[Steve Bentley]

Well you have a couple of issues there. First you don’t have a lot to work with as far as perspective goes. You can only see the top rail of the bucket really. Its a shame that instruction card on the front with the yellow and blue arrows isn’t bigger because that would be perfect (and might still be). But the wider you can get the tracking points apart the more accurate the angles usually are.
But that being said, often perspective doesn’t matter and its often better to get a good track off one point and then do the perspective by hand (if even needed) after the fact with a corner pin or skew.

Second, as you pointed out, the massive jerk when the lift starts. There is going to be so much motion blur during that section that it will be very different from the rest of the track and hard for the track area to lock on. You might have to do those few frames by hand and then resume the track.
Make sure the motion blur is turned on for the layer that is following the track or it won’t look right at all.
One other thing that you need to be aware of with video cameras is that if the motion is fast enough you can get a distortion since the sensor isn’t fast enough for the motion (telephone poles going by can have a curve to them). For these situations we’ll often take the last good frame and the next good frame, line them up and move them by hand through the same kind of motion while dissolving. That way the move is clean, but so blurred you would never know that it was just a set of stills. Just take the blur off and track the shot again after this fix is added. Then use the real footage or substitute the blurred doctored clip.

There is a massive lighting change as she drops below the floor level. you should stop the track there a few times and restart so the system can accomodate the new look of the points you have chosen to track. Thatr yellow color is really going to affect the green dots.

I see you have some green dots to track but if you are adding other spots, keep in mind that reflective things don’t make for good tracking points (the lift tubing for instance) because the reflections keep changing so the color of the area you have chosen to track will keep changing and these reflections can change faster depending on whether the reflections are facing camera or are around the curve on a tube.

One of the tricks I use for trouble tracking shots is to isolate an edge with a slit mask. So a vertical black mask precomped onto the footage that only lets through a few pixels , and then track in either the X or Y only. You have to do this a few times in both X and Y but it will let you isolate something that has good contrast but coudn’t be tracked because it is just one long line. There is the straight part of the plate in front of the talent. Good contrast between the white of her and the gray of the plate. But it’s just a straight edge. Put a thin vertical mask on it and the system will have no problem tracking it’s vertical movement. Then pick something else in the area and put it through a slit mask to get the horizontal movement (like the vertical of the same plate where it crosses her dark midsection).

[Steve Bentley]

You could also just make a spline the shape you want one layer to be with points on the spline where each sphere goes. Put that spline (or splines) in a cloner and have it generate all the layers – then collapse that cloner with make polygonal. Then use the atom array on the resutling objects and just make the spheres bigger than the tubes.

[Steve Bentley]

Without seeing the file it could be a couple of things.
First, did you render with an alpha and if so what kind? Depending on what you are viewing that image in the alpha might not be interpreted properly and it can give either a white line or black.
Does the texture map have an alpha? if it does make sure that the color of the image extends past the edge of the alpha.
Is your picture-as-texture-map on a white background? Those edge pixels might be leaking in when the map doesn’t quite match the cover of the book.
Try right clicking on the texture tag and choose fit to object and that might snug it up.

A quick fix is to just crush the alpha levels of the image coming out of C4d. In AE you can use a matte choker and in Pshop you can just use Levels to crush the blacks in the alpha and bring that edge in a little (drag top left slider to the right a little in levels)

[Steve Bentley]

Not really sure what you are after – can you post some examples or images – that might help.

[Steve Bentley]

But this is a destructive method and you can’t go back to adjust., So I usually make a copy of the symetry set up before I collapse it and then hide that for editing down the road.

[Steve Bentley]

One other thought. You are going to be wasting a huge amount of light and resolution because your projector can’t do that format and encompass all screens in one shot. So there will be tons of pixels above and below the screen. This would be better with three projectors, one for each flat. The larger you have to blow up the image the dimmer that image becomes and for that size of screen, 4k isn’t a lot.

[Steve Bentley]

You can collapse the symetry to polygons the same way you would for a primitve and then delete the original object. Depending on the object type you may have select it’s points and delete those. The two objects will be joined into one so just select the points of the first one and hit delete. The animation track will remain.

[Steve Bentley]

I dont think they need a display tag (but it couldn’t hurt- animate the visibility – btw, the animation of this attribute doesn’t show in the view port but will show properly when rendered – in the viewport it just sort of pops on when the value is >0)

Yes you could use a cylinder (not a disk) and then fillet the caps to get that rounded lozenge look.

You can just hide the circles (or flatted spheres or lathed arcs or filleted cylinders ) inside the cubes and then scale them or move them in the y until they poke through the amount you want when you want. You should child these inner objects to the outer parents so that you don’t get the issue that is visible in the reference animation. That way when the parents rotate they will carry the children with them.
When you nestle an object inside a parent you can easily center the child on the parent just by zeroing out the childs position and rotation values.

I wouldn’t worry about my last comment about the displacement map, it more a point of interest – there’s no need for the displacement, a hidden object inside will give you the nearly the same result and with a lot less render time. I was just interested in the kind of care that was taken in some aspects of this animation and then the lack of care in others.

Let us know if you are going to use PLA as there is a trick that will allow you to use easing on the motion.

[Steve Bentley]

But then (this has turned into a regular Law and Order episode), the single yellow circle “surfaces” and the reflection on the blue surround changes – I don’t think that would have happened if an object just pushed through the surface – so some of this might be displacement maps too.

[Steve Bentley]

Thats a combo of (probably) motion blur on the object being scaled or moved, and radial blur found in the effects/blur menu – set it to zoom.