This content has been marked as final. Show 12 replies
I think you need to precomp the 2 layers and apply the motion blur to the precomp. Then it would like like 1 layer with no gap.
Thanks, David. I tried that too. The problem is that I need "Collapse Transformations" on to 'reveal' the 3D nature of the comp. That blows the flattening away. Bummer.
This is just an artifact of how motion blur is calculated in AE. In real life, motion blur happens to the entire image as seen by the eye or the camera. In AE, on the other hand, each layer is blurred separately for efficiency. It's just as if you place two layers side-by-side with their edges meeting, and applied a blur to each layer. You'd get a transparent seam where the blurs overlap, since each layers' pixels are at less than 100% opacity. You can get even uglier artifacts in some cases when a 3D layer's motion blur actually intersects with a layer that would otherwise obscure that blurred layer. Layers can even appear to render out of order in some cases, thanks to this phenomenon.
If precomposing isn't possible the only way to work around this problem (other than resorting to using something like Force Motion Blur on your entire comp) is to add a layer behind the seam to create the illusion of a continuous surface. It can be really tricky, but it is possible to make it look more natural.
Thank you Aaron. Sometimes it's enough to know I'm not just doing something 'wrong'.
>I know this is comparing apples to oranges, but if I perform the same
>setup (2 planes, moving camera, motion blur) in Softimage, mentalray
>blurs plane 1 into plane 2 but never 'reveals' the seam with a decrease
Oversampling on a per-ray evaluation basis is drastically different from how AE does things. Aaron's explanation quite nicely covers it - AE simply assumes motion blur to be an accumulation of different pixel values at given time intervals for the whole layer whereas in a 3D program the ray would correctly intersect all scene geometry and evaluates all its properties at a given point in space and time. If you will, the concept is completely different.
You'll probably get less edge artifacts if you use the Alpha Add layer mode. Worth a try...
I think Alpha Add depends on there being empty alpha behind the seam. If there is a background layer, the Alpha is already 100% opaque, so there's nothing Alpha Add can do differently from "Normal".
Um...why not Command/Shift/Y (Mac) on one of the solids, increase the size of it to cover both solids & delete the other?
I'm actually 'building' things in After Effects using solids. One is a klieg light rig you see overhead in concerts, another is a hydraulic press stamp .. these are all made with solids and masks and stencils with textures to be 3D objects circa late 80's. Very rudimentary, but get the job done quickly in AE instead of built, rendered etc.. in Softimage.
It's not a matter of increasing the sizes of the parts .. doing so would make the sides and corners intersect and go 'beyond' each other.
Start by setting the blending mode for all 3D layers to Alpha Add. This will minimize the leaking edges.
Next, Select all 3D layers and pre-compose them. Do not pre-compose any lights or cameras in your comp. Collapse transformations in the pre-comp.
Now apply CC Force Motion Blur to the pre-comp. You'll probably want to double the shutter angle in the effect.
There you go. This is as good as it gets with motion blur and edge matched 3D layers in AE. Don't expect a fast render.
Thank you all for your help. In the end I've started doing what Rick and others have suggested - CC Force Motion Blur.
Stinks on these 1080 HD comps though. Too bad it's not adaptive .. I crank the samples for objects that are close but the background items with little or no motion get the same attention paid to them by a machine that may very well need to go in the freezer during the render!
I had no idea that would work, Rick. I'll have to give it a try sometime. I assume it disrupts the render order, though, right?