-
Capturing DVCPRO50 vs 10-bit Uncompressed
-
[Graeme Nattress] “or if you capture DV50 to uncompressed it goes like:
DV50 -> UC10bit 4:2:2 0> 32 float 4:4:4 -> effects -> back to 10bit uncompressed.
And that’s why it only just wastes disk space.”
Twaddle…I have to add that to my vocabulary.
See, this is where the theories and practice take a turn for me personally, but then I loop back to theory.
In theory, and by talking it out or writing it down, you would think that this would be a ‘worse’ workflow as you are adding ‘another round’ of compression. It is here this hardware out of the deck comes in. Using the deck’s hardware to treat the dv50 in footage in hardware to uncompressed and then work with that footage, all uncompressed all the time, works better for me.
Taking dv50 into an uncompressed timeline will either add scaling or a scan line shift then that effect gets rendered with all the rest, which sometimes doesn’t work well inside FCP. Capturing from a deck you get 720×486, not 480 with this simple conversion being done in hardware, not FCPs software, along with other decompression that happens in hardware from the deck. As far as proof, all I have now is my eyes. I have taken a project and finished it dv50, then recapped to 10 bit, re-keyed and cc’d and at the end, the 10bit file is much cleaner. Remember, I am talking about the rigors of editing, not simple capture and playback. I have been trying to get my hands on what exactly the deck is doing, but no one will tell me and I don’t know any engineers that build VTRs. I am sure it’s all proprietary and I won’t get any real answers anyway. So there it is, back to theory.
Jeremy
-
I’m with Graeme.
And, in addition to what he spoke about, I think there is also an extra step involved at the card level when capturing DV50 to an uncompressed format, which if memory serves correctly, involves a D to A (digital to analog) conversion, which would tend to be a source of potential artifacts that D to D would most likely avoid. Am I correct about the D to A thing Graeme???
David
David Roth Weiss
Director/Editor
David Weiss Productions, Inc.
Los AngelesPOST-PRODUCTION WITHOUT THE USUAL INSANITY™
A forum host of Creative COW’s Business & Marketing, and Indie Film & Documentary forums.
-
[David Roth Weiss] “I’m with Graeme. “
That’s cool. Healthy discussions are good for me, and this forum. I love learning if my intuition and subsequent methods of practice are wrong or misguided.
So, what ya got for me?
All I have is my own testing and practice, nothing else. I have shown split screens to folks that I work with (clients/dps/designers/whatever), and they all pick 10bit. That’s my test. Not very scientific, I know. No matter how hard I try, I have never been able to get a real live Panasonic engineer on the phone to tell me how it works and what’s going on in the deck.
As far as an a/d conversion. All captured footage goes through an a/d as the light that comes through the lens is ‘analog’ info converted to digital off of the chip.
Just curious, but do you finish in dv50 from footage shot on dv50 tape, David?
Jeremy
-
[JeremyG] “Just curious, but do you finish in dv50 from footage shot on dv50 tape, David?”
Never finish DV25 without switching compression to 8 or 10-bit, but I do leave DV50 as is because the color depth is sufficient for graphics and text.
When push comes to shove, hard drive space these days is the least expensive of all the things we as editors need to be concerned with, and rendering on the new machines is pretty fast too (at least SD rendering is), so if you feel comfortable bumping to 10-bit, I don’t think its any reason to blow a fuse even if it is “just twaddle.”
BTW, “twaddle” is one of those great Brit words that means nothing, but says a lot. They do have a way with words…
Happy thanksgiving Jeremy and Graeme…
David
David Roth Weiss
Director/Editor
David Weiss Productions, Inc.
Los AngelesPOST-PRODUCTION WITHOUT THE USUAL INSANITY™
A forum host of Creative COW’s Business & Marketing, and Indie Film & Documentary forums.
-
Jeremy,
I’ve never had the chance to use that desk, But I believe you when you say that the 10b Unc picture you get from the desk is better than the one that you get capturing DV50 and transcoding in FC.
My point is that if this is like that is just because the desck is running a better software than the one you are using when transcoding in FC.
