The Rabbit Hole

« first day (396 days earlier) ← previous day next day → last day (572 days later) »

Mareck

04:20

Hi
@troy_s it's funny to see all what's happening since Andrew post his video, it's remeber me that peoples saying great things can be crucified, but be reassured one day you wil have a book speaking about you ^^.

More seriously I have two question in mind

First one I don't remember if I already asked you, rec.709 is an other name for BT.709 isn't it?

Second one, isn't it confusing to have the filmic log encoding in the color space section of the image editor?

troy_s

04:37

@Mareck Correct. I believe the full specification is Recommendation BT.709 or so

@Mareck Well what happens if you save a TIFF with Filmic Log Encoding and then you give it to someone and they need to load it and linearize it?

;)

(In an ideal world, we would be using EXRs and DPXs, but DPXs are still broken. You can run a whole pipeline almost exclusively on those two file formats.)

Mareck

I don't say it isn't usefull but just that it end in the color space.

What can be confusing I think, but just to say.

And thanks for the BT, one less doubt ^^

In fact it's more the name of the section that is confusing, would a correct name be transfer functions?
It's just for me, to confirm knowledge and understanding.

troy_s

04:52

@Mareck OCIO leaves it up to you to name as you wish. There is a "family" description that can be used by an application to group transforms.

@Mareck Other than that, there is no real organizational component. It is just a bag of transforms.

Mareck

05:15

I haven't realised that you gave me the info to properly answer a question on SE.
Must go, see you.

10 hours later…

Mareck

15:18

@troy_s just for my knowledge and to don't say mistakes, about image editor.
Let say I've saved an image with the filmic look, I import it to image editor, if I set the "color space" to filmic, is it true that I say the transformation where:
linear unbounded to BT.709/Log C (or filmic log) then BT.709/Log C to linear 0/1 bounded?
If so is it why we don't have a 1/1 between scene referred and "color space" filmic log encoding?

3 hours later…

troy_s

17:57

@Mareck Colour space should be reserved for something that encapsulates a colour space. I typically adhere to the definition in the ISO 22028-1 standard, which says a colour space must have the three components we know. I suppose given that we can define all three things in Filmic, it could be considered one, but I find it to be a little tricky. I am simply hesitant to call it such.

It is more about the transfer functions, with a "borrowed" white point and primaries set. :)

LogC is Arri's log.

Not Filmic's.

I am not quite clear on your other question though.

The range covered in Filmic (-10 to 6.5) would be decompressed from Filmic log to scene referred linear when you correctly identify the colour transform as Filmic.

It isn't quite 1:1 with say, an EXR because it has the desaturation baked in as well as would have anything beyond the -10 to 6.5 range chopped off.

2 hours later…

Xernist

19:50

Hey people of the internet, i got a question: How different lightsources get added together? (Example: 2 different lightsources hitting the same spot on a (of course not perfectly) white diff. surface before getting reflected from that surface to the camera.) Of course the different intensities are added up but how are the albedos getting combined? Im a totally beginner in this field coming from 2d digital painting but want to learn more about realistic-ish rendering in general.