Welcome to our community

Be a part of something great, join today!

Log in
  • Hey all, just changed over the backend after 15 years I figured time to give it a bit of an update, its probably gonna be a bit weird for most of you and i am sure there is a few bugs to work out but it should kinda work the same as before... hopefully :)

RED Color - A look at REDWideGamutRGB/Log3G10

Phil, what's your preference rank?
And I want no part of B, judging from my monitor.
 
The RWG/Log3G10 aspects are locked. They don't impart any particular "look" in of themselves though. The shape of Log3G10 does define how wasy it is to grade an image, and some extensive work was done there to ensure it worked as well with CDL as other log curves with similarly large dynamic ranges.

Graeme
 
Is Log3G10 in the same vein as Cineon (like how REDlogfilm was trying to be mathematically the same) or is it more like RED's/Graeme's take on maximizing DR and contrast to make a less accurate albeit friendlier starting point (ala Arri's LogC, Sony's Slog, Canon's Clog, etc.)?

In other words, RLF holds all the DR already, so what's the difference(s) with Log3G10? Better contrast in the high end (resulting in the better roll-off we're seeing)?
 
REDLogFilm uses the exact Cineon math, which encodes linear float values up to 13.5 (if my memory serves me right) or ~6.2 stops above mid grey (which is 0.18 in linear float). That used to be a decent enough dynamic range, but it's no longer appropriate for today's cameras. Log3G10 encodes 10 stops above mid grey (hence the 10 in the name - the 3 refering to mid grey being encoded as 1/3) and thus is a much wider dynamic range than Cineon ever was, encoding float values up to 184.32. Given a typical mapping to HDR, that would also mean it could hold a peak white of 18,432 nits which is way beyond what the SMPTE 2084 spec needs, making it also very suitable for HDR use, whereas Cineon would peak at 1350nits, respectable, but way below what mastering standards need.

Log3G10 is more akin to ACES Proxy / cc in that it's designed to encode float values over a good dynamic range, but it's designed to hold up to grading operations (such as adding saturation) with greater fidelity, and also to nicely handle the typical grading operations as expressed as CDL.

By the nature of the Log3G10 encoding storing vastly more highlight DR leads to much better highlight rolloff, and by extension easier ways to define the precise nature of that rolloff. I'm certainly seeing vast benefits in how the highlights roll off here.

Graeme
 
I see... So what happens when you a lower DR image into the same extended gamma space? Would manipulation/control be better as well (because you could "stretch it" more accurately across the greater DR)?

...I know that's not how it works, but I'm hoping for a noticeable (even if mild) benefit for MX...

OH, and when can we expect this in a RCXp release build?
 
You get benefits across our range of cameras. I've tested with footage all the way back to Red One.

Graeme
 
Thanks Graeme for your work on this new development in the color workflow. I have a shoot later in October, shooting Red. I'll be coloring as well, so I hope to try this approach. Although I must admit, I'll have to read this thread (and other info) a few more times before I feel confident in how to implement it. :)
 
Graeme Nattress said:
The RWG/Log3G10 aspects are locked. They don't impart any particular "look" in of themselves though. The shape of Log3G10 does define how wasy it is to grade an image, and some extensive work was done there to ensure it worked as well with CDL as other log curves with similarly large dynamic ranges.

Graeme
Click to expand...

Excellent. Thanks. :)
I'm seeing a couple of improvements in my signal analysis. Hats off.
 
I just had some clips shot with Scarlet Weapon that had blown highlights, and when going to redlogfilm, ISO settings over 1000 would clip the highlights more and more, but with 3G10 there was no extra clipping with any ISO value.
 
Aaron Lochert said:
I prefer F in every single one.
Click to expand...
If I could choose only one I would pic F.
I dont mind A but it feels like dragon color, red gamma 4 and that we already got. Would be interesting to try F on my own footage. Well we se later what it will land on. Love those updates!
 
Graeme Nattress said:
Log3G10 encodes 10 stops above mid grey (hence the 10 in the name - the 3 refering to mid grey being encoded as 1/3) and thus is a much wider dynamic range than Cineon ever was, encoding float values up to 184.32. Given a typical mapping to HDR, that would also mean it could hold a peak white of 18,432 nits which is way beyond what the SMPTE 2084 spec needs, making it also very suitable for HDR use, whereas Cineon would peak at 1350nits, respectable, but way below what mastering standards need.
*edit*
By the nature of the Log3G10 encoding storing vastly more highlight DR leads to much better highlight rolloff, and by extension easier ways to define the precise nature of that rolloff. I'm certainly seeing vast benefits in how the highlights roll off here.
Click to expand...

So I'm presuming Log3G12 gives 12 stops above mid-grey (without remapping for it; that's what the transforms are for)?
 
Mike P. said:
So I'm presuming Log3G12 gives 12 stops above mid-grey (without remapping for it; that's what the transforms are for)?
Click to expand...

