[Mike Most]

If you’re going to use an Arri website generated LUT with a film original, you want the matrix OFF. The matrices that Arri offers for those LUTS is based on the Alexa camera’s particular characteristics and is designed to restore proper saturation and specific color values when using that camera, and with the assumption that the in camera matrix was off while shooting. What you want is a LUT that does not incorporate a matrix because a film scan already has proper values (it’s not the result of a Bayer pattern capture system, at least not at the camera end).

Second, LogC is NOT identical to Cineon log. It is slightly different in overall gamma and in specific areas of the curve. It is very similar to Cineon log, and can usually be used interchangeably with it if color grading is being done, but it is NOT the same (as opposed to, say, RedlogFilm, which IS identical to Cineon nog…).

[Mike Most]

>I know, it’s weird to ask for LESS updates, but sometimes too much is just that- too much. Personally, I >find this crazy pace of updates is a bit of a burden

I don’t really disagree, except for one thing: camera support. The rate at which new cameras seem to be appearing, and the fact that many if not most of them seem to have their own unique formats (even more so when you’re talking about cameras that support Raw output) is sometimes even faster than the Resolve update cycle. Support for new cameras is very useful for some of us that deal with a wide range of shows that seem to be very unafraid to employ cameras and formats that have been on the market for about 20 minutes. So while the “general stability” stuff is welcome but not quite necessary much of the time, the new camera support comes in very handy.

[Mike Most]

It’s an interesting approach, but ultimately not really accurate because the Parallel node simply does a mix of its inputs rather than populate the red, green, and blue channels specifically and individually. A few years ago I wrote a Shake script called Technilook (you can probably still find it somewhere on the Net, I think I originally posted it on Highend2D) that used some specific math to create “pure” red, green, and blue records and combine them in a channel specific mixing node. As with this example, the red saturation tends to get a bit out of control using this approach, so I put in a user control to adjust that. Most of what I did was based on some things that Rob Legato and Josh Pines did in prepping the movie “The Aviator” for Marty Scorsese, but I never really had access to their code (although I recall asking Josh about it a few years later….).

Just looked – yeah, it’s still there on creativecrash.com…

[Mike Most]

I give up. Clearly my experience means nothing since I didn’t pay for every machine out of my personal bank account. I guess the fact that I set up and am responsible for maintaining every one of them – not to mention that the configuration is approved by a major vendor – is just not valid enough. I bow to your superior experience, intellect, and arrogance.

[Mike Most]

And my statement is based on almost a year of running a GT120 and a GTX580 together in at least 5 Mac Pros running Colorfront software, never turning them off for months at a time, and never having one fail. And, BTW, having that configuration designed, blessed, and supported by Colorfront and its reseller for both us and numerous other customers.

Regardless of what you sometimes seem to think, Jake, I don’t make this stuff up. I have at least as much experience as you do and when I say something here, I’m not guessing. That doesn’t mean I’m always right, but I really don’t take well to your insisting on turning every statement I make into some sort of contest.

[Mike Most]

No, an external power supply is not necessary. You do, however, need an adapter cable.

[Mike Most]

I have to say that I used the Dreamcolor calibration puck and software quite a few times and never had the issues that you’re describing. It always brought the monitor back to where it was supposed to be, and projects graded using that monitor always remained consistent across other deliverables and viewing systems. Of course, I was feeding the Dreamcolor via SDI through an AJA HDP2, ensuring that the calibration engine was being properly implemented.

I’m not saying you’re wrong, but I am saying that what you’re stating is not necessarily universally true, either.

[Mike Most]

Canon Log is not LogC. LogC is a particular log curve similar to but not exactly the same as Cineon Log, that is used by Arri in the Alexa. Just because something is log coded doesn’t make it any kind of a standard. Cineon Log is a standard because it was developed by Kodak and is based on print density. LogC is a variant that is very similar but slightly adjusted due to the specific response of the sensor that Arri is using. RedLogFilm is an application of the Cineon Log curve that is used by Red. Unlike any of those things, Canon Log is more like Panalog was, a “mild” log curve that is used for the sole purpose of retaining more information than you would get if you coded to Rec709 type video space. It is not designed to work with controls that are scaled for Cineon type log curves and Cineon type white and black points.

