On the subject of HD downconversion...
 

I've seen some discussion here in another thread about downconversion and specifically, how to do it in FCP. Here's my two pennies...

I've found wildly varying results while trying to downconvert with the Apple FCP suite, depending on how one goes about it. Here's the three methods I've tried and the results I got:

1-Downconverting in an FCP timeline: This method is taking your cut HDV sequence, and copying and pasting it into an SD timeline, either DV or what have you. FCP will automatically scale your HDV clips to SD size (letterboxed), but any sort of output this way seems to net the worst results out of all the methods I've tried....even with the sequence "Motion quality" turned to "Best". Downconversion is very soft, with scaling artifacts.

2-Downconverting via Export: This is cutting in HDV, then exporting a Final Cut movie via File Export, then selecting an SD "type". This, for me, nets much better results than method number one, but affords no control whatsoever. Scaling artifacts are greatly reduced, almost perfect.

3-Downconverting via Compressor: This method works best for me. In this method, I either export a finished HDV timeline via compressor to 10bit uncompressed SD, OR recompress all my original HDV media to 10bit uncompressed BEFORE the edit, then cut in an uncompressed timeline.

This last method also gives you the choice of doing a center-cut 4:3 extraction from your HDV, or going letterbox. In addition, it gives you the ability to add edge enhancement to your downconvert, which is very important, via the filters available in a Compressor preset. When downconverting HD material, the edge enhancement the camera adds gets lost when the frame is scaled down. This allows you to add it back in, and exert fine control over the process. Which brings us to the subject of...Edge Enhancement

I expect somebody is going to say, "Well, I shoot WITHOUT edge enhancement to mimic the way film works". Something most people don't think about is that in most high-end telecine transfers, the colorist adds some edge enhancement. Skeptical? Here's a frame grab (and blow-up) from a music video that shot on 35mm, and was digitized uncompressed from the Digibeta telecine master:

http://homepage.mac.com/nweaver/.Pic...lframe35mm.jpg
http://homepage.mac.com/nweaver/.Pic...35mmDetail.png

See that 1.5 pixel halo? Didn't come from the negative :-)

One last tidbit, with pics. In another unrelated thread, on another board a couple weeks ago, I posted two frames. One from the same 35mm material above, and one downconverted from HD100 material. I had the good fortune of both frames being of downtown Los Angeles at magic hour. The HD100 frame is downconverted via Compressor:

http://homepage.mac.com/nweaver/.Pictures/1.tif
http://homepage.mac.com/nweaver/.Pictures/2.tif

Can you tell which is which? Downconverted correctly, things are pretty close!

Sorry for the classroom vibe, but I've been meaning to share all this with people for a while now, but I knew it was going to be a long post!

Greatly appreciated, Nate! Thanks for the education and keep those lessons coming.

Good stuff Nate. It reinforces what my friend Council (from the soul group "Maurice and Council") posted recently. Seriously, Council had the same method and results as yours regarding downconverting and for those of us that are doing a lot of it for broadcast it's definitely the way.

Previously when mixing 16:9 SD footage (i.e. XL2) with 4:3 for broadcast and DVD distribution dropping the XL2 footage into a standard timeline worked fine, but downressing is a whole other challenge and the 10 bit uncompressed through Compressor, is the way to get great looking HD footage out to SD.

Thanks professor...and I guess jpeg 2.

Thanks. This is great to know!

which if any of these methods uses the AIC?

is the "compressor" a final cut plug in?

I use the term "HDV" above in reference to both AIC and native HDV timelines and media.

Compressor is an auxiliary program included with FCP since version 4 to do batch processing and exports. It works either standalone or as an export module of FCP.

Note all my comments refer to the FCP5 suite...Compressor prior to v5 wasn't so hot at scaling video.

hey jim,
i think i was the one who suggested going thru compressor in 10bit uncompressed and council was saying to go thru the sd timeline way. and i give all credit to nate, because on one of the threads i read, nate said to go with 10 bit in compressor. eversince i tried that way and loved the results, it's the only way i will downconvert from now on.

