Green color on Kodak 5219

[Anton Meleshkevich]

Why are all 5219 emulation LUTs on the market shift green to yellow? Real 5219 film stock can reproduce green colors good, almost without any shifting its's hue.
Am I missing something?

[Bryan Tran]

I'm also curious about this.  I thought I was the only one who noticed this because no one else has said anything about this online until now.

[Amada Daro]

Is the same thing happening with the K5219 LUT that comes with Resolve? Most of those available are simple copies of previously generated ones. I will look into this. 

[Anton Meleshkevich]

As far as I know 5219 doesn't come with Resolve. Do you mean 2383?

Edited February 2, 2018 by Anton Meleshkevich

[Lowepost]

Hi Anton and Bryan,

I am not sure , possibly due to they way the lut was built.

If you are using a "true 5219" lut then the input would be log scene exposure to then impart the properties of 5219 into the image. However, if the scene exposure was not handled properly to account for the spectral response of the film, errors can arise. Remember, log scene exposure is not identical with balanced log Cineon CV input. It may or may not be depending on the methods used. To get log scene exposure from a 5219 film scan, the data is "un-built" back through the film tone scale, conversion to printing density, removal of inter-image cross channel effect. This is not for the faint of heart. So, if Cineon balanced CV were put into a "real 5219" lut it is possible to get color shifts. Also, maybe all the existing luts have their genesis from the same error prone source. If used properly, 5219 can produce reasonable greens. The main problem with saturated greens from film in general is the print dyes. To make green it needs lots of yellow and cyan dye in the print. However, each of those dyes tail into the green some and have the effect of reducing the saturation. Due to the dye set it is unavoidable. In a lut, to get a yellowish green there is insufficient cyan dye to balance the yellow dye. That seems to be a lut color balancing or modeling issue.

Hope this helps

 

Cheers Mitch

[Bryan Tran]

Thank you Mitch! That's really helpful, but I have a couple of questions.

What is "log scene exposure" and "balanced log Cineon CV input"?.

Please forgive my ignorance, as I'm just a film student.

Also, I've noticed yellowish greens from 2383 LUTs as well. Even with the LUT provided by Light Illusion, which is based on data from a highly accurate spectrophotometer and densitometer. Maybe @Steve Shaw can chime in? I remember watching Dunkirk in 70mm and seeing "green" greens, in fact, very blueish greens, in the beginning. So I was wondering why I can't seem to get the same with these print LUTs.

I know I shouldn't be using LUTs anyway, but besides the yellowish greens, I love the way footage looks when grading under a PFE LUT.

I'm rambling, but thanks again, Mitch!

[Lowepost]

On 2/3/2018 at 10:04 PM, Bryan Tran said:

Thank you Mitch! That's really helpful, but I have a couple of questions. What is "log scene exposure" and "balanced log Cineon CV input"?.

 

The log scene exposure is the negative log (base 10) of the linear lightness of elements in a scene. By convention, the 100% scene white (as 1.00 brightness) is - log(1.0)  = 0.0 Log exposure and the 18% gray is -log(0.18) = - (-0.7447) = 0.7447 Log exposure and a 20% gray is -log(0.20) = - (-0.69897) = 0.69897 Log exposure and a 2% black is -log(0.02) = 1.69897 Log exposure.

The gamma of film is nominally 0.60. So the Log exposure needs to be multiplied by 0.60 to obtain Log scan densities. In Cineon terms the 0-1023 CV yields a scan density range of 0.0 - 2.046 where each CV is 0.002 scan density (referred to as printing density). To find the real Log scene exposure range that can be recorded the 2.046 scan density is divided by 0.60 = 3.41 Log scene exposure. Actually since the min. density is set at 95, the range is 95 - 1023 = 928CV =  928 * 0.002 = 1.856 which is then divided by 0.60 = 3.0933 Log scene exposure.

What is meant by "balanced log Cineon CV input" is the CV associated with a scan that has had a R G B constant either added or subtracted to achieve a neutral gray at near LAD  (445CV and corresponds to a 16% gray).

