|
|||||||||
|
Thread Tools | Search this Thread |
May 7th, 2009, 10:09 AM | #1 |
Regular Crew
Join Date: May 2009
Location: Manhattan
Posts: 131
|
Neo Scene - problematic pulldown
I am testing Neo Scene with the trial and it works as specified for the most part. However, in my tests with a number of original, raw AVCHD files from my HF-100 (Canon), I have noticed that at the very beginning of every resulting clip, there are a few frames where the pulldown isn't properly removed. Frames 2, 3, 5 and 6 are a blend of neighbouring fields from the original 60i source. It looks like the pulldown cadence isn't correctly identified at the beginning, but is properly detected later. The rest of the video, no matter how long, looks fine.
Am I the only one noticing this issue? |
May 7th, 2009, 02:25 PM | #2 |
Major Player
Join Date: Sep 2003
Location: Solana Beach, CA
Posts: 853
|
There are no pulldown flags in the stream, so cadence is detected by real-time field matching. It takes a few frames to lock the pulldown cadence, so the operation you're experiencing is correct.
|
May 8th, 2009, 08:28 AM | #3 |
Regular Crew
Join Date: May 2009
Location: Manhattan
Posts: 131
|
Makes sense
Thanks for a quick answer.
I suspected as much. The final result looks great. So essentially, all I need to do is keep this in mind when shooting and allow an extra half a second before the action for Neo Scene to figure out the cadence, then chop that part off in the edit. Now, will Neo Scene do its magic on a finished telecined project? For example, if I edit some 24p material that was encoded in 60i (like these AVCHD camocorders) in FCE, iMovie or something that doesn't support 24p timeline; then ask Neo Scene to transcode my finished QT Reference Movie into AIC or ProRes with pulldown removed, would it do it? In other words, does it accept AIC as input format? Also, if it does, would it detect break in cadence between different shots in the movie? I've tried JES Deinterlacer, which does a decent job of pulldown removal on individual QuickTime clips. It won't do AVCHD, though, nor can it detect breaks in pulldown cadence. |
May 8th, 2009, 08:44 AM | #4 |
Major Player
Join Date: Sep 2003
Location: Solana Beach, CA
Posts: 853
|
Yes, Neo Scene should track cadence changes, but you'll end up with a few frames at each boundary point where the cadence is broken by editing renders. These broken cadence points will require "hunting" by the telecine removal algorithm similar to that you noticed on startup. If practical it's always best to remove pulldown on clips before editing. Editing of 24p material with pulldown will always result in broken cadences.
Neo Scene can export to ProRes HQ but not to AIC. |
May 8th, 2009, 04:31 PM | #5 |
Regular Crew
Join Date: May 2009
Location: Manhattan
Posts: 131
|
Sounds good
That sounds quite reasonable. Occasionally, I'll have material that originally came from an AVCHD device that shot it in 24p and wrapped it into 60i. This would be 'Events' in iMovie 09, which are in AIC. Since final edit of such a project is already done and iMovie already has a reference QuickTime file, it would be much easier to give that QT file to Neo Scene to remove the pulldown as best as it can, rather than removing it on every single source clip and replacing those source clips, possibly re-doing the entire edit, just to avoid a few blurred frames. Chances are, some of the edit points might fall along the same cadence. The final result, in ProRes, can then be exported into whatever delivery format is necessary.
Looks like I'll be buying this Neo Scene in a few days (as soon as my trial expires...). Looking at it from the Cineform's perspective, may God bless AVCHD camcorder manufacturers for shoehorning video captured at 24p into 60i. If they had some foresight to capture 24p as 24p, Cineform would have had a lot fewer Neo Scene customers. I must say, though, the idea of transcoding AVHCD into 4:2:2 colour space makes Neo Scene worth it even if pulldown removal wasn't the main reason for buying it. Not all projects can be done properly in iMovie; some of them really need FCP, where 4:2:2 shines. |
May 9th, 2009, 12:42 AM | #6 |
Major Player
Join Date: Sep 2003
Location: Solana Beach, CA
Posts: 853
|
Yes, it seems there always has been - and I suspect always will be - a need for products to fix ideas that are ill-conceived by others, such as 24p in 60i, or 4:2:0, or interlace, or 960x720, or long-GOP editing, ....
|
| ||||||
|
|