#apertus IRC Channel Logs

Bertl: Re mini bnc, why would I want 5 in the space of 4? I'd rather have it integrate with the cables I've already got and are standard on all other pieces of pro video gear

I'll take a photo tomorrow of why mini-bnc on set are stupid :)

note that I heard the same argument 15 years ago regarding BNC/RG58 vs RJ45/ethernet :)

Bertl: Both ethernet and, more importantly, long signal video for playback.

nevertheless, BNC died with 10Base2 for ethernet :)

anyway, I'm just curious why the (as guest stated it) only 20% smaller mini-BNC is sooo bad compared to BNC (which is sooo good :)

I understand that most folks have 'old' BNC cables lying around somewhere, so it makes sense that they want to use those instead of getting new (smaller/better) ones ... but that's not really an argument for or against the connector itself

Bertl: As far as I have learned, it is the same thought tas rexbron_

Bertl: It really is that much weaker

off to bed now ... have a good one everyone!

good day!

There's my photo. Note that the recorder only comes with 2x adaptor cables. We've ordered more as backups but within 12" of the recorder, I need to connect regular, longer bnc cables to connect to the camera and to the other monitors on set.

smaller connectors, reminds me of usb vs micro usb. The smaller it is the worst it gets. Also bnc is "industry standard"...

gwelkind: I think that is a separate project.

You mean that OpenCine is separate than apertus, and that this IRC isn't the place for such a discussion, or that RAW-PP Blender extensions and OpenCine are separate projects?

gwelkind: Everything except the IRC part.

Sure, I'd love to hear about it, take your time

gwelkind: However I know my way around post production well enough, and have been asked / dealt with very similar questions many times over the past decade.

gwelkind: So the first question is the one that is hidden in your original question, and it has to do with implied positions: What do you believe the software should do?

Fair?

(Because in my experience, often the great huge tirades / explosions that follow all are based off of very different starting points of assumptions, and some of those are more or less optimal given a post production pipeline and implicit design goals.)

oh, what do I (specifically) believe it should do

gwelkind: That's a good starting point.

gwelkind: Because it will also reveal some of your thought process regarding a post production pipeline regarding this particular camera.

Blender's source is relatively well documented, and most people who do post production in linux right now are using Blender-- it just seems to me that, in order to maximize utility to as many people as possible, we would want to integrate with a familiar toolset and conform to a standardized UI

gwelkind: Hold. You have started at the wrong place.

gwelkind: Here is a clearer question: What software, if any, should be designed for Axiom?

(Although I would hope I could explain the answer to your specific question after this brief discussion)

Too broad.

Or rather, I worry that your answer doesn't answer the key question here.

Yes:

note that this is the #apertus channel and not the #axiom channel, although it has been used for axiom most of the time, since it was 'revived'

Bertl: Ugh.

Bertl: So I take it there needs to be a specific channel for something like this? I worry because the lines are so blurry.

Bertl: I'd worry in that a typical touchpoint might be this channel, and perhaps explaining what can be dealt with here and what won't is a good starting point.

(IE: Apertus and software for raw processing.)

So, I guess something that enables end users to perform color correction, assemble clip sequences and apply filters, it should abstract them from technical aspects while being modular enough to allow them to do nuke-like compositing, or script their own filters and extensions

to be honest, I haven't even looked at it yet :)

Bertl: Exactly. :)

Bertl: Heck, there are things on there that are actually smashing in the face of much of what we have chatted on about in there regarding some of the camera needs / issues.

gwelkind: Let me help...

Perhaps this is because I don't know what RAW enables, besides more control over color correction

OK, listening

Let's talk briefly about that.

Data format: We would need to deal with many different color spaces. Many different file formats (for going to, not just from)

Once we have your data ingested, we will need to tweak color. Three sliders isn't sufficient obviously. It would require much more complexity. Scopes. Color space transforms.

Then we will want to separate control of one color region from another.

(aka a secondary grade)

Further

We will need to decide if this is purely a 'fake it look' tool for dailies

If the latter, we now have to figure out how to get from an offline edit to our online grading

Which also rolls in complexities about interchange formats like Avid Log format, EDL,

you lost me on the last line

Two differnet points where color are involved

(aka Dailies, Workprints, etc.)

2) When you have fully cut your picture and want to go from the workprint to your final ready-for-delivery data.

