SpeedHQ & future considerations

[KSTAR]

Now that the amazing SpeedHQ codec, and all the glory it brings, is in our lives. Im curious what others will consider when building future systems in regards to hard drives and controllers?

Since SpeedHQ delivers such great quality with low overhead how will that impact what drives and controllers you will use?

Is uncompressed editing dead?

Im curious to hear what others think about this?

[Jim Capillo]

Most of my work are :30's, so drive space isn't a real consideration. I'll probably stick with uncompressed for the time being. Lower prices of the huge drives are a factor, too.

[Jim_C]

imo, except for folks who are doing tons of FX work, capturing component for beta, or LW work with tga stills, uncompressed editing has been dead for awhile.


And SpeedHQ can cover the beta capture as well as rtv could and can.

[paulfierlinger]

So once you get done editing with SpeedHQ, what do you all do for final output?

[Jim_C]

90% of the time I take this deliciously rich SpeedHQ(and before NT25) and squash the he!! out of it to the much dreaded but universally excepted mpeg2 thru tmpgenc for dvd's.

The other 10% I leave it in nt25(now speedHQ) for imag display thru the VT.

[ScorpioProd]

Note that though SpeedHQ is a great codec, it's not necessarily nirvana for everything. Remember, it is a compressed codec, and compressed more heavily than DV.

[Jim_C]

So it doesn't hold up as well sending it thru another compression cycle say ummm... ->DVD ?

What about NT25? How did it hold up vs DV going to mpeg2?

[ScorpioProd]

Good questions.

See, thing is, one of the things I like about the Canopus HQ codec is that like the PicVideo M-JPEG codec, you can set a "quality level".

This isn't the case with SpeedHQ. And this results in an interesting situation.

SpeedHQ generates clips that are about 25% smaller than DV. As always, you can't get something for nothing, and it is a DCT based compression system. So something has to give somewhere.

In my own testing, I found that a DV project, rendered out to DV and then encoded to MPEG-2 in TMPGEnc had a higher final quality than the same project rendered out to SpeedHQ and then encoded to MPEG-2 in TMPGEnc with the exact same settings.

Where I found the difference was in dark areas. This was a stage show, and there were scene changes with blackouts. This was what caught my eye, the blackouts had signficantly more macroblocks visible when my source was SpeedHQ than when my source was DV. Now again, the TMPGEnc settings were identical, and this was a 2-pass VBR MPEG-2 encode. Logically, if the source footage is noiser, it's not going to compress as well.

So, to prove the cause, I went back into SpeedEDIT. I loaded up the same blackout from the rendered DV and from the rendered SpeedHQ. I then turned up the brightness of my signal by increasing Y gain to 4 and Y offset to 50. This changed my black to grey, so I could see the grain structure on my output window. Sure enough, I saw significantly more macroblocking on the SpeedHQ clip versus the DV clip.

Now again, I'm not saying SpeedHQ isn't a great compression format, it is. But I apparently hit a weakness of it. It's just a law of physics kind of thing if you're going to be compressing something heavier, it's gonna happen.

So again, I encourage everyone to do their own comparison, but for me, I'll be rendering my DV projects out to DV if I need to put them in another application.

But as for HDV, that's a different story. For HDV resolution, SpeedHQ apparently has either a higher quality setting set for it, or just the intraframe nature of it gives a higher quality than HDV, and a bigger file size. Unlike the DV comparison, SpeedHQ at HDV resolution is a bigger file (higher data rate), and therefore lower compression than HDV. (Granted, comparing intraframe compression and interframe compression is apples and oranges, but still, intraframe will always be more roboust as an intermediate anyway.) And since the only thing I can compare it to is HDV, I'd strongly recommend using SpeedHQ when you need an intermediate at HDV resolution in or out of SpeedEDIT. I use it for doing VC-1 encodes of SpeedEDIT projects in WME, for instance. Or if I'm going to downconvert and MPEG-2 encode outside of SpeedEDIT.

Oh, and I also use SpeedHQ with the alpha channel for bringing in CG from programs like Oasis or India Pro.

So again, I'm not condemning SpeedHQ, it's a great codec. Just be careful with it till you know it can do what you need.

Oh, and no, I never did full comparisons like this of NT25, but it's just a legacy format now anyway.

[KSTAR]

So once you get done editing with SpeedHQ, what do you all do for final output?

Paul for me it has always been uncrompressed Quicktime on a Firewire or USB 2.0 drive. I dont have a high end deck so Im in the habit of taking a drive to one of our local studios. I have delivered uncompressed QT's to studios that use both FCP and Avids, connected to Beta SP, and Digibeta decks. The longest edit I have delivered was 48 minutes so the QT file was around 70 gigs