#Rawdigger green raw export full
#Rawdigger green raw export code

I would like to use a Log Encoding but i am unable to find a workflow that is working for me.ĪCES appears to give me a better colour result but the result of an ACES convert via the CinemaDNG IDT gives me a result that is more underexposed, by quite margin (i need to lift it 4 stops in Nuke). I need to get my linear DNGs into Nuke in a container i can work with in my pipeline. So long as the colour workflow is correct. Side by side these cameras offer remarkable range to dSLRs. Of course compared to most dSLRs that we think are wonderful their ranges are limited in the same way. The Bolex and FS700 both have linear 12 bit output, pretty much real linear and whilst you can technically see >12 stops for the 700, that is range that is just there to check a box, you can't do much with it. Or how highlight recovery is a really important part to extending headroom. It's quite telling - i won't even go into how range is increased because the sensitivities of RGB are so, so different. There's a nice utility called RawDigger which can show you prelinearised data within a DNG, you can poke around yourself. But even so, there isn't a mass of detail in the shadows in *any* of these cameras. The Linearisation LUT here helps by robbing some values from that for the mid tones. Remember again that the brightest stop is represented by 2048 out of 4096 possible values in 12 bit linear. Even with a 12 bit log curve that shadow area isn't increased much, the log curve affects the highlights mainly.

#Rawdigger green raw export code

In a true 12 bit linear file the bottom 4 shadow stops are represented by code values 0 to 15 (out of a maximum of 4096 - less when you include black levels).

I think people get a bit obsessed about dynamic range, there's a huge difference between technical range and usable range.

Looking at the very raw data, pre debayer, i don't think it's a log curve but certainly there's >12 stops of information there but at the same time certainly not 22 bits! Indeed the BMC DNGs are 12 bit but unusually they hold a linearisation LUT which helps transform those 12 bits into 16 bit linear. I've spent a lot of time digging around in DNGs, so some of this may be useful. I'm chiming in here because there's some good understanding of how resolve is working and i need to translate this to the DNGs from the FS700/O7Q combination.

#Rawdigger green raw export full

The two 11bit values are multiplexed on the sensor chip into full 22bit values. If one describes a multiple for example, you can describe even more than just the added possible values.

The same is true for how you can combine two digital values. Separately they can only give me limited info, but combined they provide additional and exponential information.Īdding one to the other did not expand the possible data outcome linearly. If i could describe whether something is red or not, as one or zero - a one bit value - and i could also answer whether something is green or not in the same way, those two 1 bit values could describe any of the following: Research combining two 8bit bytes into a 16bit value. It is true that adding another 8 bit steam does still mean you only represent an additional 256 values - but you can use algorithms so that the second set of values is able to describe numbers equal to 16 bit numbers. The information is almost quoted directly from white papers from the sensor manufacturers for both the bmcc and alexa.Ī single 8 bit stream can describe 256 different values. Just as it is possible to represent a 16 bit value with 2 8bit 'bytes', it is also possible to represent 22bit values with two 11bit streams. Tom, that is like saying 500,000 and another lot of 500,000 equals 50,000,000,000 bits of information.