AWS Thinkbox Discussion Forums

Q: Does OcioLutFile assume it's linear data, is draft float?

Heya Folks!

As ever we’re having the fun with some lut files, and for reasons of understanding and quality I had two questions:

One is that there’s an ocio colourspace operator where you can specify in and out colour spaces. In the Ocio Lut from file command though, there isn’t such a command. From doing a comparison with the nuke vector field and applying a generated lut, it seems to assume the data it’s being given is linear colour, or it seems to treat the colour this way. Is this the case?

Secondly I was wondering if draft operates in float colour? Related to the above question, I’ve had to stack a few different colour operations together and while they’re largely gamma / midtone based operations such as stacking to and from cineon colour transforms before or after an ocio lut from file command, I’m cautious in case we ever get a lut file from a client that’ll clip the colour data if used in combination with other operators! It hasn’t happened yet admittedly!

Cheers!

John

Hi John,

Here are my answers to you:

You can create an OCIO lut file with different in and out colour spaces using an OCIO baker (note that this is not possible to bake luts in Draft yet). So a OCIO lut file can have any colour space in, depending on how it was created. If you know you have a OCIO lut file that is linear->X and you’re starting in colorspace Y, here’s what you can do:

you load a linear->Y lut, inverse it (now Y->linear) then apply Y->linear, then linear->X.

Yes, Draft operates in float colour.

If you have any other questions, please let me know.

Kindly,

Julie

Thanks for the response Julie, that’s what I’m currently doing so it’s great to know that I’m not doing any damage by stacking operators.

Much appreciated!

Awesome!

Have a great day,

Julie

Pretty off topic but you should almost never use a Linear -> Y LUT. LUTs generally are clamped at 0.0 -> 1.0 so you should use mathematical transforms for things like Linear to Log to avoid data loss. One of the best parts of LOG is that it compresses negative and positive values below 0 and above 1 respectively into 0 to 1 which makes it an ideal starting point for all of your LUTs. And since it’s log it also dramatically improves the perceived resolution of your LUTs. So if you do build a library of OCIO luts use Draft’s mathematical log conversion to first convert from linear to log and then perform your color operations there. When done you can then apply a Log to Linear mathematical function to restore the original linear if going to something like EXR.

Thanks for the info Gavin! I will remember that…

Kindly,

Julie

Privacy | Site terms | Cookie preferences