[Blaise Douros]

Don’t discount audio as a method to separate them. I’ve been on shoots where we miked the main interview with a different mike to make him more present and authoritative-sounding, while secondary interviews got a more normal mike.

Color and type of lighting can be a differentiator, as Mark has said. I might have the main guy lit with very dramatic lighting that emphasizes his face over a dark environment, while the secondary folks get a normal environmental interview.

Punching in on a 4K shot on a 1080 timeline can be useful, but I like the visual change of a truly different angle. For second camera, I like to do a tighter, longer-lens shot that is further off-axis opposite the keylight, so it hints at a little more rimmy/backlighty look. Maybe put that second one on a slider. I regularly use a Sony FS7 as my main camera with an RX100 mkii as camera 2, and that works great–they both shoot Slog flavors, so they play well together.

I’m not sold on Mark’s production monitor idea unless film or tv is the subject of the interview, so the monitor helps sell the story. But different strokes.

[Blaise Douros]

Hey Duke, saw your shoutout ☺ There are more qualified colorists than me who can answer some of these questions, and there are a TON of variables out there in any color pipeline. With that said, here’s a couple of facts you may find interesting:

There is a known and documented gamma difference in Quicktime versus other media players. Quicktime lifts the gamma slightly, so that the blacks are a little less black. VLC does not do this. So VLC will be more accurate file playback to show you what colors are really in the file you just crunched out.

As far as WMV…yeah, don’t use it. It’s just not standard anymore. Find some MP4 settings that look good to you, and use them–and view the results in VLC. It’ll be closer to what you’d get if you played that file back onto a decently set up HDTV.

While there aren’t any ultra cheapo color checkers out there that I’m aware of, I’d take some time to go through your OS’s color calibration process. It will at least get you to a point where you know roughly what your monitor is doing. Your system defaults could be doing god-knows-what.

I use an X-rite i1 Display Pro color calibrator, which runs about $250, to calibrate our iMacs and attached HP monitors–I’m not delivering big features or anything, so it’s good enough to make sure my two machines are in sync and somewhat standardized so that my assistant editor won’t give me the side-eye when she opens my projects. I purposefully use HP monitors as my second screens so I can A/B instantly with the iMac monitors, so that I can see what it will look like on a couple of different display types with slightly different gamma. It’s not a fancy color pipeline at ALL. But it is good enough so I can consistently deliver stuff that looks good, since I rarely if ever expect to go to an outside colorist.

[Blaise Douros]

Aha. I think I see the problem here, or at least, one problem. You may be relying too heavily on the histogram to expose your image. Histograms aren’t that useful in video, because they don’t tell you what part of the image is correlated to your exposure graph on the histogram. It just tells you how much of the image is falling within a certain exposure level. That’s why scopes are useful–it tells you what part of the image is where.

A properly exposed histogram is not always a mountain in the middle of the scene. If you were shooting a low-key scene, then the majority of the histogram would sit at the left in the blacks, with a spike further up where your bright spot is.

So when you say that the exposure on your first examples were perfect because you set it up perfectly in the histogram, that’s where you’re running into the problem. A properly exposed image is exposed correctly for the scene–not a one-size-fits-all get-the-peak-in-the-middle approach on the histogram. Because I can tell you, just like you can tell, from looking at the actual image result, that your “perfectly exposed” example was not perfectly exposed; it’s several stops under. Your “overexposed” example was actually much closer to a good exposure for the scene.

I don’t know the GH4 at all, and have never shot with it, so maybe there’s some weird exposure compensation setting that got bumped that is screwing you over. But reading your post, I get the sense that you’re judging your exposure too much by how you believe the histogram should look, and not enough by the actual image.

As I understand it, the GH4 is capable of displaying zebras; if you don’t use them already, you might want to look at doing so, because they will be a far more useful tool for your overall exposure than the histogram will. But again, you will need to adjust your use of them to reflect your flatter profile.

