Is FCP5 better than Vegas for HDV?

Ben1000 wrote on 1/16/2006, 5:26 PM
Howdy...

I recently saw a demo of FCP5 handling HDV without an intermediate codec, and looking pretty swift doing it. I understand that many Vegas folks use an intermediate, and that seems to be the accepted way to go for Vegas...

Is FCP5 doing some sort of magic that allows them to do mid-GOP transitions and cuts and still do it in realtime, and then render only the changed sections of the video?

or do they require a complete re-render even if only 1 frame is changed? I understand render times are long...

Basically, I'm looking for a definitive answer on native HDV editing on FCP vs. the same on Vegas, and if Mac has pulled ahead of Vegas with this HDV editing feature...

Love to hear thoughts from those who have used both..

Best,


Ben

Comments

Spot|DSE wrote on 1/16/2006, 5:48 PM
Apple does indeed work with native hdv better than Vegas for purposes of on-timeline editing IF you have a super fast Apple. However, there are still issues of dealing with the B and P frames. For this reason, Apple's team still recommends you use their AIC. (Intermediary codec)
Additionally, they have smart rendering for only changed segments. Frankly, this is the ONLY place that FCP is more efficient than Vegas, and on the other side of the coin, Vegas is much faster and efficient to process clips. There are GOP issues when you smart render as well, which is another reason that everyone recommends an intermediary. (Canopus, Avid, Ulead, Adobe, Sony, Apple)
Coursedesign wrote on 1/16/2006, 8:00 PM
...and Cineform's wavelet-based intermediate codec seems to be quite significantly better than Apple's AIC.

Wavelet encoding is just a more advanced technology that works well in practice, too.

It imitates nature, and as so often happens, this ends up being the best way... :O)

Wavelets were introduced for signal processing in 1965, and got a very enthusiastic reception from those who saw that the classical Fourier analysis they had been using for electrical signals was a very useful technique but not that efficient.

Their introduction of wavelet analysis in 1965 as a "latest and greatest new thing" got a very very amused response from the Numerical Analysis folks who had been using the technique since 1805 (when Carl Friedrich Gauss invented it).
Spot|DSE wrote on 1/16/2006, 8:05 PM
yes, I should have mentioned that of ALL the intermediary possibilities, CineForm is currently the best offering due to their use of wavelet technology. How long they'll keep that is anyone's guess. J2K is coming on hard/fast/strong, and I suspect someone will pull their head out somewhere, and create a J2K intermediary.
Coursedesign wrote on 1/17/2006, 1:57 PM
That seems very possible.

JPEG2000 is also based on wavelets. The difference between JPEG2000 and CineForm lies in how the wavelets are compressed (wavelets by themselves provide no compression).
winrockpost wrote on 1/17/2006, 3:20 PM
huh ? damn this stuff gets confusing ,the more i think i learn the the less i know.
Spot|DSE wrote on 1/17/2006, 4:41 PM
The biggest difference aside from the wavelet compression, is that J2K is always scaleable. You could for instance, choose to edit with say 10 scans of the frame, and therefore be dealing with decent footage, but at a fast data transfer rate. Then you render with 100 scans. Too simplistic an explanation, but it's nice to know there are "dial-able" values in the quality.
Coursedesign wrote on 1/17/2006, 11:34 PM
That is inherent in wavelet technology, which works very much like how our brains perceive what we see with our eyes.

We see a forest at a distance like a green mass with few features, then as we get closer, we see more and more detail.

This is different from a photo which only gets blurrier as we get closer.

With wavelets, as we add data, we don't get more pixels, we get more detail information (which is not the same thing). It's more of an "organic growth," rather than the algebraic growth of detail that comes with higher pixel resolution.

JPEG2000 is also the choice for the new DCI (Digital Cinema Initiative) standard, which I thought looked really outstanding on a full size public theater screen.
David Newman wrote on 1/18/2006, 9:50 AM
We agree that JPEG2000 could be an competitor to CineForm in the intermediate compression space, but not for some time. J2K uses a very nice arithmetic coding scheme that currently requires hardware for any decent level of decoding performance. It was designed for still image compression to surpass classic JPEG, so concerns over sequential encoding/decoding speed was not a design consideration. At CineForm that was our highest priority. So these two wavelet codecs are designed with different aims. Out of speed, quality, and size choices, J2K chose size & quality whereas CineForm chose speed & quality -- both have a strong emphasis on quality. Note: anything that pairs size and speed is not suitable as an intermediate.

David Newman
CTO, CineForm
blog: cineform.blogspot.com
Coursedesign wrote on 1/18/2006, 10:15 AM
Dan,

Your clear writing is a joy, and I really have to take my hat off to you for the thorough, no B.S., side-by-side comparison of different HDV editing workflows on your web site.