RED Conversion LUTs

[Greg Cotten]

Hello everyone,

My friend, Tashi Trieu, and I have made conversion LUTs for REDlogFilm and REDgammas 1-4. We used Nuke with MatchGrade and my own Mac app, Lattice, to generate 1D tone maps for converting to and from REDlogFilm and the various REDgammas.

Our Nuke script and master LUTs are available on GitHub along with a link in its description (on the bottom of the page) to a "release" ZIP file with 1D and 3D Resolve LUTs for forward and reverse transformations.

I hope this is useful to some of you - please let me know what you think and if you have any suggestions for improvement!

Greg

 

[Björn Benckert]

Hello everyone,

My friend, Tashi Trieu, and I have made conversion LUTs for REDlogFilm and REDgammas 1-4. We used Nuke with MatchGrade and my own Mac app, Lattice, to generate 1D tone maps for converting to and from REDlogFilm and the various REDgammas.

Our Nuke script and master LUTs are available on GitHub along with a link to a "release" ZIP file with 1D and 3D Resolve LUTs for forward and reverse transformations.

I hope this is useful to some of you - please let me know what you think and if you have any suggestions for improvement!

Greg

Nice!

But important to know flopping the images back and forth would only work in nuke with camera source in and NUKE 16 bit float. As if the images is for example exported from RCXP as 10 bit DPX and as redgamma4 for example there is a high risk that values are out side of the color palette and those will not be recovered when flopping back to redlogfilm with these luts.

Still a quite handy set of lut's I think.

Edit: now I downloaded. Yes going from redlog film to red gamma is not risking as much as the opposite. But possibly you should make reverse lut's as well no?

 

[Ben Scott]

Great stuff, thanks Gregg!

 

[Greg Cotten]

Nice!

But important to know flopping the images back and forth would only work in nuke with camera source in and NUKE 16 bit float. As if the images is for example exported from RCXP as 10 bit DPX and as redgamma4 for example there is a high risk that values are out side of the color palette and those will not be recovered when flopping back to redlogfilm with these luts.

Still a quite handy set of lut's I think.

Edit: now I downloaded. Yes going from redlog film to red gamma is not risking as much as the opposite. But possibly you should make reverse lut's as well no?

Björn,

I'm not sure I understand "there is a high risk that values are out side of the color palette and those will not be recovered when flopping back to redlogfilm with these luts". None of these LUTs clip so that shouldn't happen.

In the description on the GitHub page, if you download the ZIP of the "release" there are forward and reverse LUTs for every REDgamma version.

 

[Björn Benckert]

Björn,

I'm not sure I understand "there is a high risk that values are out side of the color palette and those will not be recovered when flopping back to redlogfilm with these luts". None of these LUTs clip so that shouldn't happen.

In the description on the GitHub page, if you download the ZIP of the "release" there are forward and reverse LUTs for every REDgamma version.

Yes I'm wrong. I wrote that before I actually downloaded those and yes going from redlog to redgamma is no issue. But having a redgamma image that has peaking white and then bringing it back to redlog could possibly clamp the colours a bit, or am I mistaken?

 

[Greg Cotten]

Yes I'm wrong. I wrote that before I actually downloaded those and yes going from redlog to redgamma is no issue. But having a redgamma image that has peaking white and then bringing it back to redlog could possibly clamp the colours a bit, or am I mistaken?

If you have an ungraded REDgamma image and apply a reverse LUT to get it back to REDlogFilm, there shouldn't be any clamping. There could be banding introduced if the bit depth of the source image is 8-bit.

 

[Tashi Trieu]

These LUTs are very good for projects where clients have viewed in a REDgamma encoding on set and have lit accordingly. If you want to grade in REDlogFilm but achieve the same creative starting point as REDgamma encoded dailies, these LUTs will help you do that.


We have also conducted an analysis of the accuracy of these LUTs.


The normalized mean absolute error of the LUTs as compared to a REDCINE-X debayer of the same gamma space are listed below:


LUT MAE (%)
REDlogFilm_to_REDgamma 0.0824788%
REDlogFilm_to_REDgamma2 0.0340227%
REDlogFilm_to_REDgamma3 0.0177164%
REDlogFilm_to_REDgamma4 0.0244737%

 

[Brandon J.F.]

Thanks for sharing these, Greg! Very useful.

 

[Elliott Balsley]

Sounds like a great tool. I will try these out soon. I have seen several threads of people asking for this exact thing from RED.