pictureliner.blogg.se

That A7s shot in reply #5 looks to have some IR contamination.Ĭode: // LogC to Cineon (with highlight roll-off to preserve dynamic range) The Film matrix is rarely used except when trying to mimic film. I think also, when it comes to the Alexa 'look', a lot of it has to do with the Alev sensor's red channel response. Rec709 primaries don't look so good under print emulation so some hue rotation and increased saturation is needed to bring it to life - the Film Matrix applied to Log-C does this mathematically to get close to the look of the color in the film scan - but only when used with Log-C shot on the Alexa/Amira because, as pointed out, the Alexa has a few tricks up it's sleeve with regards to saturation. The Film matrix is simply the average difference between a scanned negative and print and Log-C under a print emulation lut of the same neg and print (I'm speculating but I remember reading somewhere that ARRI used 52 as a benchmark!?). the linear matrix adds significantly more saturation to highlights as it's working on a much larger signal range. Saturation is quite different depending on the matrix used i.e. There are a couple of AWG2Rec709 matrices for Log-C, one for linearized Log-C and another for tonemapped Log-C. The Film Matrix isn't the only Arri matrix applied non-linear.

The workaround for removing it (which should also be performed on LogC) on MixingLight involves the RGB Mixer, which is a bit on the loose side, but since that was posted a number of better options are now available.Įdit: I didn't see Jake's post, but what I've written is along the same lines. Alexa Film Matrix is unusual in that it is a 3x3 matrix that is to be applied not to a linear signal, but specifically to LogC. This, as well as Alexa Film Matrix, can be easily applied now via DCTL. By adjusting for this before converting to LogC Alexa Wide Gamut (and then applying Alexa Film Matrix), the results are consistent with an actual Alexa workflow.Ī direct conversion from LogC to Cineon can result in the clipping of the last 2 stops of highlight information, but a workaround (of sorts) is to use gamma to compress the part of the linear signal from 1.0 upwards until it fits below the 13.5 limit. Basically (according to Art Adams) the Alexa mimics Film by hitting peak saturation at around 35 IRE and holding until around 65 IRE, before falling off, whereas other digital cameras such as Sony and Canon don't hit peak saturation until 65 or 70 IRE.

I've tried recreating this process with footage from different cameras, and always had issues with excessive saturation, but I believe this is because the Film Matrix expects the actual Alexa saturation response. Of course, the LogC encoding must then be converted to Cineon in order for the Print LUT to achieve its intended response.

I believe the intended purpose of Alexa Film Matrix is to transform LogC Alexa Wide Gamut to something approaching a more traditional Cineon scan.