Personally, I’m at the point where if I’m not shooting in a standardized logarithmic gamma like Slog3, I’d just shoot a standard Rec709 profile, especially when you’re working with lower-bitrate footage like the 8-bit 4:2:0 that the GH4 records internally. The footage just doesn’t hold up to any real adjustment in post, and especially in the highlights, you get weird saturation artifacts as the software tries to interpolate color information that just doesn’t exist. There are just too many variables involved in exposing for a nonstandard, undocumented flat profile like the one you create when you just desaturate the in-camera capture settings. That way, I can use known exposure values in both contexts–I know how to expose Rec709 correctly because that’s what we’ve all learned to expose from the beginning of video, and I have lots of documentation that helps me expose a standardized log gamma format like Slog3.

Now on to your examples below. On the first two examples, I really am not seeing white balance issues. The color is perfectly appropriate for the scene. And the exposure looks good, too. Your third one, where you’re holding the grey card, looks about a stop underexposed.

I initially judged that by eyeball. When I drop #3 into PS and isolate the gray card, it confirms my suspicion; it is sitting around 30-35% on the histogram, which is low. You can even see this on the full image histogram–there’s a spike (since there are a bunch of pixels around the same exposure value) around that 30-35% mark that stands out along the rest of the image.

Now, not all gray cards and not all cameras are created equal, so this is a very generalized way to judge, and is also very contextually dependent on the type of light and the overall scene. But in a very very general sense, you want that gray card to sit around 43-50% (or 43-50 IRE, on a scope) when under your keylight, shooting in a standard Rec709 gamma space (i.e. NOT flat). You might want it to be a little lower if you’re in strong direct sunlight. But at that point, you want to confirm that your histogram is actually showing a reading from ONLY the gray card–position it in your scene, then zoom in so that it fills the screen–then look at your scopes/histogram. That gets you most of the way there, but you then need zoom back out to make minor adjustments depending on what’s in the frame–if you don’t have any deep blacks in the frame, you shouldn’t be darkening the scene down so that your histogram shows a bunch of stuff in the blacks. Similarly, if you don’t have anything in the scene that should be clipping, don’t push the exposure up just so that you have an even histogram.

[Blaise Douros]

Looking at that image, though, I’m not even convinced it’s fluorescents. Those bands are so narrow that it looks more like LED PWM cycles than fluoro. And with non-cinema LEDs being so unstandardized, there’s often no sync speed at which they don’t flicker.

Of course, if this was a fully rigged set, that shouldn’t have been an issue, period. But I’ve been in situations shooting doc-style under ambient LED lights, or LED streetlamps, where you can’t set your shutter slow enough to smooth out the artifacts, rolling shutter or not.

The bottom line, though, is that it IS an artifact created by rolling shutter interacting with pulse cycles of a light, whether LED or fluorescent.

[Blaise Douros]

No, you’re wrong. This issue is created through an interaction between rolling shutter and a light source that is out of phase with the shutter speed. On a global sensor, this would show up as frame-to-frame flickering, while on a rolling shutter camera, it shows up as these rolling bands.

Also, stop getting into pissing matches. “Professionals” often know what they’re talking about. Like Tyrion Lannister, our job is to drink and know things.

[Blaise Douros]

The only other thing I can tell you is that your screen brightness may be screwed up? If your monitor settings are weird, it might be causing you to expose incorrectly to compensate.

Duke, I’m sure you don’t have a problem using the camera. I’m telling you that exposing for flat color needs to be done completely differently from how we were all taught to expose for standard Rec709 gamma curves.

The first time I shot a project in Slog3, I royally f*cked up the exposure for the exact reasons I specified above–the only thing that saved me was that the codec had enough leeway that I was able to salvage it. But if I’d had any real grading to do, I’d have been royally hosed. I’ve dealt with many, many other shooters who have made the same mistake, too–and it all comes from the same reason: not understanding how log exposure works.

The first thing you need to know is that the histogram is not as useful when you’re shooting flat footage as it is when you’re shooting standard gamma. The graph is going to be compressed toward the center of the histogram–in post, you’ll apply curves that will increase the contrast, while allowing you to keep detail in the highlights or shadows as necessary.

