[Steve Bentley]

after playing with this for a bit it seems that the refraction of the inner object is added after any kind of feathering done – whether that be in the alpha channel or the trani channel (I’ve tried both grads and fresnel). And that refraction will always have an edge because the refracting geometry does. ( I haven’t tried a pyrocluster shader yet)

If you put the refraction index to 1 (no refraction) it works well but that removes any of the sexy glass effects. So you might have to dual pass this effect. One pass will be everything except the refraction and then composite the refraction elements on top. And even then the inner objects refraction will have to be rendered on its own with a supporting alpha to feather it.
I thought setting the refraction to 1 on just the inner object would work (with a fresnel fade off at the edges) but it seems that the refraction algorithm works on all objects within any surrounding refracting body no matter what the inner body is set to.

Personally (especially with the tendril idea), I would want to render that element out on its own anyway so that I could comp and finesse how it blends into the outer object and the scene as a whole.
Octane or Vray may handle this better. A true ray tracer may have better luck when the ray contacts that feathered boundary of the inner sphere. But that will add huge time to the renders. Or it may just be a stacking order issue with C4D as whole (refraction after transparency). Houdini could do this.

There are a few other gotchas with glass objects – motion blur doesn’t work on anything seen through the glass, neither does depth of field. Glass over nothing has all sorts of complications especially if there are semi transparent areas – those areas don’t always map to the alpha properly when there is no background. Luminous glass objects also don’t always work depending on whats behind them.
So whenever we starting thinking about doing glass, there is always a “lets think about this” moment to make sure we’ve covered all the basis for what it has to do in the scene.

What is the final object going to be/look like? Maybe there’s another way over the cliff?

[Steve Bentley]

For extra credit: if you are animating the points (vertexes) of the objects and using PLA (and you dont’ have the latest version of C4D where you can have easing on Point Level Animation) you can use nulls linked to each point in expresso and animate the nulls, which you can have easing on. The points will puppy dog along behind. It also allows you to manually animate things without switching to point mode.

[Steve Bentley]

You shouldn’t need to go to python for this. I haven’t see the demo video’s (I’ll watch later on if this doesn’t help), but I’m pretty sure there is a section in the “my motion graphics” video that shows the use of the mograph shader. This allows you to have a folder of images or other textures to be populated onto many objects.
You can also use the shader effector to map different effects to different objects, either based on a noise pattern, a fall off or the object’s initial color (or other attributes – which can be set by a shader effector itself). The process is very iterative – one effector influences another effector and on down the line.
And remember, you can clone a cloner.
You can also put a material on one type of object – cones for instance, then put that cone in a cloner along with other objects with their own material, and as long as the cloner doesn’t have a shader, the cones will have their own “look”.

[Steve Bentley]

Have you set the preview region? Set the in point with the “b” key (for “beginning”) and the out point with the “n” key (for eNding). There has to be something to preview.

[Steve Bentley]

I think every one of these is based on the same thing. (whether its the solutions below or not, they are all based on the same effect – they are similar enough in what they look like and how they react, plus people are lazy – if they find something they like they will use it over and over again.

Its going to be either CCglass, CC Mr. Mercury, or the incorrect use of CC Vector blur (without a source – why does no one know how to use this last effect properly?). Then a second application of Colorama may have been employed to rotate the color wheel. Probably some Hue saturation and some Levels or curves as well in to increase the contrast and shrink the roll offs.
Each shot is going to react differently to these effects and some of the color corrections may have to happen before the main effects themselves.

[Steve Bentley]

Try saving your rose as a PSD with an alpha channel (and as a flattened imaged with a background), or as a PSD with a transparent background (no background layer and the rose with a transparent surround).
If you go the alpha route, two things: you may need to interpret that element in the project panel of AE as premultiplied (for the alpha) or you will have to extend the pixels of the rose beyond its own edges so that you don’t get a halo of the background color of the image sneaking through the antialiasing of the alpha edge. This is easy for a rose, just put a slightly scaled up version of the rose behind the current one and flatten the file. (make sure you have the alpha in the channels menu first).
Then you can use a normal blend mode in Particular.
I’m not sure why the png with a tranparent bg is not working, but I have seen this happen before due to how it was created.
Any file with alpha capability can hold an alpha but there has to be an alpha to hold. So Tiff, EXR, TGA, PNG, etc will work but not jpg.

