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 :)

Grading in REDLOGFILM vs Gamma3

F

Francisco Bech

Member
Joined
Dec 14, 2011
Messages
16
Reaction score
0
Points
0
Hello, I am new to RED and I am doing my first grading. The footage was shot on a Red one MX at 4K. Gamma3 and 2 look nice to me to start grading, but I would like to know what is exactly RedLog Film, what are the benefits pros/cons of grading with this. I have been playing with the setting and the look result can be similar in the end to gamma 2 or 3.


Cheers
 
Redlogfilm is a different gamma setting that produces a flatter, low contrast image that allows you to see better into the blacks and gives you more latitude when grading. I use it exclusively as it lets you set a nice curve from the mids to the blacks. I find the red gamma 3 is excellent for dailies and in camera looks, but when grading I start withered color 3 REDlog film.

david
 
Thanks so much for the quick answers. I understand now much better and log is the way to go. I am editing a 1 minute teaser for a documentary shot with a Red One. The only equipment I have at the moment is an Imac 20 with premiere CS6, I calibrated the screen with Spyder pro 3, a tool for calibrating screens to work with photographs. The teaser is going to DVD and Mp4 for the web.

I just want to get the best possible results with the Imac
 
RLF is indeed the starting point used to produce RG2 and RG3. Basically they are a "grade" on top of RLF. So if you use RLF as your starting point, it's like starting with a full range image that has not had any contrast added to it at all. This can give the very best end grade results.

Graeme
 
RG3 generally doesn't loose data, but it is adding contrast, and sometimes it can be difficult to "undo" that added contrast, whereas starting with RLF there's no added contrast to undo, and you're at a much more neutral starting point. Theoretically the right LUT would undo RG3 back to RLF with very little loss of data, but practically speaking I'd not do that....

Graeme
 
Hey G,

Are there any issues grading straight from RLF versus first applying a LUT to the RLF footage? Many of the LUT's I've tried lose a lot of tonality that I'd like to maintain. The only thing I'm worried about with RLF->Grade->Deliverable is creating out of gamut colors. Anything to worry about there?

Thanks!
CB
 
For grading RLF, it's best to design a nice curve that maps the log to a nominal video gamma space, and then insert your grade before that curve. This ensures good consistent tonality and access at all times to the full dynamic range of the log in a neutral manner. Even some top graders will also add some grade after the curve (effectively grading in video space) but I'm not personally so convinced of that approach.

The best way to deal with colours going out of gamut is with well crafted secondaries to pull them in the way you want them to appear, which also means grading on a colour calibrated display of the target gamut. Automated methods all have their own drawbacks, not least that they're not under your control.

Graeme
 
Is there an average amount of desaturation to achieve that film look? in gama 3 and color 3 for example. is desaturating a key element in achieving a film look?
 
Film look is generally quite saturated. Some specific looks like bleach bypass do pull a lot of the colour out, but they do that by putting more luma in ( rather than just a plain desaturation).

Graeme
 
Graeme Nattress said:
For grading RLF, it's best to design a nice curve that maps the log to a nominal video gamma space, and then insert your grade before that curve.

Graeme
Click to expand...

What I am not clear on is the meaning of 'nice' and 'nominal' here. Is the process of designing a curve separate from the creative grade? How does one determine what is nominal, and isn't 'nice' up to the artist?

Why can't we use RG3 as a viewing LUT and insert the grade before it?
 
RG3 was designed to be graded after, not before. You'll get much better results rolling your own curve and grading before it.

Nice - I was thinking more "well behaved". Some people think a curve can do all kinds of funky stuff, but what you want for this is a curve that maps black and white to the full scale of your video, puts mid grey at ~44% and introduces the basic amount of contrast that you want for the project. Most colourists I know have their own curve that they have designed and they know works well, either in general or per project. It's a bit of an art and a science to do. By nominal, I was meaning that you don't attempt to put the look into the curve, you just perform the most basic of mapping black and white points and introducing a basic amount of contrast.

Graeme
 
Okay, so theoretically you could be applying this curve in your grade every time you start from scratch, but since all such 'starting from LOG' grades must at least transform to the parameters of a given color space, you put in that maneuver automatically and operate upon it, kind of like a sculptor starting from a humanoid form rather than a lump of clay?

Should a colorist be able to determine by eye if the curve they've applied to the log footage is preserving the tonality of the image within the target color space (presuming their monitoring system is accurate) or is this something that requires some form of measurement?
 
What I suggest for high end jobs, is always to use RAW data at it's purest form, smallest number of compression allowed which for me it's 5:1 ( just no need at all to go to the super storage hungry 3:1) depending on Resolution (which for me it's always the absolute maximum, 5kFF now 6kFF soon) set the recording to RLF, if needed for a client's monitor give put a RC3-RG3 at it present the best starting pint so not to get them all worked out about seen an other wise Flat image off their monitor.

