-
Vegas 6 and Decklink working great!
-
I’m a little confused. So most edit systems process in 8bit, but is that full 8bit 4:4:4 processing or the lesser 8bit 4:2:2 (YUV)?
It makes sense that a frame rendered in 8bit 4:4:4 RGB would loose information if saved in a 4:2:2 YUV, so I’d think 10bit 4:2:2 would be needed to compensate for the lower “U” and “V” resolution in the codec.
I’m not a video expert, perhaps someone who understands the 4:2:2 codec better could elaborate or correct me.
Thanks,
Stewart -
Dear friends, it’s a bit strange when the tool you favor doesn’t do something and you claim it’s not important. We do have a saying in spanish for that, but I don’t know if there’s an English translation for it (is about a fox trying to reach certain grapes from a tree and when he fails he says “they were not ready yet”). I would love FCP to do simple 3D transformations like Vegas apparently does. It would be foolish to say it’s not a good thing or that I can do better than that in After Effects (which is obvious).
Yes, TV is 8 bit (or actually, no bits at all for 95 per cent of worldwide audiences). But 10, 16 or 32 bit PROCESSING is an important thing regardless of how many bits per channel your final file has. That expanded color space has obvious advantages when doing effects, compositing, motion graphics, etc, regardless of the final color depth. Less banding (or no banding at all) less rounding errors (or none) in color correction and richer color gradients are evident examples.
The previous poster’s claim that FCP output is 8 bit only is a bold one. It could be true but I would like to see what part of the tech specs says that. In any case, even if true, the good thing about scientifc observation is that in FCP you can switch between 8 and 32 bit processing and see for yourself the difference. Just as you can do in After Effects with 8 and 16 bit processing (the difference of which is evident even with 8 bit output devices).
I guess that as more application are supported by Decklink boards, this discussion will show up more and more in this forum.
Just my two cents, as you say there 🙂
All the best
Adolfo Rozenfeld
Buenos Aires – Argentina
https://www.adolforozenfeld.com
adolfo@adolforozenfeld.com -
All manufacturers tend to present the tech specs they are the most proud of, and leave the rest to be talked about only when absolutely necessary, if at all.
From https://www.hdforindies.com:
Apple’s […] demo artist said {…] there is NO 10 bit RGB processing in FCP – if you render anything, even a cross dissolve, it gets bounced into 10 bit YUV, then rendered from YUV back to RGB. Yuck. RGB is handled at 8 bits/channel max. […] So maybe next year.
Vegas works in 4:4:4 high-bit internally for effects and color correction, and I suspect this is the case for all of the top NLEs, including FCP. Still, in the end it all gets reduced to 8-bit RGB.
It’s a big step performance-wise to go above 8 bits, but it’s gotta happen someday.
-
Hello, BJ.
While it may be true that most manufacturers (including Apple) play games with specs for their benefit, in this case you’re at risk of doing just the same. And you didn’t quote Apple’s tech specs, but an inaccurate report from something that someone said (hd for indies is a great site, but they do commit mistakes with technical matters frequently). There’s a full tech specs section on Apple’s FCP site. There could be half lies because of what they don’t say, but everything they do say, is mostly true. Apple does say evident marketing BS about HDV native editing, but that’s in a separate “marketing” paragraph. The tech specs page is quite holly in this sense.There is not 10 Bit RGB processing in FCP because 10 bit RGB doesn’t exist for any practical purpose. It could exist in theory, I guess, but RGB is usually 8 or 16 bit per channel. 95 per cent of professional video formats (Digibeta, HDCAM, DV25/50/100, etc) are in YUV space, not RGB. FCP got native YUV space in version 1.25. Then it got 10 bit YUV capture and editing (but not processing) in V3. In version 4, it got (up to) 32 bit floating point processing, which is just the highest one you could do with computer imagery. It is much wider than 10 bit YUV and 16 bit RGB. You do have the option to limit yourself to “just” 10 bit YUV processing if you want to. Since, again, practically every video format uses YUV (HDCAM SR being an exception), this usually ensures more precise matching of processed and unprocessed segments. While 16 bit RGB is theoritecally much wider than YUV (but a lot less than 32 bit float), there is no video format using that color space, so 10 bit YUV (as in Digibeta, no less) is the most common way of transcoding it. If you’re NLE is RGB, then you know that it WILL have to transcode to YUV, as your destination format is surely YUV.
