HaldClut issues with 5.5 rc1

haldclut

(Stefan Chirila) #1

Help please…

So I was trying to make a Hald Clut from a preset I made earlier today. Picture 1 is the preset (with sharpening and such turned off). I then copied the preset and pasted it onto a newly created hald file made with imagemagick. Then Picture 2 is me then setting the cat picture to neutral and Picture 4 is what came out of applying the created and saved TIFF 8bit Hald Clut :frowning:

I don’t get why the white balance is so terribly off.

Attached the preset and hald. Hope It’s me making some stupid mistake and not RT being off.

Or CLICK HERE to download the clut.

TEMPLATE–hald12_16bit.tif.out.pp3 (12.3 KB)


RawTherapee 5.5 is one week away
(Alberto) #2

Hi,

You are “baking in” a custom WB in the CLUT, which is not a good idea. In general, as RawPedia says, you should avoid using WB for artistic/creative effects, as this can have unwanted side-effects. This is one instance of the problem.

Your CLUT:

Your CLUT, without WB:


(Stefan Chirila) #3

@agriggio I did eventually realize that taking out the white balance settings helps get similar results between editing the raw or using the hald clut; however I still end up getting duller bluer colours (unfortunately can’t post a sample right now). Are you able to produce a Hald from those settings (with WB, sharpening removed) that looks identical to the RT edit? I’m thinking maybe it’s assigning some strange colour space?

@heckflosse


(Stefan Chirila) #4

Figured it out! It’s the output colour space. Has to be sRGB and wasn’t :slight_smile:

Cat-lion-yellowmane-byMe-fuji400Hish–D700_20120731_0235.NEF.pp3 (12.2 KB)


(Morgan Hardwood) #5

@stefan.chirila speaking from memory, the only thing that matters (or the only thing that should matter) are the values stored in the HaldCLUT image, i.e. the color profile assigned to/embedded in the HaldCLUT image should not matter, however if the pixel values in the HaldCLUT were affected by the output profile during saving then that’s good. Someone would need to test that it was so in 5.4 and that it still is so in 5.5-rc1. You could help out with that.

From RawPedia:

It should not make a difference what color profile is assigned to a HaldCLUT image. What matters are pixel values, because, remember, a HaldCLUT image is just a matrix of “out” numbers. “Assigning” a color profile does not alter stored pixel values, but “applying” or “converting” does alter stored pixel values, so don’t do that.


(Alberto) #6

I think the color profile does matter. The quote is correct, but if you use RT to generate the CLUT, then the output profile is used to “convert”. In fact, the code does something to covert the clut to the working profile when loading the file, but from my (limited) tests it doesn’t seem to work properly. I think it’s better to stick to sRGB until someone figures out what exactly is broken and how to fix it.


(Morgan Hardwood) #7

It matters if it affects stored pixel values. It does not matter which profile is assigned. That is, choice of output profile when saving the HaldCLUT in RT matters because it affects the stored pixel values. Assigning a color profile to the saved HaldCLUT image afterwards does not matter.

It works correctly in 5.5-rc1, I just verified.

haldclut_assign_test.zip (1.6 MB)


(Alberto) #8

Yes, we agree. That’s why I wrote:

But from a user’s perspective, changing the output profile in RT does have a visible impact on the CLUT. That’s what I meant, sorry if it was unclear


(Morgan Hardwood) #9

I added a paragraph to RawPedia:
http://rawpedia.rawtherapee.com/Film_Simulation#Make_Your_Own


(Stefan Chirila) #10

Yeah …not sure how I managed to switch from RT_sRGB or whatever it’s called to RT_something2000 but it very much affected the resulting clut. As of right now things seem to be mostly fixed, except I do find the clut applied to the raw file to be a tad darker than the raw file with the RT editing applied. Not sure if it’s a mistake on my part, or if it’s because I’m removing sharpening and noise removal but it’s not enough of a difference to bother me too much.