Why does the film scan DPX appear so contrasty and dark

[silvan schnelli]

I downloaded 5213 35 film scan DPX footage and ARRI XT footage from https://www.cinematography.net/Valvula/valvula-2014.html. My goal is to compare the footages and put them through a post pipeline that I am amending and tinkering with. However, when I loaded the two DPX files into Nuke, I noticed that the film scan is much darker and slightly more contrasty than the Alexa footage. I then applied a OCIOFileTransform to both, a LogC2 to rec709 conversion for the alexa and davincis Kodak 2383 PFE for the film scan. The colours look great but the film scan is much more underexposed. On the CML website the footage looks relatively the same exposed, but for me the film looks much more underexposed. What am I doing wrong or misinterpreting?

ARRI XT (16bit) with classic LogC2 to rec709 and Kodak Film with Davinci 2383 PFE 

 

[Robert Houllahan]

On 4/7/2024 at 8:09 AM, silvan schnelli said:

I downloaded 5213 35 film scan DPX footage and ARRI XT footage from https://www.cinematography.net/Valvula/valvula-2014.html. My goal is to compare the footages and put them through a post pipeline that I am amending and tinkering with. However, when I loaded the two DPX files into Nuke, I noticed that the film scan is much darker and slightly more contrasty than the Alexa footage. I then applied a OCIOFileTransform to both, a LogC2 to rec709 conversion for the alexa and davincis Kodak 2383 PFE for the film scan. The colours look great but the film scan is much more underexposed. On the CML website the footage looks relatively the same exposed, but for me the film looks much more underexposed. What am I doing wrong or misinterpreting?

ARRI XT (16bit) with classic LogC2 to rec709 and Kodak Film with Davinci 2383 PFE 

 

16bit DPX is linear not LOG.

[silvan schnelli]

@Robert Houllahan Thank you for the simple and quick reply. It is something I only realized later through trial and error. Nevertheless, for some reason the footage still appears a bit off, but changing from log to linear fixed the issue for other film footage I have downloaded.

Edited April 11 by silvan schnelli

[Robert Houllahan]

Cineon LOG is a kind of lossless compression designed by Kodak to follow the characteristic curve of how negative film works, so the density range could be put into 10 bits.

All digital sensors are linear response and early Cineon and Quantel scanners used Kodak tri-linear (RGB LIne) CCDs which were 12bit or 14bit linear, the Arriscan is a 14bit ALEV sensor and 2-flash gets it to 16bit precision.

So all LOG curve schemes have followed on from the Cineon LOG designed by kodak in the late 1980's

There is still calibrating the scanner to the film base and both exposure and lab variations and I think there is no way to just add technical and look LUTs to get a perfect image without also doing some base grading.