REC709 Look-up Tables

[JRoberts86]

Hello,

Does anybody know if there is a REC709 Lookup Table available for Houdini11?

Many Thanks

J

[bunker]

Don't know if that helps, but you can load a .blut file in mplay.

Hello,

Does anybody know if there is a REC709 Lookup Table available for Houdini11?

Many Thanks

J

[JRoberts86]

Don't know if that helps, but you can load a .blut file in mplay.

Hi Julien,

Thanks for repling to my post. Yeh, I noticed the .blut files for MPLAY and COPS just can't seem find one for REC709. I can remember there being a way of creating look-up tables in CHOPS but I can't remember the workflow for it.

Thanks again

J

[gaurav]

Hi Julien,

Thanks for repling to my post. Yeh, I noticed the .blut files for MPLAY and COPS just can't seem find one for REC709. I can remember there being a way of creating look-up tables in CHOPS but I can't remember the workflow for it.

Thanks again

J

Its pretty simple with cops.

Assuming that REC709 is really a fancy name for a transfer function defined as a power function with an exponent of 0.45.

You can simply create a linear ramp in cop and append a function node with function x^E and dial in the 0.45 as exponent just to check how it behaves.

Now right click on the function node and choose save LUT.

Cheers!!

Edited November 2, 2011 by vectorblur

[JRoberts86]

Hi Vector Blur,

Thanks for replying to my post. This workflow is perfect but I am not sure REC709 is a power function of 0.45. As this is effectively the same a gamma 2.2 as 1/2.2 = 0.454545 which is closer to sRGB.

I have tested a linear image in Nuke and the difference between sRGB and REC709 appears to be in the dark areas where REC709 seems to produce a more contrasty image, so I am little confused when I search the internet about the REC709 function and they say it should be 2.35-2.4 as this boosts the intensity of the image, giving more of a washed out look which is not what is represented in Nuke.

Saying this though I am finding it hard to find a good definition of what REC709 is actually doing so I may be incorrect in what I have mentioned above.

I will carry on hunting and hopefully come up with a solid answer.

Thanks again Vector Blur, your workflow is the perfect process for tackling this issue.

Many Thanks and Kind Regards

J

[gaurav]

Hi Vector Blur,

Thanks for replying to my post. This workflow is perfect but I am not sure REC709 is a power function of 0.45. As this is effectively the same a gamma 2.2 as 1/2.2 = 0.454545 which is closer to sRGB.

I too wasn't sure and thats why i assumed it to be only a power function but i think at lower intensity values function becomes linear with constant slope and may be that could be a reason for mismatch.

Not sure about it.

Please share if you find the right answer.

[JRoberts86]

I too wasn't sure and thats why i assumed it to be only a power function but i think at lower intensity values function becomes linear with constant slope and may be that could be a reason for mismatch.

Not sure about it.

Please share if you find the right answer.

Hi Vector Blur and anyone who is interested in lookup tables for Houdini.

You can find a detailed explaination on how to set these up on the following SideFX forum post.

http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=23700&start=0&postdays=0&postorder=asc&highlight=

Regards

J

[symek]

Good way of dealing with a subject inside a Nuke based pipeline, is to export a LUT from Nuke, which can save it in a couple of formats, including Houdini's *.blut. It pretty much guaranties your Houdini and Nuke artists look on the same image.

[Macha]

Its pretty simple with cops.

Assuming that REC709 is really a fancy name for a transfer function defined as a power function with an exponent of 0.45.

You can simply create a linear ramp in cop and append a function node with function x^E and dial in the 0.45 as exponent just to check how it behaves.

Now right click on the function node and choose save LUT.

Cheers!!

That's a neat trick Gaurav! I didn't know we can do that. Wouldn't this curve have to be different across the color spectrum? Say, different values for r,g,b?

[gaurav]

That's a neat trick Gaurav! I didn't know we can do that. Wouldn't this curve have to be different across the color spectrum? Say, different values for r,g,b?

Hi Macha,

As far as i know a single gamma correction is applied to each of the components to get nonlinear RGB. Its a standard practice to compensate for the display device's non-linearity.

To me its also a very clever compression technique to capture high dynamic range of real scene so it gets compressed with in the fairly low range of a video signal (when compared with real world).

Highs gets compressed more in order to leave space for low end of values to greet the non linear response curve of human eye receptors. Which reduces as the light increases.

So if the source (video capturing device) is performing nonlinear conversion separately on each channel, than we do need to apply different corrections per channel based on how they got encoded the in the first place.

But this is how i understand gamma correction in video space considering how poorly defined the entire video format is.

Cheers!!

Edit - Thank you James for sharing info.

Edited November 8, 2011 by vectorblur

[symek]

That's a neat trick Gaurav! I didn't know we can do that. Wouldn't this curve have to be different across the color spectrum? Say, different values for r,g,b?

The accurate rec709 as sRGB should have a curve per channel, but differences between colors are minimal, so people often omit this minor detail.

[axebeak]

Hmm, my understanding is that as long as your linear RGB values are according to a specific set of primaries, then no further per-channel adjustment is necessary (and Rec709 and sRGB primaries are the same). If RGBs are in different space then they must be converted, but this is not a part of _this_ transfer function transform. Finally, if primaries are not known then it's not possible to reproduce the colors "correctly" at all.

But, I'm not a video or color expert, so it's certainly possible that I'm missing something.

BTW, the best source of information on these matters is probably Charles Poynton's book: http://poynton.com/DVAI/index.html

Edited November 8, 2011 by axebeak