-
HDV Downconvert through Compressor – issues with high detail areas
I posted this in the HDV Forum, but thought this would be a good place for it too.
In your experience, what is the best way to go from an HDV sequence in FCP to an SD tape master? Sorry if this post gets a bit long, but I want to explain my current workflow.
Our acquisition format is HDV 1080i60. We capture in native HDV in order to retain the original timecode data from the source tape. We edit in a native HDV sequence with the render codec set to ProRes422 to avoid conforming. Our delivery format is an SD DVCAM master that we send out to have encoded for DVD replication.
Being that our delivery format is SD, we originally edited the HDV footage in an NTSC sequence. However, every HDV clip that is dropped in has to be rendered even before applying any effects, filters, or transitions – obviously because the settings of the clip and the sequence don’t match. To avoid this, we edit in an HDV sequence. We capture in HDV rather than downconverting at the deck so we can do more pan and scan and reframing – if you saw how bad a lot of the source is shot, you would understand why!
After the editing on a segment is complete, it is exported through Compressor – typically in a batch of 3 to 10 segments. Each segment is downconverted to letterboxed NTSC. Everything looks great except for areas of high detail – such as trees, bushes, and fields. Any camera movement causes these areas to jitter. There is no problem in these areas before exporting through Compressor.
I have experimented with different settings in Compressor to achieve the best look possible and eliminate the need to render once the output is placed into the master NTSC sequence for output to tape. The settings below do a great job with the exception of the high detail areas.
Video Encoder
Format: QT
Width: 720
Height: 480
Pixel aspect ratio: NTSC CCIR 601/DV
Crop: None
Padding: None
Frame rate: (100% of source)
Frame Controls On:
Retiming: Fast (Nearest Frame)
Resize Filter: Best (Statistical Prediction)
Deinterlace Filter: Better (Motion Adaptive)
Adaptive Details: On
Antialias: 0
Detail Level: 0
Field Output: Progressive
Codec Type: DV/DVCPRO – NTSC
Multi-pass: Off, frame reorder: Off
Scan Mode: Interlaced
Aspect Ratio: 4:3
Pixel depth: 24
Spatial quality: 100
Min. Spatial quality: 0
Temporal quality: 0
Min. temporal quality: 0
Letterbox
Type: Scale
Position: Center
Output: 16×9 1.78:1
Manual: 100.000
Background: r:0.0000, g:0.0000, b:0.0000Basically what I have experimented with are different settings for the Frame Controls section. They seem to have a huge impact on the output. If you notice anything out of the ordinary, let me know. I have tried changing the Deinterlace Filter to Best, but it takes about 25-30 seconds per frame to render. I have also tried turning up the Anit-alias and Details Level, but again it takes FOREVER! But I’m wondering if one of those settings would make a difference.
My question is this: what settings should I change to reduce or better yet eliminate the issues with the high details? Is there a better way to go about downconverting than what I described above?
Thanks!
MikeMike Johnson
Final Cut Pro Editor
Drury Outdoors
http://www.druryoutdoors.com -
7 Replies
-
This might be a shot in the dark an not make much sense but it’s worth a try; try using a lower quality resize method and a higher quality retiming method.
__________________________________________________________________
Please post back saying what solved your problem. It could help others, and saying ‘thanks’ is free! -
I gave it a try with a few different combinations of what you suggested. I tried fast resize and best rate conversion; best resize and best rate conversion; and best resize and better rate conversion.
None of these combinations changed the jitter in high detail areas. The rate conversion did not seem to make any difference, while the resize setting made a big difference in the overall quality of the image.
Mike Johnson
Final Cut Pro Editor
Drury Outdoors
http://www.druryoutdoors.com -
I use the Best settings for all Frame Controls. But why are you going through the step of outputting to DVCam then to a replicator? Why aren’t you just outputting to MPEG-2 while you’re at it? If you have others authoring the DVD you can give them the MPEG-2 files. I think you’ll see a marked improvement in quality.
Ed -
I just finished exporting with best settings for all frame controls and it took 4 min to export 15 frames.
We print a DVCAM master that gets sent to an authoring house. They compress to MPEG-2, build the DVD elements, and print a check disc. If the check disc is good it is sent to the replicator. Why we don’t do the MPEG-2 compression and DVD authoring, I don’t have a clue. I’ve done it many times in other positions. But if I had to guess, its probably because no one in the company understands compression or DVD authoring and they don’t want to learn. That’s the way its been done since they made the transition to DVD. I don’t know. But it sure would be a lot easier.
Mike Johnson
Final Cut Pro Editor
Drury Outdoors
http://www.druryoutdoors.com -
Yeah, Compressor is painfully slow. What Mac are you using? Do you know about QMaster? It comes with Compressor and it allows you to assign all the cores (processors) to compressing in a more efficient way. Do a search here and in the FCP forum. There have been a lot of discussions about it.
Ed -
Another thought.
You could try a two stage process – first to 720p 30, then 480p 29.97, boring but it might get round the problem. My only other thought is to apply a very mild blur/noise reduction but that would be a last resort.
I’d really love to see this ‘jitter’ effect before I made a final judgement.
Most pro authoring houses can produce better/more efficient output than compressor is capable of (unless you are using the CinemaCraft plug-in) but then again most pro authoring houses would not consider DVCAM a good enough source format.
__________________________________________________________________
Please post back saying what solved your problem. It could help others, and saying ‘thanks’ is free! -
I’ve figured out that the jitter is caused by the resizing in Compressor. I dropped an HDV clip into an NTSC sequence and after rendering there were no problems. I then exported a test through Compressor without resizing to 16:9 Letterbox NTSC. No problems here either. So I decided to export using ProRes422 at 1920×1080. The export through Compressor goes a lot faster than if I exported a QT Conversion from FCP – and I can submit a whole batch rather than do one at a time. I still have to render everything once I place it in the master timeline, but I will let that run over the weekend.
Mike Johnson
Final Cut Pro Editor
Drury Outdoors
http://www.druryoutdoors.com
Reply to this Discussion! Login or Sign Up