Smooth line/dot crawl and GOP frame artifacts by morphing frames?

[Roland R. kahlenberg]

Dang! With the left image, I’m only seeing issues along the bottom and left edges. The black dotted areas look fine. Are we on the same page?

The effect you applied seems to have solved your initial problem but inadvertently created another.

Use a garbage mask in FCP to isolate just the areas that you want the plugin to work within. Then superimpose this over the orginal image. Commonly found NLEs have a limited garbage mask (usually a maximum of 8 vertices) facility which seems doable given your specific issues.

Additionally, you mentioned GOPs adding to your initial issues but from the onset, you should never render intermediate assets to a CODEC that applies inter-frame compression which relies on GOPs for its compression algorithm. This anomaly is compounded with the low rez texture(s) that the imagery contains.

HTH
RoRK

Sell your AEPs with broadcastGEMs’ DVD series of templates. Click here for more

[Bill Russell]

>>Dang! With the left image, I’m only seeing issues along the bottom and left edges. The black dotted areas look fine. Are we on the same page?<< Hi Roland, thanks for thinking this through with me! Some more info: Keep in mind that what you are seeing here is an extreme close-up of my full animation. This animation, graciously provided, is not something I created or that can be redone, but it is central to our HD documentary, appearing often. The artifacts are not a good thing the way we have to use it. The graphic was created in India from composite 3D maps, to track a historical route through the mountains. The animation was rendered to a GOP frame codec -- yup. As the 3D image moves, the 3D texture mapping jumps to a new state from time to time -- that is what you are seeing with the large "hour glass" shaped below and right of center: It jumps in the left frame, but "morphs" in the right frame. This sample movie is just a tiny slice in both space and time, to show how the applied inter-frame morphing result "erases" these jumps. Unfortunately, there is interlace line crawl (which I don't understand, since it was rendered 23.973... but anyway, it is not included in the CU slice here, so you can't see it) and dot crawl (which, as you say, you can see along left edge and bottom -- but it is actually in places all over the whole frame not shown). Also, in the first frame at the very top you see a hint of the GOP frame artifact. Those blobs appear at regular intervals over large areas of the frame. The "morphing" helps to hide them because they blob in and out of existence instead of jumping in then breaking up. >>The effect you applied seems to have solved your initial problem but inadvertently created another.<< Well, I think you may be referring to the white "spots" you see kind of floating on their own in the morphed result? Full size at full speed that stuff is not noticeable at all, and the whole thing really is an improvement on a 42'' 1080p display. I haven't really experience undesirable consequences -- for that I'm quite giddy -- just haven't been able to achieve as much of the effect as I need to fully solve all the problems. For instance, the line crawl (which does not appear in this CU sample). It is better, but still there as a subtler "running water ripple" where once there was a jaggy crawl. So I had used the 8 point garbage matte in FCP (I'm familiar with it) to add a blurred layer over the line crawl to hide it yet more, then "coated" the whole frame with light small grain to hide the blur. That's as close as I got. Piling these TMTS noise reducers on top of each other is such a kludge and offers no control, but it seems there must be a technique or tool that can accomplish this result, hmm. So that's my angle here! Thank you!!!!! “THE LOST SKELETON OF CADAVRA” –

And more…