Then in your post, for super fast and simple images out do RC3-RG3 as I most of the time do for pulling frames and then add form there, so I have a super fast grade, if I need much more accuracy in the grading then I work from the originally captured RLF.
 
KETCH ROSSi said:
What I suggest for high end jobs, is always to use RAW data at it's purest form, smallest number of compression allowed which for me it's 5:1 ( just no need at all to go to the super storage hungry 3:1) depending on Resolution (which for me it's always the absolute maximum, 5kFF now 6kFF soon) set the recording to RLF, if needed for a client's monitor give put a RC3-RG3 at it present the best starting pint so not to get them all worked out about seen an other wise Flat image off their monitor.

Then in your post, for super fast and simple images out do RC3-RG3 as I most of the time do for pulling frames and then add form there, so I have a super fast grade, if I need much more accuracy in the grading then I work from the originally captured RLF.
Click to expand...

Thanks Ketch. When you work from the original RLF, do you perform the curve transform to video gamma differently every time (on a shot or show basis) or do you have a set curve saved for that purpose?
 
Brett Harrison said:
Thanks Ketch. When you work from the original RLF, do you perform the curve transform to video gamma differently every time (on a shot or show basis) or do you have a set curve saved for that purpose?
Click to expand...


Every time is a different shot, never the same!

Okay, my take is a bit different and some what originated form the days of photography, so it might be unconventional, but I ma sure that Graeme can agree that the concept is the correct one just the same.


What I assume with the RC3 - RG3 is as if this were a pre-processed RAW, so no longer PURE RAW, and even so not entirely baked in, but some what indeed baked in, as when you try to eliminate entirely it's look, then you are no longer working on the PURE R3D RAW files as you would if recording on REDLOGFILM to begin with.


I never set any curves as per shot, or for an entire shoot, as I believe in individually setting a look after the image is acquired in Redlogfilm at it's purest state, when it comes to grading/coloring even so I know I'll hire an experience Colorist, He/She will have to closely follow my instructions as I know exactly what I want, and always capture and lit my shots accordingly to the look I'll want to go to at the processed footage.


Even so for needs and speed etc. applying a look or a curve might be good, is never so for me, and fortunately I don't ever have the need to argue with clients what the image should look like or not while I am recording and they might be viewing a monitor, either they trust me fully or they can find themselves some one else to do the job, but again, a look applied to the external monitor in RC3 -RG3 always puts them to rest.
 
Graeme, just a question..

Does a Red camera perform the R3D debayering on camera?, or the file remains a Bayer pattern image until it reaches Redcine-X algorithms?.

Of course I understand that it has to be some kind of debayering on camera, otherwise you wouldn't see anything on the little monitor, but I mean specifically, the pure raw R3d ??..
 
Brett Harrison said:
So the LOG format was designed to emphasize these parts of the film curve and allow the film scan to 'decompress' these curved parts of the image where gradations change rapidly, where lots of stops are burried. This is so each stop in the film scan file container provides equal numbers of data to each stop.
Click to expand...
The peculiar thing is: different companies interpreted what was a "standard" log film scan. I can recall projects 10 years ago where we received scans from eFilm, Cinesite, Technicolor, ILM, and Digital Domain... and they all looked different. Even worse, a week later they'd do another scan of a new scene and it would look different again. And different scanners yield different results: do a log scan on a Spirit 4K, a Northlight scanner, an Imagica scanner, and an Arriscanner, and they're all different.

The reality is that "a log scan" is kind of anything you want it to be. There's a wide range of what can be acceptable in terms of sufficient range for detail and color once it gets into the color correction room. I'm perfectly happy with RedLogFilm, and that generally gets it in the ballpark for me, provided the image has been well-lit and well-exposed. But I am in favor of doing a test prior to the session if possible, just to provide an early warning in the event there's anything wonky with the footage.

Jeremy Selan, one of the color scientists with Sony Pictures Imageworks, has an excellent technical paper that cites the fact that log values are not standardized within the film industry or in the digital camera industry. He does say that different facilities have their own in-house standards that does stay consistent, and their viewing LUTs are pretty consistent between different companies.
 
Pablo Moreno said:
Graeme, just a question..

Does a Red camera perform the R3D debayering on camera?, or the file remains a Bayer pattern image until it reaches Redcine-X algorithms?.

Of course I understand that it has to be some kind of debayering on camera, otherwise you wouldn't see anything on the little monitor, but I mean specifically, the pure raw R3d ??..
Click to expand...

There is no debayering done to the R3D file. Inside the R3D is the raw Bayer pattern sensor data. Debayering only happens when you view the R3D.
 
You must log in or register to reply here.
Back
Top