-
More “demystification”…
-
[Bill Davis] “This was once one of the MOST popular hubs of that.
Realistically, sadly, (honestly, crushingly to me personally) not so much anymore.
“Well apart from the fact that it has the clearest discussion of all the forums about the Colour wheels “issue”
\”Traditional NLEs have timelines. FCPX has storylines\” W.Soyka
-
[Bill Davis] “Now, apparently it appears to need to be “fixed” merely so it more closely meets the prior preferences of a certain class of colorists that might prefer that it work more like what they are accustomed to. “
That’s simply not true. This is not a preference issue. I spent a lot of the weekend putting the Color Wheels tool through its paces for an upcoming blog post of mine. To put it bluntly, it’s not right. Period. Yes, you can use it, but to make good corrections is tougher than it needs to be and you quickly induce clipping if you use the midtone wheel’s “brightness” control. It works as expected in Rec2020PQ – that’s 0 to 10K nits – a theoretical standard that doesn’t yet exist in the real world. It does not work correctly in Rec709 or Rec2020 – what all of us use right now.
– Oliver
Oliver Peters – oliverpeters.com
-
[Steve Connor] “Well apart from the fact that it has the clearest discussion of all the forums about the Colour wheels “issue””
Sorry, but that’s not even remotely close to true.
I’ve learned 10 times more about this issue in other forums compared to this one. The arguments on BOTH sides have been sharper, science based, and take into account things that appear to be relevant (such as how Apple might be trying to accommodate both Rec 2020 AND Rec 2100 factors in their software approach). (My ignorant layman’s shorthand)
The only thing so have come away from these discussions feeling totally confident about is that this whole topic is nowhere as simple as the initial “everyone knows you just divide this by that and you’ll then be accurate” Frame this thing kinda took on when it first appeared.
It appears that software engineering science is pretty damn complex and doesn’t lend itself to simple solutions.
Who knew?
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
I’ll try to unpack this Bill in terminology that you might better understand.
I put 20 additive dissolves into my HD timeline. I put 20 additive dissolves in an SD timeline but they all behave more like cross dissolves and look different to how the additive dissolve looks in the HD timeline? They are not quite like the cross dissolves which look identical in both timelines – sort of a hybrid dissolve. You can sort of fix it by editing the start and end points but it adds lots of fiddle to making it look like an additive dissolve.
Would you describe that as a bug/ mistake or undocumented feature? Would you want the NLE maker to fix it or just accept that the dissolves are different in SD and need extra steps to behave? You see the issue is not about color science as Simon has so eloquently illustrated. It’s about inconsistency. I think you are grasping at straws by copy pasting rationalisations that are not coming to terms with that fundamental inconsistency.
-
[Bill Davis] “This was once one of the MOST popular hubs of that.
Realistically, sadly, (honestly, crushingly to me personally) not so much anymore.”
Seems like its pretty much the same peeps that were here in the beginning of this, except maybe Andreas and the tone for FCPX is more positve than it ever has been for sure.
-
[Michael Gissing] “I think you are grasping at straws by copy pasting rationalisations that are not coming to terms with that fundamental inconsistency.”
Michael, with all due respect, I am “grasping” at absolutely nothing.
I am listening to multiple voices, and trying to discern who’s opinions make the MOST rational sense. That’s ALL.
My personal bias about this tool is real, but it comes from a history where for months upon months upon months people told me this tool “worked wrong” – when in fact, it was improving my results at every turn.
You will forgive me then for treating THIS debate the same way.
If Apple is “wrong” about the SCIENCE of how they have elected to process Color in 10.4 – then that will come out. Because SCIENCE is absolutely impartial. The math works or it doesn’t.
If, on the other hand, they have merely elected to apply perfectly accurate math in a way that a different interface designer might have done differently – but it still gets the results the user wants – that’s a whole different kettle of fish.
Overall, Apples stuff, in my experience just WORKS.Typically brilliantly. That’s a fact their marketplace results have verified.
A thousand times in 2011 I read: X just doesn’t work right. Ignoring all that was, in hindsight, one of the smartest business choices I have ever made.
If X is ACTUALLY somehow “broken” regards to their color implementation, I have every confidence that they will acknowledge that and fix it. They have done so with reproducible bugs and mistakes over and over again.
Heck, I’d be VERY surprised if somebody either in Apple PR, working for some virtual internet clipping service, or even on the Pro Apps team itself : doesn’t have eyes in this very thread.
These are, after all, VERY public discussions – archived forever.
If they change things, deeply, cosmetically or otherwise – OR if they don’t – it should speak volumes.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
[Bill Davis] “Michael, with all due respect, I am “grasping” at absolutely nothing.”
So why another new thread? Why keep posting quotes that have already been refuted? Why not adress the issue of inconsistency which is the core of this?
I’ve been here since day one so I do not need to be constantly reminded about your heroic decision in 2011 and how happy you are to be vindicated in your personal decision. If that was relevant to this debate I would be the first to acknowledge that in my response but it’s simply not relevant. I too am vindicated by the decisions I made in 2011 which are totally different to yours and equally irrelevant to this issue.
-
[Michael Gissing] “So why another new thread? Why keep posting quotes that have already been refuted? Why not adress the issue of inconsistency which is the core of this? “
Because I don’t believe I am. (Posting such quotes)
I’m not convinced they have been.
I’m hearing mosty opinions by non engineers here – and reading what appear to be facts with specific references elsewhere.
