[Steve Bentley]

These might be perfectly flat poly’s (I know, I know, all poly’s are perfectly flat) but I mean to the camera’s eye. A perfectly flat poly will only reflect a single pixel of a reflection map adn sometimes the same goes for a shadow map. So as the angle of incident changes as the camera moves, it’s like you are dragging that reflection map across an opening that is only 1 pixel wide and that one colored pixel is getting plastered over the entire poly no matter how big it is. That pixel will change drastically as it moves.
Try adding either a subtle bump map of noise to the texture (standard noise scaled to 1000 and set to 2 in the bump strength) or try increasing your angle value in your phong tag on the object to “round” things out.

[Steve Bentley]

The hilights etc have to be baked into the layers that you use the puppet tool on (Duik or not – You can do everything Duik is doing without it – he has just made an elegant interface and pipeline that codes the expressions for you and links dependent layers – so it’s not doing anything magical – its a great script but doesn’t change the puppet tool one bit)
Puppet tool makes a mesh of the pixels and if the hilight or shadow is not part of those pixels they will not move in concert with how the puppet tool bends the underlying image.
If you want an edge light I would recommend an inside glow after the puppet tool. You can animate the direction the light is coming from to keep up with the deforming object. Same goes for the shadow (just a dark inside glow).

[Steve Bentley]

Quality wise you should not see a difference.
Math wise, scaling can be a nightmare. Scaling is logarithmic. Scale something twice as big, it increases by 4 times the area. So if you are trying to pull out of a shot, and if you use scale to simulate this, you will pull out faster and faster as you go even though your scaling factor is constant.
With a camera you have better control over that and the eye is used to what a camera looks like when it pulls out and the speed the camera and the widening landscape progress at.
Add to that, you can change the focal length if you want to change the feel of the move – pushing in with a telephoto lens looks very different than pushing in with a wide angle lens – there really is no way to cheat that using a scale function.

The downside of cameras is they are still a little clunky in AE vs a true 3D package.

Remember that you can look at your scene from another view point (top, side, even isometric – on an angle) and in these views you have to pan with the camera tool to bring objects inside the view port, and this doesn’t affect your main camera’s position. Objects outside the main viewport are just wirecages of themselves. You have to pan or zoom them inside the viewport, in a top view for instance, if you want to see the color pixels.

Use nulls to nest your main camera so you have better roll control and so you don’t end up with gimbal lock.

The upside of cameras is that you can nail parallax shots. With 2D you are always guessing how fast something should move relative to something else. Plus you have built in depth of field.

[Steve Bentley]

Then you can use the “start” attribute to retract the line so that its tail gets eaten up as it disappears. Think of it as the start point being fixed and the “end” of the line extends outwards as it grows.

[Steve Bentley]

Oh I see what you are doing now. To keep the dashes in place you have to use the “end” to change where the line ends as it draws on. You can use the “start” to draw lines on or off but you have to use the same animation curve or keyframes in the offset attribute too so that as the start position moves along the line the dashes get offset by the same amount.
Normally you rewind the “end” attribute to zero, set a keyframe and the later in time set the “end” to a higher number and the line will draw on with no displacement of dashes.

[Steve Bentley]

That shouldn’t be happening. Are you animating the path? changing the number of points over time?
You could try animating the offset in syncwith your start to finish animation but again, I don’t see this problem here in V2018

[Steve Bentley]

Based on the footage I’m not really understanding the problem.
Use two layers in AE – one for the bookshelf and the easel and just draw a mask around them (it seems its a locked off shot so that should be a still mask). Inother words you are going to mask these objects so they show up in the comp, don’t try and get rid of them.
Then do your key for the people on a second layer.
If you apply the same color correction effects to both of these layers, even though the matte of the easel etc will be horrible in the layer with the people keyed properly, the masked layers of the easel will fill in all the transparent bits. (people go on top of masked easel layer).
You don’t have to do your green spill correction in the keyer effect, you can apply any color correction and spill killer to an adjustment layer that sits over top of both the people and the masked easel layer. Then both of these layers get precomped and that precomp goes over your background. I would stick a background into the precomp just to work with for reference when adjusting your color correct and then turn that background off. Thay way you aren’t looking at the final and popping into tweak colors, looking at the final, popping in to tweak colors over and over.

[Steve Bentley]

The key frames that turn the wheel need to be in the application that is adding the motion blur/ Its the animating that does the blur – once it is just a set of pixels you would need realsmart motion blur or some other package to pick the motion out of the footage to determine the blur – and that doesn’t always work as you might expect.
So if you did the animation in Premier and brought that footage into AE, all AE knows is that its just a movie. It has no motion data to work with.

[Steve Bentley]

Hey Eric,
In R19 the vector blur does seem to be fixed. I rendered vector motion and checked the final resting frame with a still frame with all blurs removed and they are identical. The change from blurred to non blurred in the animation seems reasonable too. So with an upgrade you should be golden.
There are also a lot of new features that should speed up your workflow including a better preview with reflections in the camera port!

[Steve Bentley]

Well we’re pretty used to exporting nulls and lights and cams to AE these days so it wouldn’t be that much of a stretch to do the same to an app that would let you move about in that world in real time and then send that motion data back to the main C4D file.
But and its a big but, where we really find this kind of app sings is when you have all your performance animation done and then you are just playing cameraman. I don’t know how you would get enough data playing that you could see a character (or 2 or 5) “acting” so that you could react to that performance as cameraman.
Don’t get me wrong, I’d use this kind of thing a hundred times a day and pay for the privilege, and someone is going to make one sooner or later (if we’d had the time we would have created something already).
It should be tablet capable too as you get more control with a bigger object to tilt (fewer round off errors and more accurate angles)