As for effects being processed in 8 bit RGB, that’s mostly only true for AE plug-ins. FCP’s own FXScript allows developers (and the effects themselves) to choose between RGB and YUV, depending on the timeline settings. According to some people, a few effects are indeed truncated to 8 bit processing, But advanced effects like the Color Correctors do not go through FXscript anyway, and are plugged into FCP’s engine directly. Thus they are capable of 32 bit float. Version 2 of Motion just got 16 and 32 bit processing too. And the same goes for Apple’s new plug-in arquitecture, image units.
8 bit uncompressed as Vegas, Premiere, FCP and Avid do it can look wonderful and be more than enough for a lot of cases, but that’s not a good reason to say inaccurate things about the ones which do support wider color spaces.
All the best
Adolfo Rozenfeld
Buenos Aires – Argentina
https://www.adolforozenfeld.com
adolfo@adolforozenfeld.com -
the resolution and bit depth are two different things: let say pal is 720×576 in rgb R, G and B channel must all be the same size (720×576) beacuse if youwould make on half resoultion it would look really off that is why in the old times YUV (colorspce was invented partly because reduced bandwith and partly because backward compatibilty with black and white standard Y) in this case Y consist of black and white image and two other channels make up the croma information so the croma can be 1/2 or even 1/4 of the orignal resolution (PAL DV 4:2:0 or mpeg2 compression), the bit depth resembles the different levels that can be stored in every channels (one channel is 2d array of pixels) one pixel (8 bit 256, 10 bit 1024)
hope this clears it up a bit – if not make it evene more complicated
-Kaspar -
Stewart: As I undertstand it, 8 vs 10 bit processing is not about color sampling resolution (4:2:2 vs 4:4;4) but it’s about wider color spaces (millions of colors versus trillions of colors). While it is true that FCP is probably the only sub $100,000 NLE doing true 10 bit per channel and 32 bit float, it’s also true that for compositing and graphics you get the advantage of 10/16 bit processing by using AE/Combustion/Motion 2/Shake/Fusion with codecs like Blackmagic’s (up to 10 bit, free!) Sheer video (16 bit+. BitJazz), Micrcosm (16 bit+, Digital Anarchy) and None16 (16 bit+, Digital Anarchy, free but unbearable huge files). The 16 bit ones will have to be converted to 8 or 10 bit depending on your NLE, but they are used by many compositing, color correction and motion graphics artists as a digital freezer, or in other words, to mantain the highest possible quality in an inter-application worklow (like going from a 3D app to After Effects and then to a NLE application). I understand Premiere can handle the 10 bit files without hurting them if you don’t process them. But I have not seen Premiere since… a lot of time. Vegas, I mostly know what I am reading here. Apparently not, if it’s 4:4:4 8 bit RGB.
Adolfo Rozenfeld
Buenos Aires – Argentina
https://www.adolforozenfeld.com
adolfo@adolforozenfeld.com -
In my previous post I said that 10 bit RGB doesn’t exist for practical purposes. I should have said for *most* practical purposes. There surely is a use for it in pure compositing/graphics tasks. Not so much for actual video formats. Sorry about that one.
Adolfo Rozenfeld
Buenos Aires – Argentina
https://www.adolforozenfeld.com
adolfo@adolforozenfeld.com -
[Adolfo Rozenfeld] “In version 4, it got (up to) 32 bit floating point processing, which is just the highest one you could do with computer imagery. It is much wider than 10 bit YUV and 16 bit RGB.”