To wit: (within the last few hours in that other discussion)
Again from Phil P.
Guys, there’s no sweat and no harm in discussing; this is how everyone gets smarter. I personally enjoy this, as it forced me to brush up my video standards knowledge, having spent most of my recent time dealing with (digital cine) camera files.
Indeed, the wheels, from what I can tell, function differently in standard 2020, PQ and HLG. The reason is simple: There is a standardized OETF for 2020 that pegs diffuse white to studio monitors that are typically set at 120 NITs— the same as 709. And because there is no notion of ‘specular’ white in SDR (remember that Rec 2020 is also an SDR format), it doesn’t matter if your display is set to 400 NITs or 100 NITs — you’ll see a natural-looking image as long as the display is calibrated to that OETF.
Still, in a professional environment, you should set your monitors to 120 NITs and operate in a dim room, even if your displays are capable of 400 or a 1000 NITs. The reason is that you might get into issues even on good monitors, with the dark portion of the tone curve, and your perception of contrast will be altered. Film editors need to operate on displays or projectors set at 48 NITs, which requires a totally dark room.
When it comes to HDR, there is a conundrum facing software designers, the same as it is facing users: To what target display are you mastering to ? Because different so-called “HDR” displays are not capable of the same dynamic range, the industry tends to gravitate around different mastering levels — 400, 1000 and 4000 NITs for HDR. If your encoding is for Dolby Vision, that system handles well the problem at the display level. Other systems less.
So there you are, doing an edit in WideGamut (2020 chromaticities) and HDR PQ (OETF). Maybe you got footage from a properly converted ARRI raw file and you have a pretty good PQ-encoded image that extends to 10,000 NITs. What do you do with it ? Do you grade it to full range, or to 4000 NITs ? Or to 1000 NITs ? You’ll be tempted to grade it to whatever your own HDR display is capable of, but contrary to standardized broadcast displays, we’re not there yet, in a number of HDR setups. The industry has yet to come to terms.
So put yourself in Apple’s shoes. They need to support HDR and in a way that makes sense. What makes sense at this stage is to treat your HDR material as full range and to apply linear transformations to it, that are mathematically invertible. They use Offset, Lift, Gamma and Gain in PQ. In SDR 2020, from what I can tell, they use offsets for master and shadows, and gain for midtones and highlights, but going through a weighting function (that’s why the effect looks non-linear). This is consistent with general practice for SMH, although there is no “standard” per se (nothing SMPTE, EBU, ASC or AMPAS have recommended). Resolve uses arbitrary bezier polynomials to split the Luma range, and that’s their approach.
In the meantime, their design intent is that if you’re grading a 709 project, things will look and perform normal. If you grade in SDR 2020, things will operate normally, similarly to 709. When you grade in HDR, things will behave as best they can, given the current nature of the medium…
——-
Phil continues, but I’ll stop there because this whole conversation is getting boring. The point is I feel like I’m getting higher quality explanations with more explanatory detail – all buttressed with references to textbooks and SMPTE publications encouraging me to NOT rely on opinions about this stuff, but rather go determine the truth for myself (to the extent, at least, I’m capable of understanding it.) ELSEWHERE.
This is supposed to be a conversation seeking the TRUTH about the allegation that a piece of software may have BAD DESIGN at its core.
It’s software I make my living with.
So making good choices about it is important to me.
If you have counter material from equivalent sources arguing that Adobe or AVID or Resolves methods of doing this stuff are better – bring them to the table. Seriously. Please.
What will help EVERYONE is the truth.
Not defending anyone’s opinion silos – most assuredly including mine.
Creator of XinTwo – https://www.xintwo.com
The shortest path to FCP X mastery. -
[Bill Davis] “If you have counter material from equivalent sources arguing that Adobe or AVID or Resolves methods of doing this stuff are better – bring them to the table. Seriously. Please.”
Phil is talking about monitor calibration and the as yet to be standardized HDR requirements. I actually understand this stuff as it is part of my bread and butter. However, Simon, Oliver and I are talking about how the gamma function in rec 709 behaves non standard compared to every other grade software that we know and have used. Simon has shown how the behaviour is non standard to other systems but the biggest issue here is that the behaviour changes within X depending on color space. No amount of SMPTE standard quotation or monitor calibration requirements can or will change this.
So one last time. X does the gamma function with the new wheels software the same as every other grade system since forever as long as you are in rec2020 color space. In rec709 (the more common space at this time) it simply doesn’t and there is no logical reason why as it is not only different to everyone else but it is inefficient and inconsistent.
All other grade systems are consistent with gamma regardless of color space. This actually has nothing to do with monitor calibration so the nits are irrelevant.
-
[Bill Davis] “If you have counter material from equivalent sources arguing that Adobe or AVID or Resolves methods of doing this stuff are better – bring them to the table. Seriously. Please.”
To start, let’s not conflate color management with color grading!
I’d highly suggest reviewing some of Baselight’s workflow materials. There’s a lot to learn about color science that’s applicable to any platform:
https://www.filmlight.ltd.uk/workflow/truelight.phpAnd of course, Baselight has an excellent technology for managing colour (they are English!) inside their colour grading system:
https://www.filmlight.ltd.uk/products/truelight/overview_tl.phpWalter Soyka
Designer & Mad Scientist at Keen Live [link]
Motion Graphics, Widescreen Events, Presentation Design, and Consulting
@keenlive | RenderBreak [blog] | Profile [LinkedIn]
Reply to this Discussion! Login or Sign Up