Editing workflow for HDV/SD

[racer-x]

Now that I've got my HV20, (great camera!) I need to change my workflow. I want to edit in full res HDV to transfer back to tape (it's the only way I can view full resolution right now) for future use. I also need to edit in SD to make playable DVDs for distribution.

Now I could capture in both formats, edit in both formats and encode in both formats. That's way too much work. After playing around a bit, I've found a workflow that fits me very well. It may not be for you, but it's near perfect for me. I'll share my workfow with anyone interested.

1) Capture HDV footage with HDVSplit (free)
2) Edit the m2t's directly in Premiere Pro 2. It works great, just as easy as if they were avi's.
3) Encode to HDV spec Mpeg file. I don't care much for the built-in Adobe Media Encoder, so I Frameserve to Main Concept 1.5 Standalone Encoder using the Debugmode Frameserver (free). Main Concept has a good HDV preset.
4) I then Frameserve the same Timeline using the Degugmode Frameserver, but this time I use Avisynth (free) to open the Frameserved Signpost avi and resize it to SD spec 720 x 480. I then send the AVS script to Standalone Encoder for DVD distribution.

This has worked flawlessly so far. The edited HDV footage gets transfered back to tape and looks perfect. The SD footage becomes fully Progressive because the virticle resolution is less than half. It's very clean and looks better than any SD footage I've ever shot before. No need for any Intermediate codecs of any kind. No unesisary rencoding, the footage only gets captured once, edited once and encoded once into each format.

This workflow may not be for everyone, but it sure works for me.

[reco]

I have the same camera, and thanks for your suggestion.

I am creating HD WMV files and I can play them on the XBOX 360.

I am also trying to create a HD-DVD on a DVD-R but I had little sucess so far.

[ronnylov]

I am thinking of buying a Conon HV20, seems to be worth it's price. When looking at HDVSplit homepage they refer to http://hdvforever.com/hdv/hdrhc1/freecapture/default.htm for installing DVHS driver. But that page is not accessible. So can you still use HDVSplit? Or does Canon provide a DVHS driver for HV20?

In my opinion I would keep it interlaced also at SD resolution because interlaced gives more fluid motion. With avisynth you can resize each field separately and then weave to interlaced again. But if you want it progressive, why not shooting in the 24P mode which is one of the special features with this camera?

When you frameserve from the NLE to an external compressor then all frames will be reencoded. But if the editing software use smart rendering then the internal compressor will only reencode the parts of source video that need reencoding, like cut points within a GOP, parts with altered video picture (filters, adjusting brightness, transition effects, titles). So you will actually loose quality in the parts that could have been left untouched and the rendering time will increase also if you recompress everything. Now I do not know if Premiere Pro use smart rendering of HDV but if not then I would concider using another NLE software. Smart rendering can only be used when output format is the same as input fromat, like when editing HDV source files to a HDV destination file.

Other than that your work flow looks good.

I would probably also create a 720P version for playback on my computer (which have a LCD monitor with 1280x1024 resolution so no need of more than 720P).

[racer-x]

You can use HDVSplit as is with no need to install anything. It will detect the HV20 no problem.

You can resize and keep interlaced if you want, but I like progressive. I shoot mostly sports, so I don't use 24p much. I'd rather they had a 30p option.

Premiere Pro re-encodes all frames, as does Vegas when editing HDV. My method is faster and produces better quality than the internal encoders. Remember that frameseving is sending uncompressed frames to encoder.

I also do 720p versions for uploading to web and such.

Anyway, it's just my method. I like it, but that doesn't mean it's the best. If anyone has a better method, than feel free to post.

[ronnylov]

Thanks!

Too bad that Vegas also reencodes everything, I was hoping it had smart rendering as I am used to when editing DV. I will buy the PAL version of HV20 as soon as I can afford it. So I have 25 fps interlaced or 25 fps progressive options. 25 fps progressive is a bit non-fluid too so I will probably use interlaced mode. Too bad it does not have a 720P 50 fps option but I can do that in avisynth when converting from 1080i to 720P.

[racer-x]

Here is a quick example of 720p encoded to H.264 @ 3200 kbps (720p @ 60fps) :

http://www.savefile.com/files/724296

The name of the file is Pitch_out.hdmov and it's 3 MB.

[crewe1000]

This is actually very helpful as I am intending on using Prem Pro after a number of years of using Adobe 6.5 for SD editing. Thanks.

One question I have if you dont mind, I have tried to capture some sample footage from my HV20 today for the first time via HDV Split but it is dropping over 100 frames/packets so the footage is not capturing properly. Its a brand new tape I'm using and the HV20 has just come out of the box. Any further ideas/tips you could share that may help with this issue would be appreciated. Cheers.

[edDV]

The main problem I see with this approach is the blend deinterlacing to 29.97p that results from progressive conversion. At 1440x1080 resolution the artifacts are low compared to SD source but the motion does get blurred similar to the NFL films 24p look.

Bottom line though, if you have good light and can shoot 1080i, you have enough source resolution to trade off in downscale. Action scenes can be edited to preserve 59.94 motion increments.

[racer-x]

edDV wrote:

The main problem I see with this approach is the blend deinterlacing to 29.97p that results from progressive conversion. At 1440x1080 resolution the artifacts are low compared to SD source but the motion does get blurred similar to the NFL films 24p look.

Bottom line though, if you have good light and can shoot 1080i, you have enough source resolution to trade off in downscale. Action scenes can be edited to preserve 59.94 motion increments.

Agreed, but I don't blend deinterlace. I instead use Avisynth to separate fields and resize. If you download the clip I linked to, you will see it's 720p @ 59.94 fps. It's encoded to H.264 with an HDMOV extention. The motion blur you may see is caused by the reletively slow 1/60 shutter speed I used. If I used a shutter speed of 1/500 or higher, there would be no motion blur.........

[ronnylov]

Downloaded your test clip and it looks good. I had to change the file extention to .mp4 to play it. It makes me want a Canon HDV20 even more!

I was thinking the motion blur could be caused by the compression to low bitrate h.264. Using a fast shutter speed may cause compression artifacts when there is a lot of motion so when compressing to low bitrates it may be better to use the slow shutter speed.

Personally I prefer motion blur over compression artifacts.