Camera for short films and commercials

[Satsuki Murashige]

One shouldn't really be basing a decision on cameras, from a demo video. The camera just encodes an image. Until you deal with that data in terms of the decoding pipeline (right up to which projector/screen you use, and under what conditions), you have no way of assessing the camera data.

 

Well, fair enough. I look forward to getting my hands on one, but who knows when that will be. It can't be any more difficult to shoot on than an AJA Cion.

 

I tested a Pocket camera against the F5 and 5D-3 pretty extensively for a short film two years ago, and while the latter two could be matched fairly close in terms of color (obviously the F5 was sharper, had more dynamic range, and a wider gamut of reproduced colors), the Pocket had a very distinctive color rendering that felt like it was missing all kinds of colors. That got me thinking.

[Kenny N Suleimanagich]

Carl,

 

Feel free to use my first name. The last name is a mouthful!

 

I find your observations on the color dyes and masking really interesting - it makes me wonder about digital color science and how imagers compare to film. I was always curious about how the late and lamented Aaton Penlope Delta, with its vibrating sensor, would interpret color (fixed-sensor imagers are commonly viewed as "harsh" and oversharp, maybe for this reason?). JPB was trying to mimic film's randomness due to grain crystals, and the few tests of it were very pleasing. It added an interesting component to the world of digital capture.

 

Satsuki, I do know what you mean about the BMD color science, but on the same note that camera does suffer from IR pollution and that might have something to do with it. Whenever I use a Schneider Tru-Cut 680nm, I tend to have less problems with skin tone. But I do agree about the brick-reds and funky hues being somewhat unpredictable when compared to Canon or Sony. I was just thinking that a lot (not all, but a lot) of these tests tend to be graded by people who don't do color for a living. So raw files, ungraded, might do more justice to the footage. I remember there was a firmware update a few months after the release of their first camera that completely changed the image for the better. That was probably the first instance in which I realized how important color science is to firmware.

 

I also wanted to add that the Steve Yedlin studies have been on my mind constantly these past few days. And this discussion goes well with that.

Edited January 22, 2016 by Kenny N Suleimanagich

[Mark Kenfield]

What sort of shoots are you doing Jan? The nature of the way you're working plays a big part in the appropriateness of the camera you choose.

[Satsuki Murashige]

Good point about IR, hadn't thought about that. Though my tests were interiors lit with Kinos and no ND, and it was still funky. I'll see if I can find the frame grabs.

[Adrian Sierkowski]

You know i haven't had too much funky on mine-- but I think perhaps the biggest issue with BMs is their QC; batch to batch there is little matching.

[Phil Rhodes]

I don't think they've figured out how to create pure sensor dyes yet.

 

On a single-chip camera you aren't even trying to. It's necessary for the green channel to leak a bit of red, and the blue to leak a bit of green, and vice versa, otherwise the reconstruction becomes more difficult.

 

This is how they claim to achieve anything like the headline resolution.

 

The problem, of course, is that the result appears desaturated, so they have to do "algorithms", which basically, as far as I can find out, means "wind up the saturation". Naturally this is an imperfect process and yes, there are colour problems intrinsic to single chip cameras as a direct result. It's not necessarily the ends of invention, but the reason the original Reds looked so noisy was only partially because of direct luminance noise - a lot of it was chromatic noise brought out by this process.

 

P

[Jan Tore Soerensen]

What sort of shoots are you doing Jan? The nature of the way you're working plays a big part in the appropriateness of the camera you choose.

Short films/commercials for Facebook and TV.

 

I run a, for now, one man or small crew production. Let's be honest here, I am not a colorist, so my perception of what a perfect color gamut is, is definitely not present. I grade for scene moods, and what looks good to me.

[Tyler Purcell]

Nice specs on the Ursa Mini 4.6k, but the footage released so far in my opinion does not look very good. The highlights seem to clip hard and over-saturate to the clip point instead of rolling off smoothly into desaturated white.

How have you seen this material, what was the viewing format? Some of the stuff out there is recorded at Rec709 and compressed to .h264 before presenting. I've seen the stuff on DaVinci (RAW) on a 4k color grading monitor and it was excellent. The demo material was both dark scenes and really bright ones. I had no problem lifting the blacks and lowing the highlights and still getting detail. The 4.6k imager has the soft clipping feature that you see on the other cameras, which is more similar to Arri, than Red or Sony which both tend to have harsh clipping as discussed many times.