LAD is usually 445CV which is  is relative to the minimum scan density (0.0) by convention set at 95CV. Thus, (445-95) * 0.002 = 0.70 scan density. Since the scan density excludes any RGB minimum density (the scan density with no exposure), in this case the LAD scan density is 0.70 above the minimum base density.

In Cineon CV a 1 stop increase in exposure (0.30 Log scene exposure) is * 0.60 = 0.18 scan density difference, which if divided by 0.002 yield 90. Cineon CV per stop of original neg exposure. In the older printer lights , 1 printer light = 0.025 scan density or 0.025 / 0.002 = 12.5 Cineon CV, therefore 90CV ~= 1 stop exposure = 90.0 / 12.5 = 7.2 Printer points per stop.

So, to adjust an image in a color corrector that is stored as Cineon CV, for example, +2 PP (Printer Points) Red (brighter Red), -1 PP Blue, adjust image by adding  2*12.5 = 25 CV Red and subtracting -1 * 12.5 = -12.5 CV Blue

I hope this is not too confusing.

Cheers Mitch

[cameronrad]

4 hours ago, Lowepost said:

The log scene exposure is the negative log (base 10) of the linear lightness of elements in a scene. By convention, the 100% scene white (as 1.00 brightness) is - log(1.0)  = 0.0 Log exposure and the 18% gray is -log(0.18) = - (-0.7447) = 0.7447 Log exposure and a 20% gray is -log(0.20) = - (-0.69897) = 0.69897 Log exposure and a 2% black is -log(0.02) = 1.69897 Log exposure.

The gamma of film is nominally 0.60. So the Log exposure needs to be multiplied by 0.60 to obtain Log scan densities. In Cineon terms the 0-1023 CV yields a scan density range of 0.0 - 2.046 where each CV is 0.002 scan density (referred to as printing density). To find the real Log scene exposure range that can be recorded the 2.046 scan density is divided by 0.60 = 3.41 Log scene exposure. Actually since the min. density is set at 95, the range is 95 - 1023 = 928CV =  928 * 0.002 = 1.856 which is then divided by 0.60 = 3.0933 Log scene exposure.

What is meant by "balanced log Cineon CV input" is the CV associated with a scan that has had a R G B constant either added or subtracted to achieve a neutral gray at near LAD  (445CV and corresponds to a 16% gray).

LAD is usually 445CV which is  is relative to the minimum scan density (0.0) by convention set at 95CV. Thus, (445-95) * 0.002 = 0.70 scan density. Since the scan density excludes any RGB minimum density (the scan density with no exposure), in this case the LAD scan density is 0.70 above the minimum base density.

In Cineon CV a 1 stop increase in exposure (0.30 Log scene exposure) is * 0.60 = 0.18 scan density difference, which if divided by 0.002 yield 90. Cineon CV per stop of original neg exposure. In the older printer lights , 1 printer light = 0.025 scan density or 0.025 / 0.002 = 12.5 Cineon CV, therefore 90CV ~= 1 stop exposure = 90.0 / 12.5 = 7.2 Printer points per stop.

So, to adjust an image in a color corrector that is stored as Cineon CV, for example, +2 PP (Printer Points) Red (brighter Red), -1 PP Blue, adjust image by adding  2*12.5 = 25 CV Red and subtracting -1 * 12.5 = -12.5 CV Blue

I hope this is not too confusing.

Cheers Mitch

Holy crap, this just cleared up so much confusion for me. Thank you!

Question I have. I've seen multiple Cineon ICC profiles that use the sRGB/Rec.709 color primaries. I know ICC profiles aren't the right tool for describing the complexities of film,, but I was wondering if it was standard convention to encode/assume Rec.709/D65 - Cineon Tone Curve for Cineon files. Or would there be instances for ProphotoRGB - Cineon, P3 - Cineon. etc 

 

Edited February 10, 2018 by cameronrad