Broadcast Rec709 -> sRGB 2.2 Unicolor Transformation

[mail18]

I am trying to figure out the best practice for transforming a TV Commercial graded for 709 broadcast to online specs sRGB 2.2
I end up with a different result if I compare the sRGB version on a sRGB corrected display vs the Rec709 version on a Rec709 display.

 

If I use the Unicolor to make the transform see image for settings.

 

 

sRGB Version

 

Rec709 Version

 

The rec709 version is super close on the sRGB display to the version of the Rec709 Display.
While if I make a export from a rec709 grade on Resolve I always need to apply a Color Transformation node to transform the gamma to make it look the same.

Does anyone know the best practice for this situation in Mistika?

 

 

 

 

[jan1]

Not sure what the right answer is. But I have found the Unicolor node to yield sometimes surprising results. It seems straightforward and I would have configured it the same way you did. 

That said, I was just working on a commercial job shot on Arri, and when I apply the Arri Gamut/Log to the footage, which is definitely log material, the results are over exposed and I have to bring it back into range. I wasn't on set, so I can't guarantee that it was shot correctly, but I would assume yes. So there seem to be cases where I'm not sure I can 100% trust the Unicolor transforms.

You could create a LUT in LUTCalc or your favorite tool and load it via the LUT3D node instead and see if this is any better.

[agonzalez@sgo.es]

Hi, 

Have you tried using Gamma 2.4 with Rec 709 instead of the ITU 709. Normally the grading that is done in 709 (and the settings of the Color Monitor for this kind of grading), is normally done in 2.4 instead of the ITU 709. I think Resolve applies gamma 2.4 as well, as it is the real standard. 

 

[mail18]

Thanks Agonzalez! I am gonna try out that option!

[mail18]

Only I have been told in the past that I need to convert the Rec709 2.4 gamma to 2.2 when going to sRGB.

Quote from wiki

"It is often casually stated that the decoding gamma for sRGB data is 2.2, yet the above transform shows an exponent of 2.4. This is because the net effect of the piecewise decomposition is necessarily a changing instantaneous gamma at each point in the range: it goes from gamma = 1 at zero to a gamma of 2.4 at maximum intensity with a median value being close to 2.2. The transformation was designed to approximate a gamma of about 2.2, but with a linear portion near zero to avoid having an infinite slope at K = 0, which can cause numerical problems. The continuity condition for the curve , which is defined above as a piecewise function of , is"

 

Now I am totally lost since I thought it was 2.2 

[jan1]

11 hours ago, agonzalez@sgo.es said:

Gamma 2.4 with Rec 709 instead of the ITU 709

That's an interesting point. In Unicolor there are two choices - 'Gamma 2.4 - ITU BT1886' and 'ITU BT 709/ITU BT 2020'. There's no other setting that includes '709'

I had been using ITU BT 709 which seemed logical based on the naming. When I just tried them both, there's quite a bit of a difference. 

In my example I'm going from Arri Log C / Arri Wide Gamut to a Rec 709 calibrated reference monitor. What would be the appropriate output gamut/curve? So far I've been using 'Rec 709 D65' and 'ITU BT 709/ITU BT 2020'.

[jan1]

9 hours ago, mail18 said:

Rec709 2.4 gamma to 2.2 when going to sRGB.

It's tricky business. Technically speaking all GUI monitors are 2.2, not 2.4. TV and reference monitors are 2.4. Which often leads to the mismatch when people grade and then watch things on a GUI monitor. Of course, with the right software (or OS setting) you can also change your GUI monitor to 2.4, just to make it messier ?

That being said, most browser based viewers like Vimeo/YT expect videos to be in 2.4 and the monitor in 2.2 and should apply the appropriate conversion. However, ever since browsers started using hardware acceleration for video playback most browsers have had heaps of bugs and are all over the place on this matter.

That said, I generally deliver online video content in 2.4, rather than guessing what their viewing environment is. Doing the conversion on your end to 2.2 is likely making matters worse, not better.

Edited October 24, 2019 by jan1

[mail18]

Thanks Jan, that's a clear answer on my gamma struggle. 

[Yoav Raz]

HI all

I use the output curve gamma 2.2 and the output gamut on rec709

this gives me the best of both world.

But i use unicolor just for transforming between color spaces. mostly P3 to Rec709

From log_c to Rec 709 i use lots or color grade hand made luts.

 

PS: don't know if you remember that you can use Lut's inside a Color Grade Vector (layer)

Vector > Grade Inside/outside > Lut > go to option and you can change the opacity.

Hope it Help

Yoav

[jan1]

On 10/24/2019 at 4:39 AM, Yoav Raz said:

But i use unicolor just for transforming between color spaces. mostly P3 to Rec709

From log_c to Rec 709 i use lots or color grade hand made luts.

I'm assuming, though I don't know for certain, that Unicolor does not use any LUTs internally, but instead the proper color space primaries and transforms. That is always preferable over the interpolation of a LUT. So to the extend possible I would always use Unicolor rather than a LUT, at least when it comes to the basic transforms for scene and display referred grades. Then use LUTs for any creative elements only.