[Steve Bentley]

Noise makes any compression do more work – there are more colors to consider and you will (usually) end up with a larger file because there is more information.

FFMpeg will compress GIFs and there’s GIFsicle.

With the vector look, you could also set a custom pallet of fewer colors at a lower bit depth (we use an old copy of debabelizer to do this). If you pre-reduce the colors that are fed into the compression software, it has less to do and what comes out the other end is smaller in mb and more tailored to the colors that are there. (so like 30,000 colours or even 256 instead of the 16.7 million you get with 8bit)

As for your orange to green problem – its possible that orange is not in the gamut for a GIF and because there are no graduations or variations of that color it just gets changes in the LUT (although all the over to a green seems bizzare). Adding noise creates those variations (kind of like a dither would) and allows the LUT to approximate a color closer to the original even though the exact one might not exist in the GIF world.
The alternative to noise would be to shift that color slightly in the original until it compresses to something more reasonable.

[Steve Bentley]

For stills, lightwave’s full blown go-and-have-a-long-meal-render version of DOF has always been the most natural. It fell apart however with animation and there were visible artifacts that popped in and out.

AE’s camera blur is just a blur. It tries, but you don’t get any bokeh, but then you don’t get any bokeh really really fast.

While the Nuke plug in version seems nice at the outset, I’ve found lots of scenes where it creates an almost vector look to thinks – too sharp or something. The bokeh becomes disks far to easily. This is great for makeup ads but not much else.

Lens Care Seems to bridge the gap nicely between being realistic and fast enough. (is anything every truly fast enough?) And its also a personal preference issue. Some people like the look of a leica lens and some people like the look of big-glass pentax’s – so what is “realistic” to you?
And with the controls available in Lens Care and the custom iris (try a high contrast dandelion puff – it creates gorgeous bokeh!) you can really zero in on the ethereal.
Its not as fast as I would like (especially at higher bit depths and large frame sizes) but I find the wait worth it.

[Steve Bentley]

I’m not sure I understand what you are after. Putting an adjustment layer over any layer will affect all layers below. If you only want to kill the noise on your 3d render, put the noise killer on that layer alone.

I guess the first question is, why is there noise on your 3D render and/or why would you want to get rid of it?

What am i missing?

[Steve Bentley]

If you can upload the project we can take a look.
Keep in mind that scaling is exponential – as you “zoom in” it gets faster and faster – when you scale something 200% it gets 4 times bigger. Scale it 4 times and its 16 times bigger; whereas a camera moving at a constant rate will look, well, constant.
Also with a lens involved, as objects approach the edge of the frame they tend to warp a little as they “round” the lens. The effect is higher in the horizontal rather than the vertical due to the side edges of a landscape shaped frame being closer to the edge of a round lens. You can use the distort/optics compensation effect to simulate this – it really adds a nice true camera feel.

[Steve Bentley]

Well if you mean the best way vs an AE way, use hardware. Some of the Farrugia (sp?) boxes are spectacular. We have one here that puts out HD from SDI and the difference is indistinguishable (even flipping back and forth between sources)

[Steve Bentley]

Hey Jim,
You’re selling yourself short. You’re always coming up with great solutions.
I’ll have to give the plugin a go. The going back and forth in and out of C4d and RF can get tedious and time consuming. We’ve done a bunch of fruit but we’re usually throwing coke bottles through sheets of liquid.

I’m also curious how fast the sim processes with the plug in or what the upper limit is of particles. With C4D already clogging the ram I’m surprised it can do the job. I’ve actually burned out a few machines running sims for days in the standalone app.

[Steve Bentley]

So I take it Jim, that you are using the RF plug in within C4d and you can do your sims within C4d correct? In that case, I’ve been talking out of my hat. We use RF, and I assumed the OP was too, using RF as a standalone program. (we do it this way so we can also use RF with houdini etc as well).

If the c4d sim plugin works with dynamics that might be a good reason to use the plug in within C4D. What we normally have to do is bring a stunt object into standalone RF and let it bump into the fluid there in order to sim the reacting mesh before we import the final mesh (collisions and all) into C4d.
On the downside, I don’t think some of the sheeting algorithms are available in the plug in version.

[Steve Bentley]

You might have to join the points of the two spline’s end points to make them one big spline (which you can only do once you have connected them – select the end points of each and hit “join” from the mesh/spline menu).
I’m not sure what the two helixes would look like or if your’s meet at both ends. If the do, you also have to reverse the direction of one of the splines so that the direction keeps going same way as you go around the circuit so you can make them one big spline.
Is that what you are looking for?

[Steve Bentley]

Paint reveal might be easier – because the legs are going to intersect the trees etc only at specific points. And because the background doesn’t really change you won’t see any of the boiling you normally get with paint.

Mocha can certainly do this too and might even automate it a little for you. But there’s a learning curve if you’re not used to it.

[Steve Bentley]

First, each box would be a precomp and in the precomp would be a timeline of all the animations that box should do. Could you not then use expressions in the outer comp to sample the luminence at the spot a particular box sits at and then send the play head of that precomped box to the frame appropriate to that luminence level?