Getting the best results downconverting HDV to SD

Camera: Canon XH-A1 (footage 30f, NTSC)
NLE: Final Cut Pro 5.1.4, Compressor 2, DVD Studio Pro 4

After months of simply exporting my HDV sequence to Compressor and getting poor results (from the File Menu Selecting "Export" - "Using Compressor") I decided to search the forum for an alternative technique that would yield better results. I read a number of posts and found the overwhelming consensus to be to convert your HDV Sequence to "Uncompressed 10-Bit 4:2:2", and then to import the uncompressed file into Compressor for encoding to Mpeg-2. Although this information is scattered throughout the forum I thought it might help someone to consolidate some of it and share my experience to date.

One of the post that I read (http://www.dvinfo.net/conf/showthread.php?t=102079) explained the steps to be taken to convert my HDV sequence to "Uncompressed 10-Bit 4:2:2".

1) Select HDV sequence from "Browser" or "Timeline".
2) Go to the File Menu and select "Export", then "Using Quicktime Conversion".
3) Under "Format" select "Quicktime Movie", and click on the "Options" button to its right.
4) In the "Movie Settings" window go to the "Video" segment and click on the "Settings" button.
5) Once the "Standard Video Compression Settings" window opens , select "Uncompressed 10-Bit 4:2:2" from the "Compression Type" drop down menu and click "OK".
6) After clicking "OK" the "Standard Video Compression Settings" window closes and you return to the "Movie Settings" window. Now go to the "Video" segment of the window and click on the "Size" button.
7) In the "Export Size Settings" window select the dimensions of the "Uncompressed 10-Bit 4:2:2" you would like from the "Dimensions:" drop down menu.
8) Then check the box that is labeled "Preserve aspect ratio using:" and click "OK".
9) After clicking "OK" the "Export Size Settings" window closes and you return to the "Movie Settings" window. Click "OK" to close the "Movie Settings" window and finally click "OK" in the "Save" window to export the "Uncompressed 10-Bit 4:2:2" movie file.
10) Once the Quicktime Movie has been saved, import the "Uncompressed 10-Bit 4:2:2" movie file into Compressor for encoding to Mpeg-2.

At first the task seemed simple enough, but for some reason I almost immediately had problems. After clicking the "Size" button in step #6 I found myself unsure of what dimension to make the file ("NTSC 720x480 16:9", "NTSC 720x486 16:9", or possibly "768x576 SD"). The posts original author was editing in PAL so his settings were not going to help. In addition, I was unsure of what setting from the drop down menu to the right of the "Preserve aspect ratio using:" check box ("Letterbox", "Crop", "Fit within dimensions"). I suspected "Letterbox", but was unsure.

I tried exporting every combination I could think of but everything that was saved was distorted when played back in Quicktime. When I looked closer at the post I saw that the author states that the exported files will appear distorted if viewed using Quicktime, but when imported into DVD Studio Pro into a 16:9 SD timeline the file will be displayed correctly. I tried this but the results were still distorted and inconsistent (I later discovered that my settings were off in DVD Studio Pro and could have possibly caused the problem).

At this point I decided to export my HDV sequence directly to Compressor and use Compressor to create the "Uncompressed 10-Bit 4:2:2" movie file using "Advanced Format Conversions" from the "Settings" menu. The results were terrible, so I gave up on this approach.

Finally, I re-read the post describing the process and saw that it might also be possible to create a new "Uncompressed 10-Bit 4:2:2" sequence in Final Cut by cutting and pasting my HDV sequence into it, rendering the sequence, and using the File Menu to "Export" - "Quicktime Movie. This technique has work very well for me so far and the results are terrific.

Below I have listed the settings I have used to convert my HDV 1080p30 footage to "Uncompressed 10-Bit 4:2:2" sequence in Final Cut, how I used Compressor to encode to Mpeg-2, and finally how I used DVD Studio Pro to "Build and Format" the final SD-DVD.

1) In your HDV project create a new sequence.
2) Control click the new sequence in the "Browser" and go down to "Settings" from the drop down menu.
3) The "Sequence Settings" window opens with the name of the sequence at the top. Under the "General" tab just below the "Name" of the sequence are a number of fields and drop down menus. The first is "Frame Size:", then "Pixel Aspect Ratio:", an "Anamorphic 16:9" check box, "Field Dominance", "Editing Timebase", and Quicktime Video and Audio settings.

a) Frame Size: NTSC DV 3:2 (720x480)
b) Pixel Aspect Ratio: NTSC CCIR - 601 / DV (720x480) and check "Anamorphic 16:9"
c) Field Dominance: None (since I shot progressive)
d) Editing Timebase: 29.97 (since I shot at 30 FPS)
e) Under "Quicktime Video Settings" I have the "Compressor:" set to "Uncompressed 10-bit 4:2:2" and the "Quality:" at 100%.
f) I left the Audio Settings untouched and clicked "OK".

