Marc Wielage

I've used the Baselight Blackboard before, and it's a great panel, but it only works with Baselight.

Film scans are tough because the standards are loose in the film scanning business. In Resolve, sometimes I start with a CST node decoding from Cineon to Gamma 2.4 can help. If it's too much, I start with a Printer Light (Offset) node to adjust the overall density, then an S-Curve to tame the exposure and add contrast, After that, I have nodes for Balance, Gain ("Gain1"), and a Gain trim ("Gain2"), and usually by that point I can make a decent picture assuming Rec709 delivery. If I see a bias towards pinkish-reds or excessive yellow, I'll take care of that with secondaries right after that. Note that camera negative (OCN), internegative (IN, which is a copy used to make prints), interpositive (IP), and prints all have different looks and require different tactics for final color.

I worked on more than 46 film features in 2021, and each one took anywhere from 30 hours to 70 hours, about 50 hours each on average. The ones from camera negative are the most difficult, since this is effectively the "raw" image shot by the DP and developed by the lab. It takes time and effort to tame the image and give it a reasonable, dramatic look. In many cases, we have an older SD or HD home video release as a reference, and I'll make a judgement call on whether to match it exactly or just get reasonably close to it. Of course, if the old home video releases looked awful, I'll toss that and just go by my best instincts. My feeling is that digital is EASIER to work on, because you have a bit more range... but the client also depends heavily on you to establish a look, which wasn't necessarily the case with film. I always concentrate on telling the story as best I can, but at the same time respecting and preserving the work of the DP and not "stepping on" it too much. When in doubt, we go for less processing... but I will do a little bit of relighting when there's an obvious on-set problem (glare in the background, actor who misses their mark, an important plot detail that didn't get lit). 

I'll have more to say about this in an upcoming Master Class on Feature Films, which is now being worked on. 

On 2/3/2021 at 1:14 PM, Frank Wylie said:

I will probably catch a lot of flack on this, but printer lights originated in an era when 99.5% of all color correction happened in-front of the camera; prior to filming.  What you can alter and skew is very, very limited to now standard color grading tools. If you have really uneven or mismatched looks on a job, you would probably be advised to make a balancing grade pass BEFORE you try to use printer lights and keep your printer lights node toward the very end of the node tree.

I see your point, but I don't agree. Think of Printer Lights as a way of adjusting LGG Shadows and Gain simultaneously. It has its uses, and you don't have to necessarily do it at the end of the chain. I agree it's more useful with Log images, but I honestly use it on a lot of stuff. In particular, when I have an operator who changes the lens exposure in the middle of the shot (which is pretty much a logarithmic change), I find a keyframed Offset usually will fix it, or at least minimize it so it's not too noticeable. So there's a lot of "it depends" to this.

I'm very much a "if it works, it works" kinda guy: there's a lot of different ways to approach things, and often there are no absolutes beyond making the client happy and not winding up with a picture that looks stressed-out or distorted.

Very scary. Many, many years ago (about 1980), I threaded up a 16mm neg on a Rank-Cintel MkIIIB telecine for a music video at Modern Videofilm here in Hollywood. I went back to the control room, sat with the director, and color-corrected the image as we laid it down to tape (which was the standard-def workflow back then). After the first reel was safely on tape, the director said, "hey, could we do that again with a different look?" I said, sure, and went to rewind the film at the machine. Much to my shock, I saw shards of emulsion and plastic on the base of the scanner.

In my haste and nervousness, I had wound the film around the 35mm guides... which were supposed to be bypassed for 16mm, due to its narrower width. The 35mm guides gouged into the film and put a half dozen deep scratches in the frame, all the way through! Just as I was staring at the film and wondering how I was going to explain it, the director walked in, saw what was happening and said, "oh, I guess there's some equipment problems?" I gulped and nodded and said, "totally my fault," and profusely apologized. He shrugged and said, "eh, what we've already recorded looks fine. Let's just move on to the next roll. But don't scratch the next one." The director was totally unphased, was happy with what we did, we continued with the session, and it all ended up well.

Needless to say, I was much, much, much more careful about loading 16mm on scanners after that.

Another recommendation for PixelTools: they did a terrific job at assembling together utilities, often-used nodes, and looks in one package. And unlike a LUT, they can be adjusted to work in any color space and camera format.

On 1/13/2019 at 7:46 AM, Luca Di Gioacchino said:

Can the RBG Mixer be considered an alternative to using PPL?  It, too, offers precision.

No, in Resolve the RGB Mixer has a completely different function. I tend to use Offset very early in the grade (usually followed with a Custom Curve), to get a broad overall adjustment, and once I get the image in what I call "quasi-Rec709 space," then I can start making more precise balances. I tend to use Pots (individual RGB controls) in the Primaries to start the adjustment, but you can make a good argument for other methods. I sometimes work with film-based projects where there is no Raw data per se (that is, no Raw adjustments), so sometimes I'll use either the RGB Pots or sometimes the RGB Mixer to fix color temperature problems. In particular, it's helpful if you have an underexposed Blue channel, and you can "steal" some information from R&G to give the Blue a cleaner signal to work with. This will help minimize noise in cases where the Blue channel is underexposed. 

There are a lot of different ways to work, and the beauty of Resolve (or Baselight or Mistika or any top-flight system) is you can choose one of a half-dozen different methods. As long as it works, the end justifies the means. I do tend to start with a very well-balanced picture first and then degrade it later on if we need to go to (say) an extreme blue look or a bleach-bypass look. I'm not a fan of starting with an image that leans off to one side early on in the signal chain, because the danger is that later on, you can wind up with distortion and noise because you're overdriving the signal (or worse, destructively crushing the signal and then being unable to normalize it in subsequent nodes or layers.