Are you sure about that? 16 RGB has a 280 billion+ possible colors. How many can 32 bit YCrCb can do?
-
Adolfo is right about the number of bits being about color space resolution, and floating point is higher resolution than 16-bit video even (which is really 3 x 16 = 48 bits).
16-bit video can accurately represent 281 trillion colors.
Floating point color uses a value between 0 and 1 for each color channel, usually expressed as a decimal number. Because of the exponent part in this, the number of colors is virtually unlimited.
[Pinnacle] Liquid Chrome handles uncompressed 10-bit for abt $20K.
Vegas handles 10-bit uncompressed codecs such as the excellent BMD codec without trouble, but for assembly and simple cuts only. Effects and transitions are processed internally in high bit depths, but the output is always truncated to 8-bit RGB. I shoot only uncompressed 10-bit nowadays, and use Vegas, so I have to use Combustion for any processing that needs to maintain the higher bit depths between steps. AE Pro would also be OK (but not the basic AE version which is 8-bit only), as would Fusion, Shake, etc.
BMD offers “the world’s first 10 bit RGB 4:4:4 Dual Link (SMPTE 372M) editing solution for Premiere Pro, including full capture and playback to true RGB 10 bit files.” This seems hot, perhaps a PP user in this forum could confirm that the 10-bit RGB is maintained all the way through to the output of all effects and transitions?
I have not been able to find better tech specs for FCP than what Apple has on their web site, and those specs say nothing about the output bit depth of internal effects and transitions. Perhaps somebody else knows?
I also found the following at https://www.cinematography.net/Pages%20DW/HD_Post_Costs.htm:
Final Cut, in its current form, is not the place to be doing high end color manipulations because it is bit depth limited and the color algorithms are not particularly sophisticated. Perhaps in the future there will be a third party add in solution to this, but right now if you want clean color correction and the most extensive manipulative capabilities, you’ll be looking elsewhere (i.e., DaVinci, Pandora, Lustre, etc.).
The rest of the page linked to above shows what some feature film pros think is the best workflow for FCP.
I’m hoping we don’t turn this into “my NLE is bigger than yours.” All NLEs have strengths and weaknesses, and for the record, Vegas is totally unsuitable for any film work requiring matchback. I wouldn’t have any problem switching to FCP, it’s high class, widely supported and the FCPUG is a fantastic resource.
Today I like Vegas because I’m not currently doing film work, I like the totally different paradigm for editing, and I appreciate that it is rock stable.
Is there another NLE that didn’t “computerize a flatbed,” but started with a clean slate to try to figure out the best way to edit, without having to do it as it was done before with film strips?
-
BJ: We agree much more than we disagree and nothing like my NLE is bigger than yours. I only jumped on a Vegas thread because something I saw as inaccurate (or mostly inaccurate) was said about FCP. While FCP does support 10 bit output and the higher space for internal processing, I would never think of it as a direct to film effects tool. For starters, it doesn’t do DPX or OpenEXR. Not to mention Its’ scaling algoriths were quite poor, which is a bigger issue than color space (apparently FCP 5 changes that, which is the good thing about Apple – every time they do something poorly and users complain, tbey don’t stop until they get the best of the best in that particular area).
The jump in FCP to true 10 bit output was when Apple released their own 10 bit UC codec, which is the one favored by Decklink and AJA nowadays for FCP work.
10 bit YUV is amazing for broadcast and HDCAM level post. Of course manipulating fim in digital form or 4:4:4 HDCAM SR is 95 per cent of the time a compositing/effects thing, not an editing one. So it makes sense that compositing apps offer higher levels of processing and output.
All the best
Adolfo Rozenfeld
Buenos Aires – Argentina
https://www.adolforozenfeld.com
adolfo@adolforozenfeld.com
Reply to this Discussion! Login or Sign Up