4) Then I simply cut and pasted my HDV Sequence into the new "Uncompressed 10-bit 4:2:2" sequence, rendered it, and exported as a "Quicktime Movie".
5) I then imported the "Uncompressed 10-bit 4:2:2" Quicktime Movie file into Compressor, and encoded using my best settings to Mpeg-2 and Aiff. I posted my best settings at the bottom of the following link if you are interested: http://dvinfo.net/conf/showthread.php?t=86827
6) Finally, I imported the resulting files into DVD Studio Pro and "Built and Formatted" the disc.

The results are great. The jaggies, faded colors, and soft look of the SD-DVD that I originally burned using only Compressor paled in comparison to the smooth edged, vibrant colored, and sharp image I got from the SD-DVD I burned from the "Uncompressed 10-bit 4:2:2" file.

Hope this helps someone,
Hugh

Thanks and it dose help.

After finding this thru searching, I want to say THANK YOU SO MUCH for this guide Hugh. Made a huge difference.

If anything has changed in the last year for an even more simpler approach with even better quality and less file size, would someone be kind enough to link me?

Thanks again Hugh!

Hugh,

Thanks a lot for sharing what you've found. I am wondering: is there anybody here who can explain what goes on under the hood if I just select "Export - Using Compressor" to create an MPEG-2 encoding? The way I understand it, FCP will render a frame at a time and hand each frame off to Compressor. What I don't know is what encoding is used for this hand-off of each frame. Uncompressed would make sense, since storage space is not a concern in this case, but that's probably not what happens - otherwise the quality difference shouldn't be so drastic.

Just trying to understand what we are comparing these more elaborate workflows against.

Thanks again,

Martin

Hugh,

Thanks a lot for sharing what you've found. I am wondering: is there anybody here who can explain what goes on under the hood if I just select "Export - Using Compressor" to create an MPEG-2 encoding? The way I understand it, FCP will render a frame at a time and hand each frame off to Compressor. What I don't know is what encoding is used for this hand-off of each frame. Uncompressed would make sense, since storage space is not a concern in this case, but that's probably not what happens - otherwise the quality difference shouldn't be so drastic.

Just trying to understand what we are comparing these more elaborate workflows against.

Thanks again,

Martin

Good question, fellow Iowan; I use Export>Compressor constantly for 60i HDV footage and it looks great, so it would have to be some stunning results for me to justify the time investment in this proposed workflow. I might try a little A/B test.

I had similar issues, but I gave up on 30p entirely. I use exclusively 24p (720p HDV JVC 24fps anyway) had a LOT better results than 720p 30fps. So you might think about 24p experiments with your Canon. Getting good motion out of 30p into a 24p timeline took me 22 minutes per minute of footage to render. 30p to 60i (59.97 or what ever it is) took about 4 minutes per minute of footage, and 24p took maybe 2 minutes of render time per minute of footage and had the best results displayed on a LCD or Plasma. This is of course refering to playing on a DVD. I had excellent results from Canon's 24f mode. The render times were about twice as long as JVC as I expected. I'm running a Mac with Dual Quad Core 8x3ghz with 6 gigs of memory and running OLD Final Cut Studio 1 that doesn't even access the top 4 cores, just first 4 cores.

My issues with 30p (JVC 720p HDV that is) when converting to DVD at 30p (DVD NTSC standard is 24p/60i native, so I think 30p comes out as 24p + 6 extra interlaced frames per second mingling around) was giving still objects a inconsestant flexing look...probably the interlaced frames hitting the LCD) I did get some great conversions from 30p to a 24p DVD, but at a huge cost of render time. 22:1!!! I'll never do that again. The Better setting resulted in the usuall 30P to 24P jumpy conversion. I hate interlaced NTSC on a LCD or Plasma so I only tried that once.

So maybe try 24p again and check your render times and final DVD results or go 50/60i.

Capture HDV • Edit HDV • output using Export/Quicktime Conversion/Settings/Compression set to Prores422 and leave everything else other than to tick "interlace footage."
Drop into Compressor and create a custom setting to output 16:9 by using the existing "best for 90 min" and changing a few settings to output 16:9. Leave all else set to standard.
Pull the files into DVDSP and again change the "output" under the General tab to 16:9 for the "display mode."
You should see a marked improvement.
That said. Rendering takes time. Realtime engine sorta works, depends on how you like to work.

I've always just used Export/Quicktime Movie/ProRes 1440x1080 24p after editing in HDV24p. Drag them into compressor and compressor automatically senses that its anamorphic and progressive.

Good question, fellow Iowan; I use Export>Compressor constantly for 60i HDV footage and it looks great, so it would have to be some stunning results for me to justify the time investment in this proposed workflow. I might try a little A/B test.

Ben, did you ever try that A/B test?
When you export into compressor for Standard DVD encoding, what settings are you using?

Actually, I haven't tried it; I've been too busy producing and had more than adequate results with Compressor's *stock* settings that I just haven't had any reason to.

Actually, I haven't tried it; I've been too busy producing and had more than adequate results with Compressor's *stock* settings that I just haven't had any reason to.

