Thanks Stu,
Hehe, Tia, got it. ;o) The map in 1K will not really get better when the 1K is rendered just in 6K. Perhaps, so far I got your point correctly—if upscaling is needed, perhaps Photoshop does a better job, as you have other algorithm for that there available. However from 1K to 6K is 36 times more data—which makes me wonder if the result is of any use
(Funny that it is about a Dragon, as the RED Dragon has 6K as well… sorry, being a RED-Epic owner I couldn’t resist ;o)
Focus.
To split a scene in three parts is an idea to have the option to work with motion blur—and—lens-blur in a compositing application (... not targeting deep compositing here!). Typically the problem is that a depth—or—a motion map can hold only one information per pixel. With the mixture of motion and lens blur we have, more often than not, to deal with the need for more than one information per pixel. There is the problem (The physical Render engine overcomes this with no “sweat”!) Assuming that one area has the focus point allows then to mix the scene—with different blurs (MB—DOF).
There are reasons to work with blurry parts in an image. From a practical point of view, anything that moves fast and has no motion blur shows up a little bit jittery and/or rough. We are used to see different kind of motion blurs, and to name two main categories, one looks filmic, the other one looks video-ish. Both create an aesthetic that delivers a certain amount of information.
These are changing aesthetics, as we see more video (captured) based feature films these days, the term itself becomes blurry (pun intended). With high frame rates in the discussion for UHDTV or 4K acquisition some suggest already to go to 48 or 60 frames per second (some suggest already 8K and 120FPS). Which will change the perception again.
However, combining this with a shallow depth of field, will yield another option to create. Video was always more for Broadcast, and there a larger depth of field was wanted, to be able to move on without a focus-puller. Hence the small sensor size granted that wish “naturally”. With the above mentioned RED Dragon, the sensor size will be even larger than a typical Super35mm film, and the options to work with a shallow depth of field will be easier than with broadcast cameras in the past.
From an artist point of view, one might blur any information that doesn’t deliver information for the current scene. If done too heavy (because “one” can) limits the information density to push the story forward. Here as well, it might isolate the main character too much. All of that needs to support the story, not just to show that we can do it. To blur areas where the audience eye moves toward to (... always where the movement or light changes are), it might be frustrating to the audience to can not “see” clearly. Which is certainly normally based on a careful balanced decision, between pure aesthetic and cinematographically needs, to move the story (or the information flow) forward.
All in all, motion blur and depth-of-field based blur are great tools to work with, but the right amount is needed, not just maxed out as rule of thumb.
Your question is a cinematographic question. Whatever is needed to deliver informations, needs to be in focus. (Well—sometimes to leave it blurry might be “the” information!) If my material gave the impression that only the middle ground has to be in focus and foreground/background not at all, then sorry, that was not the idea.
If an actor moves from the background to the foreground, then the “rack-focus” would follow the actor, and with that change all “fore/middle/back-grounds” constantly while doing so, just to illustrate that with an example.
As a side note, if the “Bokeh” or light-blooming is created for the background (or elsewhere) the lights become bigger, less intensive (from a per pixel brightness point of view), but the background “seems” to be more illuminated. Which helps not only to keep the dynamic range down in the night shots; It helps to “power-down” super bright lights, and with that motion blur light streaks, which might look weird. If lights are beyond the dynamic range and focused, they are just white, if blurred (Bokeh), the intensity lowers and the colors can be “rendered” again (true for celluloid as well for sensor based capturing).
I certainly know that this is nothing that follows just a simplistic formula, like the one I described above, as each Cinematographer/DP will have his or her own idea how a story needs to be represented. (As well the overall look that is wanted… longer story)
Lenses are long story as well, especially Cine-lenses with so much more options (normally) than still lenses. Yet, sit with some DPs for a while and discuss “Glass”, you might wonder about the strong differentiations/opinions about the various lens kits they offer (I’m guilty here as well), so: I could go on and on about that of course here as well, but that was not your question. My intention was here more to open the mind for the endless possibilities given in practical lensing. So a simulation can be done. The more sensitive the perception to this aesthetic is, the better the representation in digital renderings can be, as far as the software can deliver here.
Any useful introduction of blur is limited on the “education” the audience had so far watching films and video. Those experiences changes naturally (just watch a movie from 1950 or so, you will get my point instantly). To be aware of that, allows to “tint” the experience your audience will have. My best wishes in using this wide arsenal of options!
All the best
Sassi