Thanks for the info! I've been doing it the "worst" way. One question though:

Could you export the HDV sequence as DV NTSC using Compressor to avoid the huge file sizes? Would this result in lower quality than the 10 bit?

Thanks guys!

You could, and you'd still get very pretty pictures, but the color resolution when you do it the way I outlined becomes 4:2:2.

Downconverting to DVCPRO 50 would be a great stopping place in between if you can't go uncompressed.

I tried to go DV and the exported video looks jittery and the aspect ratio is off. Do I have a setting that's not right?

What kind of material from what camera in what mode?

[edit: well, yeah, there's a lot of settings to get right in Compressor. Maybe I'll post some Compressor presets to do this. Let me get dinner taken care of and I'll see what I can post]

No Problem Nate.

The material is of a single speaker who moves but its relatively slow movement. I'm shooting in HDV mode on a Sony Z1U. Thanks again for your help.

Okay, that's what I was suspecting. You're trying to go from interlaced HDV to interlaced SD resolution. I haven't made a preset for that, nor have I done any tests.

All of my presets and tests have been on progressive source media.

Sorry about that Maurice...I always thought you were the better singer though. I think it was Nate that first inspired me to go out to Compressor 10bit rather than "export as" as well.

Nate,

Ever tried editing in photojpeg? It might be the best compromise between HDV editing and finalizing in uncompressed. For 720/24 it should be around 7MBs with exceptional quality at 75%. Should downconvert via Compressor extremely well too.

Downconverting
 

Thanks Nate! Totally agree, so far I only tried the FCP timeline method, suggested by the FCP manual (with horrific results.) Just got my JVC baby 2 weeks ago and so far only shot tests.

Has anyone tried using the AIC (Apple Intermediary Codec) for capture and editing and then doing a downconvert using #3, the "uncompressed 10 bit compressor method (which seem to make the most sense, though for large shows it would chew up space like crazy)? How did it look? I'm trying to shoot a music vid in HDV 720/ 30P but will need to put it on an SD DVD and am trying to come up with the best method. On that note, how about going straight to MPEG2 compression with compessor for the DVD files instead of the 10 bit uncompressed intermediary stage? Has anyone done that?

Yeah, I did that with my Static-X video for DBeta delivery. Looks great. All of that ground glass grain I had from the Mini35/AIC combo shrunk up on downconvert and looked even more like film grain. Looked great uncompressed/Digibeta, but looked much worse on MPEG2 encode for DVD. Looked even worse when it got MPEG2'd for broadcast when I saw it over DirecTV/MTV2.

I've done that too, but it was before I was "paying attention". It should be just as good quality as doing the uncompressed intermediary...but so should the timeline method too!. I do know though that encoding MPEG2 from a FCP timeline with downcoverting, effects etc takes more than it should by quite a bit.

For approval DVDs on a short music video project, I'd encode from the timeline. That what I was sending Warner Bros on Static...was cutting in AIC, monitoring in HD externally, and then making SD MPEGS (@23.98!) via Compressor.

What's the value of capturing with AIC versus native HDV 30? I don't believe the AIC has more than a rudimentary capture interface.

HDVXDV -> .m2t file -> Streamclip -> Anything!
 

I am far from being an expert on this topic.
However, I use HDVXDV for log and capture of
720 24p footage from the HD100U or BRHD50 Deck.

The result are .m2t files.
I then use Streamclip (It's Free!) (http://www.squared5.com/)
I open my .m2t file in Streamclip. I goto Edit -> Fix Timecode Breaks.

This is a critical step that HDVXDV does not do and the reason
audio gets out of sync using HDVXDV for conversion.

I goto File -> Export To Quicktime
For Compression Select :Apple DVCPRO HD 720p
Quality 100%
Make sure: Interlaced Scaling & Reinterlaced Chroma are NOT CHECKED.
Enter: 24 into the frame rate box/field.
Make sure 1280x720 (HDTV 720p) is selected/.
Then select make movie.

