[Steve Bentley]

Does Max do that in the viewport in wirecage? Really?

You can do it in the render in C4d with a higher phong shader angle (phong tag). In the viewport you could choose another view type (Quick shade no lines) but you will still see hard edges beyond a certain angle – set by the phong tag.

[Steve Bentley]

This is tough to get truly right in C4D – in fact its tough even in RealFlow.
What about using a displacement map and animating a grayscale movie with waves and droplets etc then using that to push the polys of the water around. It should give you better control albeit with higher render times due to the displacement level you may need to get it smooth.
Let the collider do its job to get the geometry out of the way of the foot but then augment the secondary animation of the water perturbations with a texture movie.

[Steve Bentley]

Here’s a CS5.5 version in case you havent upgraded in a while. You will need trapcode for this to work. Modify and spin the camera as needed. In the newer version of the plug in you don’t need to link the camera to the comp camera – it just assumes the comp camera is the one you want.

12414_aecs553dstroke.aep.zip

[Steve Bentley]

I second the EXR suggestion – if its good enough for ILM who are we to complain.

And I’m not sure that QT animation best files are always enormous. They are run-length-encoded (lossless) so a 30 second movie of a black object moving on white will be positively minute in MB (maybe even in kB!). It all depends on whats in the frame.

To get an uncompressed format the size in bytes will always be the bit depth times the number of channels times the number of pixels divided by 8, and at 16bit in an HD or larger frame thats going be a huge number, so without compression of some sort (lossless or lossy) there is just no way to have a small truly uncompressed format.

Its a shame apple stopped developing QT. I love the animation best compression – but would have loved it even more in 10 or 16 bit. When you are making matte elements its positively microscopic. I’m actually kinda shocked that someone hasn’t done it already. Run length encoding isn’t nearly as hard to do as the voodoo that goes on under the hood with H264.

[Steve Bentley]

There’s something not quite right about that magnified view. I think there are two parts to the tire element. The main circular tire passes behind the profile of the road (which is a black opaque element with a white edge line so the tire below the road surface is hidden). The second element is a profile of the road in blue that sits slightly higher than the white line profile. The blue profile is passed through the matte of the tire element so it only appears where the tire is and the tire edge comes along and “wipes it” away and reveals it. This gives the illusion that the tire is deforming.

[Steve Bentley]

Actually if that works that’s brilliant. I hadn’t thought of shifting the UV’s. It would be the same process as the seed change in Expresso. (i’ll give this a whirl later and upload a sample if it works)

But (and I’ll have to keep looking for examples we’ve done) we’ve tried a number of times to have different iterations of the same texture on different objects (the Mograph shader can work sometimes – even on non cloner objects), but we’ve rarely been able to get there. I think it has to do with C4d (and expresso) populating and calculating geometry every frame, while the textures sort of live forever and are global. While you can pull textures into expresso they don’t have the same flexibility or iterational (irritational perhaps) ability.

[Steve Bentley]

The simplest way is to make two layers, one with the effect and one without. Then tie the opacity of both layers in the precomp to a property in the layer in the outer comp (a check box effect would do the trick or its transparency). Then when the check box is on set the top layer to opacity 100 and the bottom layer to zero. Do the opposite when the checkbox is off. And yes you can animate the checkbox(but only with hold frames).

[Steve Bentley]

What about the openGL being on or off? I get very different looks from FXy stuff depending on whether I’m letting the hardware give me my preview or the software.

[Steve Bentley]

What renderer are you using? The standard render still has issues with motion blur (errors, blur on objects even if they are still, a blobby unconvincing blur). If you need to use Standard for speed, render the motion blur as a separate motion blur data pass and composite it in AE with a blur filter that can accept that blur data pass (Real Smart motion blur for instance).
Otherwise use the physical renderer – its slower but does a better job on the blur and does it all in one. Check that your physical camera’s shutter speed is not keyframed or that the blur level is not keyframed.

What also could be causing the error is if something transparent or reflective passes in front of the object.

It could also be keyframes nearly overlapping. So each frame renders properly but the motion blur is being calculated over a number of frames and the space between frames. If a keyframe that may not appear to render (because it lies between keyframes) is present and of a much different position, it could cause the blur to suddenly “warp out”.
You can quantize your timeline keyframes so each keyframe lands on exact frames.

Or it could be the object gets too close to camera and the blur happens “inside” the camera (closer than the cut off for the objects close to camera)

[Steve Bentley]

I’m not sure changing the seed will work because its still one texture over many objects. So the noise will change but it will change on all objects linked to that one texture. (more on this kernel of an idea in a sec)

Good try with the world space, thats usually the first best solution but you are right with it waving around the moss is literal traveling through the 3D noise field.

I wonder if you made a link list of the moss objects in Expresso and then linked a single texture to them all but one where you fed in a noise to that texture, and into that noise feed a seed number, where the seed that is fed in is the iteration number of the moss object in the link list. That way you get a different noise for each object and that seed number is individually consistent for each object.