Now, having said that, log coding is nothing magical and the container that the image exists in has nothing to do with the values in the image. So the answer to your question – at least I think the answer to your question – is that the controls in Resolve are designed to be used with “standard,” Cineon type log curves and their use with other gamma curves is not impossible, but it is not what they were designed for. The shadow and highlight controls might be of some use because of their restricted range when compared to lift and gain, but in the cases you’re talking about, you’re probably better off using “traditional” controls. In fact, one of the design criteria for both Panalog and Sony’s Slog was the ability to be corrected directly in a video environment with “normal” video scaled controls, but easily converted via a 1D LUT to “standard” Cineon log for a DI environment.

[Mike Most]

The log controls have nothing to do with linearizing the output. That is a step handled, as Alexis says, by either a LUT or a curve. The log controls are designed to manipulate information that is coded in a log format, with the expectation that you’re viewing the result through a normalizing LUT. The values in a log coded image have different black and white points and a different gamma than video images, and controls need to be scaled differently to allow for manipulation of those images in a more controllable and predictable way. The most common way of dealing with this, based on the film timing approach, is to use a combination of exposure (implemented here as offset) and contrast to alter the color balance and overall level on the image as a whole and adjust contrast to better match any under or overexposed material to the video display environment. The shadow, mid tone, and highlight controls act on very restricted ranges that are based on the Cineon log black and white points of 95 and 685 rather than 0 and 1023. The result is that you can set up basic balance on an image using purely offset and contrast, then fine tune it using shadows, mid tone, and highlight adjustments. Because of the design of those controls, you’ll find that the highlight control, while restricted to input values near or above 685, acts as a sort of variable soft clip rather than a gain control (the gain control has a pivot at 0) so it’s sometimes useful even in a non-log grading environment. It is also important to note that the contrast pivot can be very useful in dealing with exposure issues, such as lack of fill or blown out highlights. By raising the pivot, the contrast control affects the darker areas, and begins to feel more like a “black stretch” control. By lowering the pivot, it affects the upper ranges, bringing out more detail in things like overexposed facial highlights and the like. On Resolve, you get a lot of corrective power out of these things because you can implement them separately on different nodes (provided all of those nodes are prior to the output LUT).

Film DI work has been done using exposure/contrast and log scaled range controls for a long time, in fact, Lustre was designed around it, as was the Film Grade strip in Baselight. Because you’re not pulling the picture apart by separating the brightness ranges, the offset/contrast control lets you stay much more true to the original image because once you get the balance right (usually by setting up flesh tones), all of the colors in the scene become what they’re supposed to be based on the original photography. It might be counter-intuitive to those who have only used lift/gain/gamma control systems, but it’s much simpler to come up with something representative of the photography using only one trackball (the offset trackball) than it is using three. You get better color separation and thus a lot more depth. I would suggest experimenting with the log grading controls using this description rather than trying to use them as you would lift, gain, and gamma. Think of the offset control as your Hazeltine knobs, and the Contrast as a display gamma correction and it will make more sense. I think you’ll be surprised at the clean results you can achieve.

[Mike Most]

Higher bit depth is useful in post production, but not necessarily for color grading. It is helpful in the case of matte extractions from green or blue screen material because it allows for more color information and thus more precise edge delineation. But for color grading, the difference between ProresHQ and Prores4444 is minimal at best, and non-existent most of the time.

What might be confusing the issue here is the color space used. If you record using a log coded format, like LogC on an Alexa, you retain much more information than you would recording the Rec709 output, and in a more usable form. But that has nothing to do with the codec, it has to do with how the image is being processed. You can encode a LogC image in Prores4444, and you can encode it in ProresHQ. For all practical purposes, the grading will be exactly the same. You could even move a color correction from one to the other and get the same result.