OK, that's offline edit vs online grading?

Yes.

I didn't know that, makes sense though

Online means - "Final data."

The needs of cinema / motion picture work are quite different.

(and of course, Foucault "quality" varies on context)

It is an oldschool term from film

you printed a workprint using a qucikie grade (called a one light) and you edited using your workprint

gwelkind: Sense?

The term sort of makes sense here too because your editor doesn't really care about yoru image quality.

And they can't exactly edit in realtime on your high grade 112414k images etc.

right, OK, I'm with you now

Nor would you really want to be twiddling on your raw frames.

Right, of course, a functional stand-in

Yes.

Does our starting question of color correction tool, need to be logging (the early part) or online grading.

Because even the former is a rather formidable task as you may see now.

And that is assuming you have a team with the knowledge and ability to even execute it.

Fair?

gwelkind: So let's go a step back before that then and realize that we aren't quite "there yet" to discuss the design needs of either of those pieces of software. MAYBE a logging grading tool for 'one lights'... but even then.

gwelkind: You want to shoot. Then you want to edit.

I would do an assembly cut, just get things in the right order

(and I might seem a little harsh here, but I've watched at least a DOZEN airey fairy ideas of "LETS MAKE AN NLE" come and go in Libre / open source that I immediately worry about the design contexts.)

You don't even have the footage off your camera yet.

Oh, ok, well I guess I would want to extract the RAW files onto a backup medium, then convert RAW files to some stand-in format that's easier to work with, like we discussed

Ok great.

probably need something to go from AxiomFoobar to ???

Agree?

And that question likely takes us to other contextuals: What are people going to edit on?

Yep.

Or we could be a little more liberal and think about some basic output format that are suitable for editorial in typical editors. Say... Avid, FCPX, LightWorks maybe...

Sound decent?

h264 is not that bad :)

gwelkind: Well the design constraints of DNxHD (of which ProRes is effectively a knock off of)

Got it

(long discourse there I'm sure)

DNxHD

Ah, right

and some of that, if we are focusing on cinema quality, are nuanced

Bertl: dd?

gwelkind: Copy.

a unix tool to basically copy data streams, with seek and skip features

(Further down the rabbit hole)

We agree we will probably need to go from ApertusRaw to DNxHD

we have to ask some follow up questions given our context of editing...

(this is effectively like our oldschool lab scenario where the lab would bake a 'one light' into the workprint.)

or we skip that and have only a tool that dumps to say, a more capable and known format for raw such as DNG, then have another wedge to deal with that DNG to DNxHD step for example

and 'bake' it into the DNxHD for editorial.

right, wouldn't that be easier? You aren't losing fidelity converting between these formats, right? They're all lossless

It could be 640 x 480 if you want.

Ok, Ok, we're just on (1) right now, I follow you

(and if you are unfamiliar with typical pipelines, the files off the camera are archived and are only referred to long after the edit is finished and Vfx or conforming is required)

short question here: is there a standard 'handle' for each original/raw frame?

Bertl: In what respect?

i.e. something like a hash or frame number or so?

If you are designing a grading app, then absolutely.

As you would need to scrub the footage.

Even the chroma scaler will need something like that.

:)

gwelkind: Sense?

or to be precise, which identifies a frame unambiguously (I'm asking how it is done today)

Bertl: If you shot to say, ProRes... that's in the codec.

okay, let me rephrase, you do the raw shoot, create the 'edit' version and put the raw into cold storage, yes?

Bingo.

then you do some editing, and once you're done, you 'apply' this to the original raw

Bertl: Hiero and such are wedges that serve that sort of purpose.

In terms of organizing your footage and layers of processing.

in a smaller, you can do it from say, Resolve

Bertl: EDL :)

*Googles EDL*

And if you screw that up, Resolve can even match frames on media it understands.

gwelkind: At this point... I certainly _hope_ that at least part of your question is answered.

okay, so there doesn't seem to be any (open) standard for identifying the frames ...

Well... let's flip that over

What if it is not modular.

"Use X camera, Use Y software."

(Assuming you can even write that software (Hint: You can't.))

