A micro-tutorial about labor-saving total project workflows

Implied within the statement that “you want to do a Blender animation project” is the fact that “you want to get it done.” Whether or not your project is another Sintel, En Passant, or Big Buck Bunny, you want to have something to show for it. Unfortunately, a lot of earnestly-started projects don’t wind up that way. Either they are “never finished,” or, if they are ‘finished,’ the woefully-incomplete spots are painful to watch.

All of us, I hazard to guess, are at least partially attracted to Computer Graphics because we are Computer Nerds. :yes: We’re probably attracted to one (or more) of the general subject areas that represent the forum-topic layout on this site: Basics, Modeling, Rigging, Lighting, Compositing, and so forth. But … all of these things are specialties. They are silos. Only part of the greater whole. But a projectis “the whole.” It’s the whole “whole.” Both the parts that we are comfortable with (and interested in), and those that we’re not.

Our true “goal,” then, when we undertake to undertake ‘a project,’ must be something very different from this-or-that silo. It has to be: project management. :eek:

Here, in one short sentence, is “The Problem” that usually dooms a project to failure: “there’s just too much of it.” There are just too many matters begging to be attended to … stuff that’s easy for us, and stuff that isn’t. Stuff that we know how to do quickly, and stuff that hits us in the face like a red-hot cast-iron skillet. (“Pwangg-ggg-ggg!”) (Ouch.) Our tendency, of course, is to focus on the stuff that’s <<interesting | easy>> and to just hope that the rest of it somehow goes away. Well, it doesn’t. Hundreds(!) of hours of computational effort can wind up being … thrown into the trash.

So: “project management.”

And my micro-tute, simply, is this: “Notice what Hollywood does, and do that.”

Hollywood never “pushes a button and A Movie pops out.” Instead, they plan everything very carefully. They work out, as best they can, what the entire sequence of the movie will be, shot by shot. Then, they construct a set (carefully “to scale”), and they put Cameras onto it. Then, they put Actors onto the set, give those Actors lines to speak and “marks” to hit, and … they shoot a lot of film, “covering” the action from many different angles. Every strip of film that is shot is carefully cataloged and identified.

The next step in the process is where the real movie-making magic happens: editing. What used to literally be done using a razor-blade and celluloid tape is now done using a computer, but the process still consists of slicing bits-and-pieces from all the various pieces of film that have been shot and stretching them together to make a single, continuous, sensible, compelling “show.” It’s a great deal trickier than it sounds, because a good edit is often a matter of fractions of a second. Many editors candidly state that it is they who make “the movie,” and their assertions are probably correct.

So … how does all this relate to your CG-project workflow? Well, in a word, it is: “go and do likewise, but with a twist.” Get out there and shoot film … a lot of film … from many different camera positions, and more film (from any of those positions) than you think that you will need. Catalog all of it, and use the Stamp feature to add descriptive information: file-names, camera name, lens, scene, marker, timecode, frame number. Use the OpenGL Preview feature to do all of this, and work only with these renders … and with the video editor of your choice … until you get the show(!) completely finalized.

By approaching your project in this way, you’re following the tried-and-true Hollywood workflow … with just one important exception: the “footage” that you’re “cutting” isn’t final yet. It’s low-res, fairly crude looking, but it depicts “to-scale” objects moving about in “real space” with “real cameras” (shooting with “real” camera parameters) recording the action. Thus, each clip is actually an exact(!) stand-in for what will eventually be “the real thing.”

(This ain’t nobody’s “animatic.”)

When you finally reach that “Final Cut” moment, go through the entire edit and take notes, in a spreadsheet, of shot by shot, cut by cut, exactly what footage remains. Browse through it carefully, because this is your “to-do list.” Some of the shots are long and important; others are just a fraction of a second long. Some represent the first time that the audience gets to see something; the rest do not. Some models demand agonizing levels-of-detail, some do not, and some fail to appear at all. (Now, you know.) But, given all of this, now you know how to organize the rest of your project. And so, you do. The “important” shots (and the complicated ones) get lots of attention; the passing ones don’t. The stuff that (you now know …) “no one will actually see” gets ‘cheated.’

And … your project gets finished, because you pushed all of the decisions of “exactly what to do” ahead of the decisions of “how to do it.” You didn’t waste time producing what you wouldn’t use. You didn’t lavish attention on things that wouldn’ pay-off, so that you had the attention left to lavish upon the things that would.