 

And the color science still shares the same brick orange skin tones of all the other Blackmagic cameras. It's almost a two-strip Technicolor look with mostly orange and cyan, lacking pure reds, greens, and blues. I don't think they've figured out how to create pure sensor dyes yet.

From my experience shooting blackmagic exclusively for the last 2 years (2.5k, 4k and pocket, RAW, 444, 422, 4k, 1080p) in pretty much every configuration imaginable... they deliver a nice warm image almost all the time. It's VERY easy to turn warm into cool, it's much harder to turn cool into warm. I like warm because everyone today makes stuff cool (blue/green) and it's boring. I don't feel the camera has missing channels in RAW mode. However, in 422, the Blue and Red channels are more muted and do bleed slightly. It's not so noticeable when shooting a normal project, but it is when you're doing tests in controlled environments. The 422 color sampling just doesn't work well. Since most people with 4k would be shooting 444 color space, it's a moot point, because in full RGB mode, the issue doesn't exist.

 

Another thing to think about is how the camera is being colored. I've found the stock built-in REC709 LUT in the camera and the DaVinci LUT for the camera, to be warm in of themselves. When you color without a LUT, that warmth isn't there at all. In fact, the camera has very muted colors naturally. It really takes a lot of work to build the color in DaVici, more akin to Alexa then Sony or Red.

 

Pocket camera 422 mode no LUT, hand-colored:

 

 

Pocket camera 422 mode DaVinci LUT.

 

[Carl Looper]

In a single chip Bayer filter camera, the red green and blue channels can be considered as entirely separate from each other. Or to put it another way there's no leakage of colour information between the channels, but there are missing pixels in each channel. And these missing pixels need to be populated. The basic idea is that they are filled in through interpolation, but only from adjacent pixels of the same colour channel. So there's no colour pollution here. More subtle work infers a more detailed result using edge detection, and other techniques, to fine tune interpolation of the missing pixels. For example, an edge has a certain direction so it makes sense to bias interpolation (per channel) along such a direction rather than equally in all directions (which would otherwise give a softer result).

 

C

Edited January 22, 2016 by Carl Looper

[Phil Rhodes]

 

 

Or to put it another way there's no leakage of colour information between the channels

 

Ordinarily the dyes are set up to cross over slightly. It helps the bayer reconstruction, but hinders colour reproduction.

[Mark Kenfield]

Short films/commercials for Facebook and TV.

 

I run a, for now, one man or small crew production. Let's be honest here, I am not a colorist, so my perception of what a perfect color gamut is, is definitely not present. I grade for scene moods, and what looks good to me.

 

Well for one-man band/small crew (I'm assuming you mean 3-4 people) work, I think simplicity, light weight, wide dynamic range (to reduce the amount of lighting control you have to do) and versatility are probably the things you need most.

 

And at the moment, I'd say the Sony FS7 is probably the camera to beat in those areas. You've got built-in NDs, the ability to mount almost any lens ever made, 4k and HD recording options, 1-180fps variable framerates, a lightweight body, and lovely image quality for a very good price.

 

So that's probably where I'd be focussing my attention.

[Phil Rhodes]

 

 

I grade for scene moods, and what looks good to me.

 

Don't worry, so does everyone else. They just charge more.

[Carl Looper]

 

Ordinarily the dyes are set up to cross over slightly. It helps the bayer reconstruction, but hinders colour reproduction.

 

I see. So such would result in a certain amount of desaturation. I guess this would help derive a stronger luminance signal, which would be of help in appreciation of sharpness. Much like the rods in our eyes, which are more numerous, but less sensitive to colour. No doubt for the same reason - where sharpness has some more evolutionary benefit over colour.

 

C

Edited January 23, 2016 by Carl Looper

[Igor Trajkovski]

For very in-depth useful info on the FS5 and

some about FS7 when the differences are pointed out,
I'll recommend you this video:

Vocas Sony PXW-FS5 Introduction days presented by Alister Chapman

 

Not just great infos about the camera(s) but also a ton about the gamma curves,

the Sony sLog curves, what does it mean data/codec and exposure wise.

...

I agree with the above about keeping it simple and light.


Good hunting.


Regards

Igor