Scarlet / Canon / Sony frustration

[Stuart English]

I seem to remember that too. (Not that 16bits does any good for a 13 stop sensor).

Those two values are not related like that.

Bit depth describes the PRECISION at which the data you are capturing can be represented.

Stops describes the intensity RANGE over which you can capture the data.


So yes, it does have value to have 16-bits for 13+ stops...

 

[Gunleik Groven]

Definitely does!!!

I think many of the differences between R1 MX and Epic, are made possible through the extra bit-depth.

I could start to list symptoms on both cameras, but that is really unimportant. The general difference is that one feels like it is maxed out and then some, the other hums peacefully and tells me it has so much more to offer, if I push it. The highlights rolloff is one of them, though...

Those two values are not related like that.

Bit depth describes the PRECISION at which the data you are capturing can be represented.

Stops describes the intensity RANGE over which you can capture the data.


So yes, it does have value to have 16-bits for 13+ stops...

 

[Hrvoje Simic]

ok, fair enough.

Have you used or graded the C300 though? I thought it held up well to grading... Obviously not close to RAW, but it impressed me.

Yes - not much but enough, as well as the same codec coming from other cameras.

Regarding impressions: "grading" is a very relative term. It greatly depends on what kind of route one takes and how far. As I said, different types of alterations affect the codec limitations/picture "thickness" differently. Compared to even thinner 8 bit codecs this one surely has more room and is more "resistible", but I still find it very sensitive and limiting. Someone else might have individual routes and priorities which are less "codec demanding", hence the user base shooting and using the material without any issues...and someone maybe just doesn't mind some types of compression artifacts which are less visible and material destination doesn't require strict QC.

Uncompressed 4:2:2 would be a different thing, but this is Mpeg 2. Looking at a single frame grab often provides only a partial perception of some types of compression artifacts, because some codec limitations are much more visible in motion. Plus it doesn't show the post work limitations which may pop out as soon as one touches the material. Plus there is a generation loss issue. This codec was not designed for grading, but to be "good enough" to pass broadcast workflow without picture breaking apart. This doesn't mean it's totally unusable for anything but shoot/cut/export because it obviously is, it means it can look good "out of the box" but that's mostly it, and I wouldn't advise relying on post work - much.

With C300, SDI capture route is highly advisable, mostly to remove the horrid Mpeg blocking, even if it means being left with only 8 bit 4:2:2 sampling.

 

[Chuck Z]

Those two values are not related like that.

Bit depth describes the PRECISION at which the data you are capturing can be represented.

Stops describes the intensity RANGE over which you can capture the data.


So yes, it does have value to have 16-bits for 13+ stops...

RED describes their own technology as recording linear light data from the sensor. If that's the case, you don't need a 16-bit ADC for a 13 stop camera.

 

[Hrvoje Simic]

Colour value recorded is not the exact representation of the real one, it is the closest one possible.
16 bit ADC allows more dense, precise sampling of the values within the captured DR. Regardless of how many stops are there.
You can measure (quantize) range of 13 miles with yards, inches...etc, or go down to nanometer value.

More precise sampling also alows smoother roll-off at the extremes. Individual needs are one thing, potential of 16 bit sampling is another.

 

[Gunleik Groven]

Which is a good description as to what I see as the main differences between R1 MX and Epic.

I also have a hunch that the overhead (too much) precission, gives a better and less losy transorm of the meta-values when matrixes and gammas are applied.
The two cameras at same compression and same exposure with the same optics behave very differently once outside the R1MX's more "center of it's DR" priority (which so far has been reflected in the choice of sticking to RedGamma on camera).

The Epic image just is so much "fatter" for corrections.

Cheers!

G

Colour value recorded is not the exact representation of the real one, it is the closest one possible.
16 bit ADC allows more dense, precise sampling of the values within the captured DR. Regardless of how many stops are there. You can measure (quantize) range of 13 miles with yards, inches...etc, or go down to nanometer value.

More precise sampling also alows smoother roll-off at the extremes. Individual needs are one thing, potential of 16 bit sampling is another.