When you take this approach, it will surprise you. Greatly. Unexpectedly. You’ll find yourself ignoring great chunks of ‘film’ that you did generate, while going back to generate other clips from other angles. It unlocks a form of pure creativity that you’ve seen in the movies and on TV for many years but never actually recognized. And, in the computer-intensive world that is CG, it might save weeks or even months of time.


To continue . . .

You want your film to be “finished,” yes, and you also want it to look “real.” But CG shorts often do not “look real,” even if they are “photorealistic.” There’s something wrong, to your mind’s eye, even though perhaps you can’t put your finger on exactly what it is.

In the film-language of cinema, the Audience is a participant in every shot. The Audience is represented by the Camera(s). Having watched many films and TV shows, your mind’s eye knows what to expect – what the finished picture “should” look like, and how the shots “should” be arranged in order to dramatically(!) portray a particular bit of action. You know that film-language already, but only as a Gentle Reader. You must learn how to be a writer.

Here are some specific things that you should pay attention to when setting up your movie:

“Real” Camera Placement and Moves:

In the Real world, cameras are big, heavy, and highly-visible. Many of them are fixed, on tripods. Those that move, have to be moved using gear, such as “dollys,” “trucks,” and the flying-booms that you often see being used at rock concerts. They have limited range of movement, and they have zoom lenses. (“Focus pulls” are used, but rarely.) When setting-up the shot plan for a particular scene, the cinematographer must ensure that none of the “necessary gear” will be visible in other shots. You need to do the same thing: using empties, parenting, and track-to constraints, put your cameras onto “rigs” and place them so they do not quite “see” each other. Then, use them realistically.

For instance, suppose you have an ironworker pushing an ore-cart along a gangway that’s 30 feet up in the air. How can you do that, in real life? Well, you can get a “front shot” from a rubber-tired truck that’s being pushed steadily back as the actor walks forward. The only way to get a “side shot” is with a boom, unless you plan to build scaffolding to put rails on, or you have established in a previous shot that there is a parallel gangway. Otherwise, the audience subliminally expects the shot to show the approaching actor at about a 45º forward angle, and to track him with a slow circular pan as the camera itself traces a semicircular path through the air. The audience expects the framing (zoom) to tighten as the actor gets closer, then to loosen again as the actor recedes in the distance. You know the move. You’ve seen it a hundred times. Go, do that.

(Avoid At All Costs …) “Rubbery, Flabby” Action … and Unnecessary Screen Time:

A CG “actor,” driven by F-curves, usually has flaccid movements. Say he’s going to push the call-button on an elevator. You see his finger (already positioned, because the animator’s lazy) coast into motion, coast up to the button to push it, then coast away, all in a straight line, all in one continuous shot. A full ten seconds, 300 frames. “You’re looking at eleven hours of Cycle rendering-time, buddy, so you’d better admire it!” :yes:

Uh uh. :no: Real actors don’t move like that, and you don’t need to show the entire sequence on screen. But there are two fairly-easy things that you can do about it.

First, use the NLA Editor aggressively. Real movements are a combination of other moves. You “ease in” and “ease out” of moves, but you don’t do so at a continuous, sinusoidal rate. Adjust those F-curves. The finger moves swiftly forward, strikes the button, then the hand drops down. Decisively.

Second, having shot that sequence in its entirety with the OpenGL Preview feature, leave most of the footage on the cutting-room floor … never rendered at all. Unless the pushing of that elevator button is a Major Plot Point that should be accompanied by an orchestral something-or-other, it deserves a couple seconds, max. Which would be edited out from a longer filmed sequence. Leave it on the floor where it belongs, and thus keep the story moving.

Dramatic Cinematography:

Another typical problem in a CG short, in addition to the shots being too long in duration, is that they are uniformly, boringly, flaccidly medium. The camera is placed from a comfortable distance away, and it covers all the action and the surroundings, loosely-following the actor from about twelve feet away. Equals: b-o-r-i-n-g.

Real cinematography uses a language of “establishing shot,” “medium close-up = MCU,” “extreme close-up = ECU,” “over-the-shoulder = OTS,” and so forth. The shot is filmed from multiple angles and is presented in a sequence of (usually) tighter and tighter shots which reflect either what the audience would want to take a closer look at, and/or what you as the director want them particularly to see. Things that are irrelevant to the intended visual dramatic impact of the shot are either not-filmed, or they’re edited-away. The shots follow the line of action and stay on the same side of that line.