[Steve Bentley]

Good point Roei,
Once you have your motion path perfect, the speed graph is a good choice for adjusting the timing. However, the speed graph is anathama for me to the degree that I will precompose the keyframes of the moving object and use time remapping (and another value graph!) to adjust the speed, keeping all the temporal resolution and the motion path intact, and with my beloved value graph.

[Steve Bentley]

Thanks for the movie Angie.

I guess the problem I’ve always had with the speed graph is the magnitude and impossible-to-see detail of those spikes, especially if the motion isn’t so “nine-old-men” and needs to be a graceful sweep or a complex, multi-axis roll. If you could take that graph and blow it up to building-plan-sized sheets, what would those curves look like? Is detail being lost in their compressed shapes? The shoulder and toe of the curves, that are so important in the value graph, never seem to get enough screen area in the speed graph to see what’s actually going on. I feel like the shapes of the curves have been compressed to accommodate the small graph area and when I use the handles the peaks can really get away from you.
Because the graph is showing speed (distance over time), but the graph itself is on a timeline, is that graph actually showing “acceleration”? (speed over time)

Personally I don’t see a difference in final product between the two methods – in the hands of experienced people (you can get there either way – please correct me if I’m wrong here because I’d like to know what I’m missing out on). But I do see some horrendous easing in projects using the speed graph. 9 times out of 10 when there is an easing issue, the user has gone the speed graph route. Granted this is magnified by lack of user experience and perhaps with the value graph being simpler to grasp, less experienced folk can more easily produce better work.

For me its more a visual feedback issue. I can push one of those peaks farther than you have without seeing much change in the animation, conversely I can change one of the peaks by the smallest amount and it becomes a huge change in the animation. And what’s happening in the instantaneous rise to the peak (before the gradual falloff) that doesn’t seem to anything? It always feels like I’m mousing with boxing gloves on when using the speed graph, AND I can never make the graph editor big enough to really have the control I need.
I remember when AE finally included a value graph and we left behind the timeline based speed graph that appeared between keyframes; everyone I know let out an audible, “Finally!”. To be fair, that old speed graph was made even worse by the too-narrow band of graph area in the timeline.

I find it interesting that we all started with a that timeline speed graph way back when and yet once the value graph came along, everyone (at least everyone I know) went running to the value graph instead of what they knew already. Change is hard for most folks, so why did they change? Why did they switch to something that was new to them? Why haven’t other time-based programs adopted this method (Maya, C4D, etc.)

Admittedly there is no one right way to do this and its a potato vs potata thing.

But if there is a feature of the speed graph that is not available in the value graph, I’m all ears. Or if there is a situation where you just can’t get there with the value graph, I’d like to discover that too. Where does the speed graph make more sense. Or will the answer be, “always!”

I’m not trying to be argumentative, I truly want to understand the draw of this method for you. It works for you and you speak so passionately about it, there must be something to it that I’m missing.

Regarding my past post on the subject and your questions. I was referring to layers, expressions and plug ins referencing the value coordinates only. There is only one plug in I know that references speed, everything else, including the layer’s attributes are indicated by their values and the timeline creates the temporal aspect. So when doing expressions and plugs ins etc where you have to use coordinates (and never speed) to plan your layers’ animation, why not stay in that same system when you start adjusting (OMG, I’m going to say it) their speed?
Did I just lose the argument there?????

[Steve Bentley]

Or put them all in a precomp as is, and just rotate the one precomp in the outercomp.

[Steve Bentley]

I just completed a post for the very same issue. I think it has to do with the anchor point being outside some camera clipping limit. In mine I can always see them in the camera views but I can’t see the widget in the ortho views if the anchor point is shoved too far into depth (either way).
The responses have not been helpful so far.