As you say the people of Panasonic know their own stuff better than nobody else. They must have a better algorithm to do the resampling. Chroma rebuilding and whatsoever.
If you wuld have a filter that would do the same maths with your DV50 file, you could get exactly the same results in FC.
Happy Thanksgiving. Here the That Luang. The more important Lao festivity.
Cheers,
Rafael
PS: I love this kind of discussions. I really miss them here.PPC G5 2x2Gh 4GbRAM/BlackMagic SD/PMBP 17″Core2Duo 4GbRAM
JVC DTV-17″/FCS2/AE CS3/COMBUSTION/SHAKE -
Well, as the camera codec only records 480 lines, where are the other 6 coming from? And of course, an uncompressed timeline need not be 486 lines, but can be 480.
What it comes back to, is not that such a flow makes you happy in the edit suite, but have you checked with, rigorous experimentation, the output of the SDI port on the deck compared to the Apple DV50 decode? When the same question was asked about DV that’s what I did. I spent a lot of time and enlisted some good help to perform that experimentation as best as I could. Would you care to do the same with your DV50 gear and write an article on the results?
Graeme
– http://www.nattress.com – Film Effects and Standards Conversion for FCP
-
[Graeme Nattress] “Well, as the camera codec only records 480 lines, where are the other 6 coming from?”
Embedded with black out of the deck.
[Graeme Nattress] ” And of course, an uncompressed timeline need not be 486 lines, but can be 480. “
Yeah, but all hardware expects a 10bit signal to be 486. It’s tough to play 10bit out @ 480. FCP will drop frames.
[Graeme Nattress] “but have you checked with, rigorous experimentation, the output of the SDI port on the deck compared to the Apple DV50 decode?”
No, as I mentioned before. Just my eyes which is not really rigorous and totally subjective, BUT I will say that everyone that looked at the not so scientific test results always picked 10bit. Always. Like I said, this not only has to do with capturing footage, it has to do with the entire edit. The final finished product. The footage does not look nay better when you capture dv50 vs 10bit, but after a cc or green screen and putting gradients, text, and solids, the 10 bit looks much better and more smooth due to the extra little bits.
[Graeme Nattress] “Would you care to do the same with your DV50 gear and write an article on the results? “
I’d love to, and I’d love to find the time and the know how. I’ll have to dig up your article and reread. It’s been a while since I looked at it. This would take it even a step further as it’s 4:2:2 to 4:2:2. The problem is, is how do you measure subjectivity? Mathematically, I think you can do the tests and get the results, but why is it when I look at a finished dv50 and a finished 10 bit timeline, the 10 bit looks better, even though statistically, it might be the same? I am not sure how to measure that.
Thanks for talking this out.
Jeremy
-
I’ve never had problems playing out 480 lines over hardware. Never a dropped frame warning.
For the test, just capture the same source clip thrice, once over SDI to uncompressed 10bit, and another to uncompressed 8bit, and finally over FW. If you can do that on a small, colourful clip, that’s probably best.
Different 4:2:2 codecs are handled differrently by FCP, some will be chroma smoothed, some not, so you still have to analyse what the chroma is doing. That can significantly effect a key, so if you’re noticing better keying, it’s vital that the methods of chroma subsampling reconstruction are verified. If you have time to capture stuff, I’ll take a look and see.
Well, you don’t try to measure subjectivity – what we’d look for is any mathematical differences first, both in bit depth, resolution, chroma sampling, compression artifacts etc. That might be enough to explain what you’re seeing.
Graeme
– http://www.nattress.com – Film Effects and Standards Conversion for FCP
-
Awesome. I am away right now, but back in the office next week.
I’d also like to throw ProResHQ in the mix. Haven’t tried keying with that yet, although a green screen project is coming up and I’d like to put that codec through some sort of testing.
I’ll see what I can dig up next week. I have a few clips from recent projects in mind.
Jeremy
-
Thanks. ProRes should be included too!
Graeme
– http://www.nattress.com – Film Effects and Standards Conversion for FCP
Reply to this Discussion! Login or Sign Up