So in flat footage, if you’re exposing your histogram with the blacks just kissing the left of the histogram, like we’ve all been trained to do, then you are underexposing the image. Similarly, if your white card is sitting at 95 IRE, then you’re way overexposing. That’s why it’s crucial to do the test that I described above, so that you know where you need to make adjustments in your exposure for different kinds of conditions. You CANNOT expose flat/log footage the same way you expose a standard Rec709 gamma image–not if you want to get good results. And many of the problems you’re describing with the footage not looking the way you want it to look are completely explained by this.

When I say you may not understand how to expose log footage, this is not something to take offense at, because I’m telling you from my own experience, and the experience of some really top-notch DP’s I know, that it is super common. I learned it the hard way, and then I had to teach a DP with twenty years more experience than me how to do it. It’s not because he’s incompetent–the guy is a much better DP than me, with oodles more experience and twice the talent! But he hadn’t used the tech, and I had. Now he’s back to kicking my butt.

Have you tried color correcting that image that you labeled as three stops overexposed? Incidentally, how did you determine that shot was overexposed? Because the result looks correctly exposed.

[Blaise Douros]

Look at your waveform monitor on the first example you posted, with the screenshot. You’re completely underexposed. Your skin tones are sitting at 50 IRE max, and the usual ideal range highlight range for white peoples’ skin is around 71 IRE (obviously not specular highlights).

I think I know what the problem is, too. You’re shooting in too flat a profile, and you’re not used to it, and I think you’re making an extremely common mistake.

See, when you shoot flat footage (or log footage, which I’ll get into in a minute), everything starts to look washed out and low-contrast. And what are we conditioned to do from our prior experience when something looks washed out? Drop exposure so the monitor looks good. So you drop the exposure a stop or two, and the flat or log footage starts looking a little better, it makes you feel a little less stressed about it on set.

Your eye, and your prior experience, are lying to you.

See, the whole point of flat footage is that it looks washed out so we can unsqueeze it later in post. but when you drop the exposure so it looks better in the monitor, you underexpose the whole scene. When you try to bring up the highlights, you’re actually having to bring them up too far, and you’re seeing the results here–clipping.

So here’s what you need to do: you need to shoot the same scene, twice. Once with settings zeroed out, standard color profile, so you can get your exposure correctly dialed in with NORMAL color. Then, switch to your flat settings, change NOTHING, and shoot the same shot. If you have an 18% gray card that you can hold in the image, then that’s even better–more on that below.

Now you have a baseline for comparison–you can see where your waveform should be for a correct exposure with your flat settings. You also have a chance to make notes and build a template of settings to desqueeze the exposure from the flat settings–a DIY LUT, basically.

If you’re shooting in a true log format, then part of this process becomes a lot easier–you can use a LUT to do the initial desqueeze, and then season the look to taste. There are also exposure guides for log footage that can help you on-set; guides that tell you what IRE an 18% gray card should be at for proper exposure, that kind of thing. Because you’re building a de facto log exposure through your camera settings, you don’t have an easy solution for this (unless someone has already created it), so that’s why you need to do your comparison work ahead of time.

The fact of the matter is, there’s only so much color and exposure information present, even in the fabled GH4. It doesn’t matter how high the bitrate is, there just isn’t as much information in 8-bit 4:2:0 footage as there is in 10-bit 4:2:2, for example. Which is why you need to expose GH4 footage much more carefully than you would a more robust codec–you just don’t have the grading latitude that you would in a more robust one.

[Blaise Douros]

It’s hard to make a recommendation without knowing what technique you used in the final shot. However, I would tend to say that the 3D camera tracker might have done the job here, with the caveat that you’d need to get into the advanced settings and specify your lens FOV, use the more detailed tracking settings, that sort of thing.

Shallow focus is tough for VFX work, so I would tend to say a narrower aperture would have been helpful here. You don’t need to go full deep-focus-at-f/22, but a little less blur overall would probably help. Tracking a blurry surface is damn near impossible, since there are no sharp details for the tracker to grab onto.

[Blaise Douros]

Re: the Slog3; since you’re planning to get the A7sII anyway, you can play with it on that camera.

The Shogun would be a great monitor, but won’t work well as a shoulder-mount solution–for that, you’ll need to use the eyepiece.