The shots are “cut” very tightly to establish the pace and to eliminate things that the audience doesn’t need to see. For instance, if you’ve got the camera pushing forward and zooming in at the same time, for about (say) fifteen feet, you can start the move then cut-to the point where the camera stops. The audience gets the idea, and you save render time.

Dramatic Light:

Y’know, I think that Cycles is a wonderful invention … but (IMHO, and so-to-speak) the lighting that it produces fairly-uniformly sucks. :wink: It’s wonderfully soft and flat. Wonderfully. Soft. And flat. The set is filled with soft-boxes and there’s not a spotlight to be seen. Equals: b-o-r-i-n-g.

The language of visual drama is the language of light. And shadow. The full available tonal range is used, without ever quite going blown-out white or dark, and it is used dramatically. If you’re feeling ambitious, a combination of BI, Cycles, and even OpenGL techniques can be used together, to produce a shot that has dramatic key-notes against a uniform base exposure. The much-maligned “ambient lighting” (“light that came from everywhere drove shadows from the room”) can be used, too. And, the shadow-only spotlight, a pure-CG invention, can be used to inject shadow without adding light.

Light, also, has color. Blue light from the sky, yellow from the sun, icky green fluorescents. Too many CG artists light their entire sets with not a “gel” in sight. But, go to a real theater and you probably won’t find a single lamp up there that doesn’t have a gel of some sort.

“Hope this helps.”

For someone who has ideas about a short and was planning to do something about it, I find that extremely helpful. I didn’t think of using OpenGL to plan editing, and now that I think of it, this is the best option, and the less time-consuming.
Other than that, there’s an art for manipulating cameras, manipulating light, manipulating meshes, creating meshes, and all that …

Your suggestions Sundial are really well thought out, and I think they will be really useful to anyone considering doing a short film.

Discovering openGL renders was truly a gift for me. You can never seem to get timing right in many animations because your viewport can’t display accurately the 24-30 frames per second… or you just plain out got it wrong to begin with. What used to happen to me is I would work out the animation, then render normally… it never failed that something went wrong. So hours worth of rendering time was wasted.

For those who don’t know what openGL rendering is, it becomes a lighting fast render… If something took you 2 min/frame to render… open GL will render it in like 10 sec frame. You get a dead- on accurate animation based on your frame rate and you can quickly and easily see where you have messed up. Your output on these openGL renders will be pretty much what you see in the viewport… but the key is that you will get accurate representations of your animation without the painfull rendering… You render just like you normally would, but instead you simply select open GL render instead of F12. Make OpenGL a top priority to learn if you decide to animate.

Now what sundial is correctly suggesting, is to take openGL even a step farther… shoot various camera angles of the same scene which gives tremendous flexabilty when you get into the editing process. You know before hand, that you won’t use all these camera angles, but it cost you next to nothing in time to openGL render some different approaches to the same scene. Then you simply bring these various openGL renders into your editor and work out which one would be the best to use.

While I’ve used open GL, I’m committed to trying to use it more in the actual editing process on future projects I do.

Again I think all your recommendations will give current, and aspiring short film makers some good tips to think about.

OpenGL Preview is your “BFF” Best Friend Forever" in that department, because it is both “fast” and “exact.” (Furthermore, for a lot of shots, you can use what it produces!!)

Use the “Stamp” feature on all outputs (timecode, file, scene, marker, frame#), use lots of cameras, and use “Scenes” to prepare the individual shots since each “scene” has its own output-file designation, active camera, start/end-frame and so on.

Set up a rough set, carefully to scale, and put rough stand-in objects, carefully to scale, on it. (Start right out by using linking and libraries: eventually, the stand-ins will be replaced by the real things, in those files.) “To scale,” and a rigid definition of what “one Blender Unit (BU)” corresponds to in the real world, is crucial.

Block out the action, set the marks.

Position, and name, each camera. Not sure which of two similar angles you like best? Splurge!! Cameras are free! (Don’t move 'em: add a new one. Once you’ve placed one and shot with it, use the Outliner to make it non-selectable with the cursor so that you can’t easily move it. You may never shoot with it again, but there it stays.)