Correct. Log3G12 is for 12 stops above mid-grey. Log3G12 for 10 stops above mid-grey.

This whole method is designed to fully contain the total possible image information, dynamic range, and color possible by current RED Digital Cinema Cameras.


Side note. For those who looked at the comparison images. Yes, option A is indeed straight DRAGONcolor2/REDgamma4.

I'll reveal my preference and my thoughts on things shortly, but most of these options are easily attainable within the final release of what Graeme has been cooking up if not directly so.

Especially for C-H. B is more or less an older version which has since been updated for the better.

If you've been tracking this thread and knowing my general preference for an "out of the camera look" those are all pretty much in bounds towards what I'm after. I

G is indeed a common trend and I think many will go that route.
 
Phil Holland said:
Correct. Log3G12 is for 12 stops above mid-grey. Log3G12 for 10 stops above mid-grey.

This whole method is designed to fully contain the total possible image information, dynamic range, and color possible by current RED Digital Cinema Cameras.
Click to expand...

Why would one choose 3G10 over 3G12? Wouldn't you want that extra two stops of highlight latitude?
 
Marcus Friedlander said:
Why would one choose 3G10 over 3G12? Wouldn't you want that extra two stops of highlight latitude?
Click to expand...

Whoa whoa, you're not getting two extra stops of highlights, you're just setting mid-grey lower with respect to your raw processing (I think). And I'd guess that after a certain point it probably gets redundant as you'd just be normalizing it in order to get it to play correctly down the line anyway (granted I could be completely wrong, as I only spent 5-10mins turning knobs and pushing buttons.)

Phil Holland said:
Correct. Log3G12 is for 12 stops above mid-grey. Log3G12 for 10 stops above mid-grey.

This whole method is designed to fully contain the total possible image information, dynamic range, and color possible by current RED Digital Cinema Cameras.

Side note. For those who looked at the comparison images. Yes, option A is indeed straight DRAGONcolor2/REDgamma4.

I'll reveal my preference and my thoughts on things shortly, but most of these options are easily attainable within the final release of what Graeme has been cooking up if not directly so.

Especially for C-H. B is more or less an older version which has since been updated for the better.

If you've been tracking this thread and knowing my general preference for an "out of the camera look" those are all pretty much in bounds towards what I'm after. I

G is indeed a common trend and I think many will go that route.
Click to expand...

So does that mean there's going to be a 3200k transform and a 5600k transform (assuming that there's something Graeme could tweak to get a better image possible under those two major light conditions?)
 
Last edited:
Actually as Graeme explained, the 3 in 3g12 and 3G10 means that middle grey is set at a value of 1/3 in both. I guess that 3G10 was the winner because it allowed for slightly more precision in those stops above grey, and perhaps because it's likely that 99.9% of shots won't be at an ISO where 12 stops above grey would be possible.
 
Anton Shavlik said:
Actually as Graeme explained, the 3 in 3g12 and 3G10 means that middle grey is set at a value of 1/3 in both. I guess that 3G10 was the winner because it allowed for slightly more precision in those stops above grey, and perhaps because it's likely that 99.9% of shots won't be at an ISO where 12 stops above grey would be possible.
Click to expand...

Thanks for chiming in Anton. Can you clarify what you mean by 1/3? Thanks!
 
Marcus Friedlander said:
Why would one choose 3G10 over 3G12? Wouldn't you want that extra two stops of highlight latitude?
Click to expand...

Mike P. said:
Whoa whoa, you're not getting two extra stops of highlights, you're just setting mid-grey lower with respect to your raw processing (I think). And I'd guess that after a certain point it probably gets redundant as you'd just be normalizing it in order to get it to play right down the line anyway (granted I could be completely wrong, as I only spent 5-10mins turning knobs and pushing buttons.)
Click to expand...

Anton Shavlik said:
Actually as Graeme explained, the 3 in 3g12 and 3G10 means that middle grey is set at a value of 1/3 in both. I guess that 3G10 was the winner because it allowed for slightly more precision in those stops above grey, and perhaps because it's likely that 99.9% of shots won't be at an ISO where 12 stops above grey would be possible.
Click to expand...

Sort of correct. This whole REDWideGamutRGB and Log3G10 approach was developed to "contain" everything a RED camera can produce image quality wise. Yes, with aggressive processing you can hurt yourself in any log type format, but overall for the majority of typical shooting situations and even the more extreme situations at higher ISOs Log3G10 has you covered.

Unless your camera is seeing much, much, much more DR than it currently is. I'm talking over 20 stops, there wouldn't be much advantage that I can think of. Not in these modern times of Rec.709 and Rec.2020. Until like Rec.3030 or whatever singes your eyebrows with light and radiation, not much reason for anything bigger.
 
You must log in or register to reply here.
Back
Top