(Botch up a lot of things you and I haven't even imagined because we are human and can't see certain shooting contexts.)

gwelkind: And that is even assuming (your original question) that Blender is even acceptable at doing such a thing. (Hint: It isn't in many respects.)

gwelkind: Only because I am about 1% informed on color and data and other things, and even _I_ can see how much of a monumental task it is to replace an entire ecosystem of tools.

gwelkind: And frankly, that is likely because people have never actually tried designing something themselves. (And by design, I extend that into say, graphic design etc.)

It's another thing entirely to make it and encounter the obstacles.

gwelkind: Even after this brief little chit chat, go read https://www.apertus.org/opencine and look at it with new eyes.

Bingo.

But frankly, the job is too big if you try to do everything for everyone.

So small steps.

1) Set in / out points in footage.

So this is also now a logger / trimmer for removing heads and tails of slates etc?

So now what format are we talking? Float?

Realtime now hits a new design constraint

And we haven't dealt with color transforms (OCIO would be needed, as well as a whole set of GUI elements)

:)

Hahaha, right, well the task at hand seems much more reasonable then

(as the contextual needs aren't dealt with. How do you do visual effects work in OpenShot? Oh yeah...)

gwelkind: Well I've used Blender enough to say "Extensively" I suppose.

gwelkind: And I can tell you it has some pretty ... uh large potholes.

And believe it or not, some of the most glaring potholes no one even knew existed (Hell... color management wasn't even going to be where it is today until they tried to use the F65 :) )

Absolutely.

For example, how the hell do you use timecode in Blender?

Man, I wish I could tell you.

How could you take an offline edited elsewhere and load it into the VSE? You can't.

(By hand)

It's why so many "ZOMFG WE NEED NLESSSSSSSS" discussions fall to crap\

(that probably sounds like a ridiculous egotistical statement, but darnit... I've been watching this cruft for so many years)

Cut.

XD

Beautiful. Thanks for that troy_s, learned a lot from this discussion and it's tangents

gwelkind: Glad it didn't seem like a mad lunatic's dribble.

gwelkind: The _reason_ it is done like that often is more than just pure anachronistic stubbornness.

btw: timecode = XML of files and their in/out points?

gwelkind: Not always XML

(With many variants that cause fits in softwares in very weird and unique ways.)

(FCPXML was a very well documented and useful format until FCPX as I understand it. FCPXML was very well specified.)

i never though fcp was something professional... same level as ms-word in terms of printing... where pros use at least latex or so

gwelkind: Try this http://mango.blender.org/production/first-original-footage-frames-in-4k/

guest: Many features have been edited on FCP and Avid.

guest: The tragic part they are missing is that YES they are edited on there (and the editing tools are well known for their streamlined efficiency (or at least prior to FCPX))

:)

they could do it on paper, for god sake.. fcp is there for "no" reason :)

guest: Well... handling a large project in terms of editing is no small feat (heck there are _editors_)

guest: Editing is about cutting. And managing cuts. And organizing.

i was expecing lightworks to come with some nice features

but they cant do DNG and 4K yet :/

guest: Why would you need 4k or DNG in an NLE exactly again?

(Hint: You don't.)

so where you mix your edl and original fullres content?

guest: A Resolve suite, a Baselight station, Lustre, you name it.

lets say just editing and color grading, no vfx and we can leave out audio too

Editing is different from grading.

The NLE comes before grading.

1) Shoot

3) Rough/Fine/Final Cut.

5) Post production (music / sound / vfx)

7) Delivery

what is between 6 and 7 then?

Where 2, 3, 4 are the domain of the NLE.

The picture is locked.

A bigger deal on projects with more visual effects.

hahah, yeah, understood

And often people like to say "But we are a small project". Even on a small project, chances are you aren't scoring the film. Nor doing the sound effects.

Maybe someone else is doing the music and sound.

Nor would you expect an NLE to say, track markers for visual effects or a grade.

(Or it is operating on crappier data.)

have to go.. will read it tomorrow

guest: It can be done, but the tech always exceeds the software.

You just get your software up to 2k realtime and boom, 4k Sony shows up (which is 8k Bayer)

But until you actually _try_ to accomplish things, you can't quite appreciate why things often are done the way they are.

Bingo.

It teaches so much.

really must go, see you later

gwelkind: Anyways... did your original Blender question get answered?

I saw the GUI mockup on the website and got the impression somehow that OpenCine was trying to be a standalone post processing suite

I still have to read your OCIO link, but I'm off to dinner with some mates. Talk to you later, I hope, tysm!

gwelkind: PM