How I did it: setting up a crane shot

I’m working on a short animation which depicts the operation of an American Civil War-period blast furnace. During the sequence where we describe how raw-materials (charcoal, iron ore, and limestone) were loaded into the top of the furnace, I needed to show a worker pushing an ore-cart down the charging trestle which leads to the top of the furnace. And so, as I tried to animate my camera flying-along beside the worker as he pushed his cart, I realized… “this just doesn’t look right!” But what were my instincts telling me?

Answer: “WWARCD? = What Would A Real Cameraman Do?”

In a real shoot, there must be something for the camera to sit on, and that would not be a magically suspended track: that would cost a fortune (and be in the shot). No, in real life we would use a crane, with a very long boom sticking up in the air with our remote-controlled camera mounted on the end of it. As the actor made his walk, the crane-arm would rotate in a circular path, while the camera remained locked on the actor’s face (or some other point of interest). Framing would be maintained as-necessary by pulling zoom.

Your mind’s eye knows this, having seen it hundreds of times. Consequently, nothing else will do.

And so, that’s exactly how I set up the shot:

  • To give the camera something specific (and adjustable) to focus-on, I created an Empty, FocusEmpty, and parented it to the actor. - I positioned another empty, Dolly, where I figured the crane’s base would be. - I then added a camera, parented it to the Dolly, positioned it at the starting point, and added a TrackTo constraint which would cause the camera to remain locked-on the FocusEmpty that the actor was, in effect, “carrying.” I also designated that empty as the DOFObj, so that it would remain in-focus at all time. (An electronic dgital “focus puller.”) Now I have control of the framing in all three dimensions, when depth-of-field is added. - The camera’s position was determined by IPOing RotZ on the Dolly. This took a little tweaking (and positioning of the Dolly) so that the shot would smoothly track the actor in the frame. - Finally, a Lens IPO was added to the camera to accomplish the zoom. The actor’s going to turn a corner with his cart, and the dolly has to stop some distance away to “see” (and to cleanly frame) where he’ll wind up. Pulling zoom will finish the move (and subtly tighten the frame). Zoom is where all the framing fine-tune is done throughout the shot.
    Once this was done, the shot suddenly “looked exactly right,” because it now matched the cinematic reality we have all become accustomed to by watching movies. I had to deal with the shot as a real cameraman would, making only the adjustments that are available to him in real life. And the shot, suddenly, looked “real.”

The zooms and dollys had to be smooth, which meant a small number of control-points on the IPOs. Most of the keyframe-points I put in at first had to be taken out, as only two or three were really, so to speak, "marks that I needed the camera to ‘hit.’ "

I found that I have to set up all of the shots that way: the camera moves have to be “real.” Planning the shots is very much like what you really have to do in real lfe; otherwise, it just doesn’t “ring true.” You can look at a very preliminary, rough animation and “see it” right away. Therefore, you gotta fix it now.

Blender’s “rapid animation” features are extremely helpful here: I’m a long way from lighting-and-rendering yet, doing all the work in “solid” rendering-mode with one window set on camera-view. Just mash “Alt+A” and you can see a walk-thru at speed. Pressing and holding “up-arrow” to zip through ten frames at a time is almost as good. When you think you’re done, Ctrl-click on the “Render This Window” button to crank out a full-size (but lo-res) animation strip, ready to “play” or to cut into your growing animatic. (The “Stamp” feature of Render is helpful to “tag” the strip with frame numbers and camera-names for ease of rough-editing.)

With my workflow, I will actually generate all of the potential shots in “solid mode” quick-renders, and cut the whole thing together, before moving forward to any of the materials, lighting, or other steps. These steps will be done with the assumption that they’ll gradually replace the “roughs,” one-by-one, in the animatic sequence until the completed film is thereby complete.

cool :slight_smile:
gotta remember that one

can we see the shot?

Yes, I’ll be assembling “something to show.”

I have attempted a setup using the description above.

Anyone care to download and evaluate the BLEND file?
This scene is not a good test for DOF.


camera_dolly_rig.blend (220 KB)