I then make sure my FCP settings are DVCPRO HD 720 24p
in easy setup menu.
Import that clips in and can do a great multicam HD edit.

Streamclip does beautiful downconversions to SD/DV/DVCPRO.

Is there a more efficent method out there?

So, please do some experiments and any further feedback on this
would be great. Hope this helped someone out there.

There's a cheat where you can digitize 720p24 via AIC in Final Cut if you don't need your audio.

That's the only value I know.

I agree. Streamclip does use good math for downconversions if you're coming from m2t. My original post was more about how there are drastic differences within the FCP suite when it comes to quality.

Hey Nate.

Great stuff as usual, very useful to know the difference in quality with all the different conversion methods. BTW, I found the "Uncompressed 10-bit" codec in Windows version too (http://www.blackmagicdesign.com) and it's great if you need to exchange clips between Mac and Windows.

Nate, I replied to your email of the 18th, I don't know if you saw it, maybe it got trapped by the anti-spam filter?

Thanks again.

i know it's been a while since this thread was active, but i wanted to give it a bump because it was so useful. hey nate (if you're still reading this), how much "Sharpen Edge" do you suggest? thanks again!

I just read this about the methods for down conversion of JVC HDV footage - (topic save)

I am cutting a show that is in a HDV timeline in FCPro 2. Yet some of my footage is NTSC stock archives in 4:3. The show will be delivered in 4:3 letter box NTSC.

So, I am left with a dilemma. How do I down convert all the HDV footage, yet retain the full quality of the NTSC archives? Going out to compressor will result in a long clip with no way for me to replace the NTSC content shot by shot when placed in the NTSC sequence.

My plan is to copy paste the show into an NTSC uncompressed 10-bit sequence and then replace all the NTSC archives at 100% scale with letter box crops. But from what I read this method will result in poor down conversion of the HDV.

Is there any way to media manage the sequence so all the HDV is rendered to uncompressed 10-bit letterbox clips?

I ended up down converting the HDV footage to 10-bit uncompressed within FCPro. The results are superb. (gasp) But I used Media Manager to recompress the sequence to 10-bit. By using the Custom sequence settings in Media Manager you do have some control over the recompress, but no filters like Compressor. OTOH, you can continue to edit the shots. I had an hour long doc here. Plus I was able to delete the rest of the material not used.

The downconverted recompressed letterbox looks great and I eliminated the terrible aliasing and flickering in the details of the HDV footage when using a FCPro sequence downconvert. Looks much more film like. This show will get a national audience so, we'll see how people react. Lot's of beautiful footage from Italy and Poland. All shot 720p30 to Firestore as HDV. Edited natively in HDV to fine cut then recompressed to NTSC for mastering to Digibeta.

Dave Beaty

just came to same conclusions as you did. however that doesn't mean i'm happy with the results. I'm ready to download the trial version of adobe suite and give it a watermarked whirl. I've seen good things from adobe premier and dvd burning lately... but to be fair, I havn't checked out FCS 2 yet. For me, it just looks way too compressed and artifacti. yes, compressor, 2 pass best.. 6,7 and 7,7 bit max etc.. nothing close to a store bought DVD, and I think we should be able to get close the quality of one... not obviously home done.

OK, I'm an idiot... I went to film school before DVD's were invented so i never knew some basic info about DVD encoding. It's as basic as in America the steering wheel in on the left hand side of the car.. but I never knew it.. here it is... if you want EXCELLENT SD DVD.... SHOOT 24P. WHY? DVD is 59.97 interlaced frames. Progressive scan DVD players only has progressive LIKE feature when the source material is 24p, encoded at 23.97 frame rate. the two IDENTICAL 23.97 frames from the original 24p is recombined in the DVD player then sent to the HD-TV set. so for SD DVD, shoot 24p HD/SD for best results...

Hi Dave, did you find that with your workflow the results were better than using regular Compressor? Would you mind posting the exact workflow? Much appreciated. Thanks

It's actually:

http://www.blackmagic-design.com/support/software/

Gotta have a - in between the blackmagic and design or you go to a "sublime" website...