Then … shoot. Get “coverage.” Experiment freely. Keep everything. On a Mac, whack that “Time Machine” button frequently.

Personally, on my Mac I’d use Final Cut Pro or its very-capable kid brother, iMovie (the latter comes free with your Mac) to do the editing, because it’s a fast and flexible tool and of course because I am very familiar with it. (VSE is very powerful, too.)

And … this is where the movie starts to come together for the first time, and the last. “And you haven’t even rendered it yet!” (And you won’t, until you’re done.) You’ll never look at a movie or a TV show in quite the same way again . . .

You can be “goofy wasteful” on the OpenGL shooting that you do, because you can see previews in real-time and you can generate footage for detailed editing at less than a second per frame. (And, as I said, some of those shots are something that you can use directly in the film. Some passing-fancy that’s on screen for a second or two doesn’t warrant “an hour per.”)

When you’re rendering, you won’t render a single frame that you do not already know that you need. The final-cut edit says that you need “Frames 103-107, 225-460, and 801-824, from Camera X in Shot Y,” and that is all that you do. Drop them in at their pre-determined places: you already know they match. You’re not making shot-decisions now: the movie’s done, and now you are merely generating and assembling it.

(The ability to do this is “what you’re (so to speak) shooting for,” and you can see that it requires a lot of discipline and precision. The ability to decide, and to stick with that decision for weeks or months.)

I happen to be very fond of Dan Alban’s Digital Cinematography and Directing, but an Amazon search on “Digital Cinematography” produced over 100 titles for me.

Also … let’s be specific about what I’m talking about, and where to find it.

  • Right there on the bottom of any 3D window you’ll find two small icons: one looks like a camera, and the one next to it looks like a movie clap-board. Although the “tooltip” for the two is identical (2.7), the left button OpenGL renders a single still frame, while the right button OpenGL renders the entire image sequence. The settings on the Output section of the Render tab concerning file-location and file-format will be used. (Usually, I render to an AVI or MOV movie-file format for convenience at this stage.) Be sure that you have selected the proper viewport!

  • The Stamp is accessed by the section by the same name, also on the Render tab. Checkmark the box to turn on stamping, then open the section and choose what you want to see. (I use Time(code), Frame, Scene, Camera, Lens, and Marker.)

  • To see the animation that you rendered, choose Play Rendered Animation on the Render pull-down menu at the top of the Blender screen.

You can get a pretty good guess of what the sequence is going to look like by using the “VCR Buttons” at the bottom of the Blender screen, but this is a much rougher approximation that is mostly good for modeling and camera placement.

The single most-important “tip” throughout all of this is: “Edit, Then Shoot.™”

The Timeline with its ability to “Bind Camera to Marker,” is handy for your very-first explorations, but I still pull a complete film from each camera position separately, and do the “real” edit in the editor program. (You need to shoot some frames “on either side of the action” from each camera so that you can finesse the edit just-so.) Keep the Timeline scene with its marker-binds as a quick reference for what you were first thinking-about at the time, but don’t try to generate the final that way.

Keep notes. Everything, every day. I use … a loose-leaf notebook and a number-two yellow pencil. :slight_smile:

And I think that people can become attached to the Blender naming convention of Scene. Remember that it doesn’t have to be a an actual complete scene, in fact a “Blender Scene” is often a just a single shot. Especially when you are making an effects movie, with real vision to comp into.

Indeed. To me, a “Scene” is basically: “the thing at the top of the (visual …) data-structure hierarchy, within which all other things live.” Significantly, though, it contains its own file-output destination, its own range of frame-numbers, active-camera selection, and several other pertinent bits of “global” information that you might want to change.

In my workflow, I set up a base-scene, which is linked to the ‘set’ (scene …) and various objects used in the shot. Then, each camera-angle that will be used to film that action is its own scene, each one linked to the base. Each camera typically (these days …) belongs to the scene that uses it. Everything is scrupulously named and recorded … in a big fat three-ring binder(!) populated using number-two yellow pencils.

The “objects used in the scene” start out as stand-in objects which are to-scale. One by one they’re replaced with more-detailed objects … and I have a little trick of switching them around by re-naming the groups to which they belong. I try to "start-out as the project will eventually end-up, but to do as little ‘work’ as possible in the beginning … letting me go straight to ‘creative experimentation’ and to stay in that mode as long as possible.