As far as focusing goes, I have to go extremely light, fast, and low-profile for a lot of my doc shoots, so I don’t use a follow focus knob. I just have my grubby hands RIGHT ON THE LENS. I know, it’s barbaric. But I’ve gotten used to it over the years. Especially when using a zoom lens, it keeps me one step closer to the lens so I can react fast.

The only other add-on I have is the K2M XLR add-on, which allow me to bring in four channels of XLR audio if I need to.

Again, though, I don’t shoot much narrative or even commercial type work. It’s mostly documentary style or fast-turnaround corporate. There are lots of guys who will give you a much more comprehensive kit list for the FS7.

[Blaise Douros]

I agree with you, Todd, in that I much prefer Canon glass to Sony. I was a hardcore Canon guy for many years. That’s why I’m using the Metabones Speedbooster to convert my Canon glass for use on the FS7.

I think the one place you’re also right, Todd, is that the C300 has a nicer right-out-of-the-box look. Canon’s processing algorithms pump up the reds in their imagery which tends to make people look very nice. Sony leans heavily on the green channel for luma data, and it often makes their footage have that very slight greenish cast which I associate with the older Sony cameras.

That said, the FS7 Mk1 (which is what I use) doesn’t have that problem. The ungraded, out-of-the-box look in Custom mode is actually really nice, and I use it all the time when I’m on a project without the time for detailed grading. I’ve shot it in 1080P next to the 5DIII, and it makes the 5DIII footage look like hot garbage in terms of detail.

Now, over to you, Craig.

First, lens adaptation: the Speedbooster is a cool thing–it’s an EF to E mount adapter with electronics to allow the camera body to control the EF glass. It also has a focal reducer that collimates the lens down to the Super-35 sensor on the FS7; what that means is that all the light that normally would fall on a full-frame sensor is concentrated onto the APS-C/Super-35 sensor in the FS7. The result is a full-frame equivalent FOV, with the added bonus of an additional stop of light. So F4 lenses become F2.8 lenses. Metabones also makes a non-speedboosting adapter with no optics. I have both, and the Speedbooster basically lives on the camera body, since I have no E-mount glass.

As for color, there’s a learning curve in shooting log footage, and because Slog3 gamma has such a wide gamut, it will take some practice. In the end, LUTs will be your saving grace in one way or another. The FS7 allows you to apply a LUT to the viewfinder, while still recording Slog3 to the card. This allows you to see the normal exposure while still recording the log footage, retaining the dynamic range for post. Alternately, you can learn how to just expose for Slog3 in the viewfinder using the waveform monitor.

In post, you’ll use LUTs to get yourself most of the way to a good grade. You can then tweak it from there. It took me a few projects to really understand how to expose and process log footage, especially under run-and-gun documentary-type conditions. But once you do, it gives you a lot of power in post that you never would have with a camera shooting a straight Rec709 gamma profile: you’re basically shooting HDR.

And of course, you don’t have to shoot in Slog. You can shoot in Custom mode with a Rec709 gamma, and the camera will act just like a standard video camera.

As far as accessories: I’m not a V-mount or Gold mount battery user, so I just bought several of the largest batteries available for the system. Other FS7 shooters have mentioned that the V-mount add-on makes the camera balance a bit better on your shoulder, as it adds some counterweight. So if you have some investment in big batteries, I’d say go ahead.

Forget the handle arm that comes with it–Sony’s arm assembly is junk. Buy the Shape FS7 handle.

The standard viewfinder on the FS7 MkI isn’t too hot. If you have the cash to get a replacement, go for it. Otherwise, it’s fine–not horrible, not great, just fine. I am not sure if they’ve improved it on the II, but it looks pretty similar.

As far as walk-around ENG-style lenses, the only one I use that fits that bill is the 24-105 F4L. It’s a really useful focal length on the Speedbooster. I have not used any E-mount lenses, including the kit lens which I didn’t buy because of its boneheadedly stupid 28-135mm focal length on Super-35, so I can’t speak to autofocusing or servo zooms. Frankly, the 24-70 (I think you mentioned that you had that?) would work for most of that stuff, especially paired with the 70-200.

Does that answer enough of your questions? What else do you need to know?