Fri Jan 06, 2017 4:14 pm
Hi Dave, so I just wanted to follow up on your post and Tony’s response. I think there was a little misunderstanding about your final question about the cadence anomalies and I wanted to clear that up. If I’m reading it correctly you’re no longer asking about the clean cadence function but want to understand the cadence detection capabilities in the Express and AV. I hope I’m on the right path there.
While it’s true that the Express and AV do not have the clean cadence function, they are running very similar algorithms based on the same IP for cadence detection, deinterlacing, etc. There are a couple of minor changes we’ve done for the new hw architecture and to support the higher bandwidths of 4K but the quality of conversions, cadence detection, etc are all similar or better.
Also, I just want to make sure everyone is on the same page with what we mean by clean cadence, etc. The clean cadence conversions in the 2D and 3D run for interlaced 59.94 inputs to interlaced 59.94 outputs and it does a 2 step pulldown removal and reinsertion all in one pass. During the removal it handles broken cadences, etc to get to an internal 23.976 before reinserting a clean cadence out at 59.94. You could perform the same conversion in 2 passes (or 2 boxes) in the AV or Express by doing pulldown removal in 1 and insertion on the other.
As to cadence anomalies, all the units (2D, 3D, Express and AV) can handle broken cadences, there’s no magic formula for what it will do and depending on how broken it is, we may create skips, dups or synthetic fields for orphans depending on where scene cuts fall etc. If the cadence isn't correct we have to do something, the important thing is that we are doing something on purpose based on knowing what the cadence is and what the fields contain. What we guarantee is that we will not blend across scene cuts and that the algorithms are not locked into looking for just a 3:2 cadence, we’re looking for fields that come from the same film frame to match to ensure you get clean results. We look at every pixel in every frame across multiple frames to make sure we find the correct match. Even if the content is mixed and there is video we will "fall back" to do a frame rate conversion to create a frame at the 23.976 frame rate, so that you will get a clean result. All automatic, all running all the time.
In your case with the random cadence switch, we will handle that, sometimes it will handle it perfectly, depending on how often it switches you could get a skip, but we will handle the anomaly and are not thrown off by changes to the cadence, we will adjust to it and continue to do the pulldown removal. I would encourage you to try it out, that is going to be the best for you to see the benefits of this technology.
Hope that makes sense and helps and let me know if you have more questions.
- Ray
