, imagine a flip-book - a pad of paper that you flip through the pages quickly and see something move.

            For the same data rate:
            Interlaced video has poorer pictures (because each is half resolution), but more pages. So when you flip through, the motion is more fluid.

            Progressive video has better pictures, but half as many pages. So when you flip through, the motion tends to "jump" more from picture to picture.

            Again, for the same data rate:
            Turning progressive video into interlaced video is easy. Turning interlaced video into progressive video, is significantly more difficult. Because each pair of fields that create a frame have been sampled ("snapshot-ed") at different times, sophisticated motion detection and frame blending alogrithms must be used. After all, if in one field my arm is to my side and in the next field my arm is raised, if we blend those two fields, it'll look like I have 3 arms - one solid and two "ghost" arms! And trust me, I only have two arms.

            Okay, that's my brain dump for today.
            I recently captured old films with my brand new ADVC300 with out of the box settings using Premiere. I then used QuickTime Pro to trim the beginning and end of the movie and converted to h.264 with 1500kb/s setings and the results look fine. My question is, is the result of this deinterlaced automatically or is it interlaced? I can not verify this with wmp since i got some error and I should reinstall it.

            I played it back thru VLC and the h.264 encoded actually looks better than the raw DV i got out from premiere (original has jagged lines sometimes when something is moving) Which leads me to the ollowing question. Is there some kind of automated deinterlacing in the advc300 that makes the lines appear, because the artifacts i see on the captured DV is like the "weaved" described in wikipedia.

            Then one more question. What is the best thing or codec to do/use to encompress the DV AVI file gotten from premiere raw that still maintains the interlacing so it can be used with things like DVD's

            Also if someone could suggest some easy to use app to cut he excess parts off the captured videos and maybe lip them to smaller pieces as well. I don't trust the premier in that too much since it ALWAYS does the annoying and time consuming RENDERING no matter what you do.

            Comment


            • #21
              On your encoding to H.264, it depends mostly where the deinterlacing happens. If Premiere is smart enough (or you force it to) deinterlace before it hits the encoder, the encoder gets progressive frames and doesn't have to worry about deinterlacing. On the other hand, if Premiere just sends the fields to the encoder, the encoder will have to reassemble them into frames.

              This is very similar to the whole concept behind standalone deinterlacers and scalers for plasma/LCD televisions and projectors. Most times the TV/projector has a mediocre deinterlacer and scaler, so people will put a dedicated deinterlacer/scaler
              in the signal chain before the output reaches the TV or projector. That way the TV/projector's deinterlacer/scaler is bypassed.

              Of another analogy...

              You can send me a URL to a non-English website and let me go run it through Babelfish and get some funky translation that may be 80-90% correct understanding, or you can translate it to English for me and tell me, so I get 95-100% correct understanding.


              There's no automatic deinterlacing in the ADVC300 that I'm aware of at least... There really shouldn't be, because DV is interlaced by spec.

              As for apps to cut/trim, VirtualDub does a good job. Nero Vision is a little weird with its cutting, but I have used it on occasion when I was in a bind.

              Comment

              Working...
              X