I wanted to apologize for not having responded to any of the questions people have asked concerning my original post. There must be something wrong with my notification settings.

Anyway, it has been a while since I posted this workflow. I have not bothered to look for an updated process, since the one I described above still yields great results. I originally used compressor to compress my HDV, but kept finding the results rather inconsistent. I decided to follow a couple suggestions I found on this forum and haven't looked back.

I performed a couple of A-B tests, and there was no question that there was a massive improvement to the DVDs encoded using the "uncompressed" workflow. Whatever Compressor was doing to my HDV footage as it downrezed and encoded it to mpeg 2 was resulting in jaggies, and a loss of color saturation and sharpness. I wish I could explain why dropping your HDV sequence into an Uncompressed 10-bit sequence, and then exporting to Compressor produces better results, but my workflow was just a simplification of techniques I found on this forum. The results spoke for themselves.

Give the workflow a try. It only takes an hour or so to put together an A-B test, and I am sure that the results will convince you that Compressor needs a little help converting your HDV to SD. If you are using the latest version of Compressor (I was using Final Cut Studio 1) and your A-B tests reveal little or no difference in output quality please let us know. It would be awesome if Apple's update to Compressor allows it to produce better results than it previously did, but knowing Apple (there is still no Blu-ray authoring support for DVD Studio Pro) I have my doubts.

Thanks,
Hugh

The reason I'm skeptical is not just that I'm content with the Compressor settings (as are my clients). I've been doing this long enough to be suspicious of any process that seems arbitrarily over-engineered. Some screen grabs and scopes comparisons might help make the case.

The reason I'm skeptical is not just that I'm content with the Compressor settings (as are my clients). I've been doing this long enough to be suspicious of any process that seems arbitrarily over-engineered. Some screen grabs and scopes comparisons might help make the case.

All that is needed is a simple side by side comparison. If you are happy with the results you are getting then don't bother. When I wrote the first post I was using Final Cut Studio 1, so perhaps the new version of Compressor corrects the problems I was having.

Thanks,
Hugh

I just tried your method and I have to say I can't really notice a difference. If anything I almost think the version from the uncompressed looks worse. On the positive side it look a LOT less to render.

I'm going to have to do a longer test to know for sure..

Also, if you want to maximize image quality then don't use AIFF but encode your audio to Dolby Ac3; it will use less space on the disc and leave more room in your precious data rate for a higher-quality video encode.

I'm in the process of learning FCP(5.1). I shot a wedding in HD with an xha1, edited it and it runs about 45 minutes. I should have tried a short clip first but, unfortunately, I didn't. I'm learning you know. I went through Hughs' workflow and after 12 hours of processing in compressor, it stopped at 88% complete. Looking at what was processed as QT looked very good but NO audio. What have I missed and should this take so long to process? The processed file was 1.2GB. Thank you and Happy Thanksgiving to all.

To resize from HD to SD first establish a new sequence and choose the desired prefs,
for SD, e.g. PAL (in my case) and codec (preferred MJEPG A or B).
After activating that new sequence drop the HD sequence into that SD sequence as if it were a single clip, as you do with nesting sequences.
Re-render and export as QT or go directly into compressor. Done.

Alternative: Export QT movie (Can by "related movie", no need to spend HDD estate).
Open that QT in MPEGStreamClip.
Resize in MPEGStreamClip while watching the preview window and the progressive bar.
Save SD file. Open in Compressor. Done.

Personally, I prefer alternative 2, because I can either do de-interlacing either in MPSC or with DVMaker or RS-Fieldskit. DVMaker produces a wonderful motion blur out of the two interlace frames. This de-interlacing yields to far better down scaling of interlaced source material.

Hope this helps. P.

Thanks Peter for your response. I've been out of pocket for a few days. I'll check that out. Jerry

All that is needed is a simple side by side comparison. If you are happy with the results you are getting then don't bother. When I wrote the first post I was using Final Cut Studio 1, so perhaps the new version of Compressor corrects the problems I was having.

Thanks,
Hugh

So I finished a 60min. project and leveled all horizontal planes in landscape that needed it, ones that would benefit from matching to other cuts in the project. Using image-wireframe in the canvas I had a blast reworking many of these clips. The movie really took on a slicker look, I was happy as a clam even if I had to spend extra time to do it.

Ok, so everything worked out fine except one thing. Every clip that I used -image-wireframe on to straighten the horizon is now blown up to make only the center 40% of the picture show in the movie. In other words, the exact amount I blow up the frame is not exactly what is now blown up. There is some issue with aspect ration in just these tweaked clips.

I tried nesting the entire sequence and that did not work either. Something about dropping clips that have had wire-frame work into new different type sequences is not a good thing.

The only thing I can think of now is to go back and re-frame every shot that has this blown up effect, do it all again.

But I do like this was of going HDV to SD. I just hate what has happened to my work I spent some serious time on.

Any ideas would be welcome. But I might just be SOL I know.

Do you know about using image-wireframe to straighten out and level frames? Beautiful featrure in FCP, but ruined by this method of HDV to SD I'm fraid?