[Steve Bentley]

I’d worry less about the C4d Mapping. In the end you are putting 2d content onto a 3D surface. So you can certainly have a a floating 3D cube casting a shadow on the surface and it will look like it’s floating out from that surface but its the warper in MM that’s looking after correcting the projected image that would be distorted by the shapes on stage. Dragging a particle emitter left to right, MM will look after it undulating across the various surfaces. You can setup edges in your “2D” c4d comp or in AE that the particles can react to without going through the pain of building a true 3D representation of whats on stage.
But you can also build simple versions of the planes in C4d so that you can have particles bounce around. These don’t need to be perfect and can be done by eye. With the exception of the POV spot and edges lining up, 3D mapping is very forgiving. Shadows, contact points, lighting all reenforce the illusion so as long as you get close with these it usually sells. You really only need to be super accurate when you are mapping an image of the building you are mapping onto.

There are a number of plug ins that allow you to import OBJ files into AE and you can use those to help composition or (in some plug ins) texture them directy with sub comps (Video Copilot’s element 3D is brilliant at this).
You can also position planes in 3D space in AE and then sub comp them leaving all attributes in the outer comp so you have flat surface to compose on in the inner comp but can be angled and 3D in the outer. And a mask on this solid can be used to trim the shape to match your on stage geometry.
For positioning you can export nulls from c4d that come in as nulls to AE, as well as lights and cameras. If you animate these that data comes in as well. But be warned there is no rotation data you can play with in AE on lights or nulls. But you can bring in a camera and it comes in with everything and is fully rotateable – we use these all the time, not as cameras, but as nulls we can hook geometry to in AE and then play around just as if we’re in C4D. Just make sure to turn the cameras you are using as nulls off (the eyeball icon).

We’re probably starting to annoy the CC gods with length so here’s an email you can send to if you want to send designs etc and I can point you in the (hopefully) right directions. This can be brain melting stuff (should I predisort this? or is the distotion happeing in MM? or should I make it flat?) steveb@sharktankdigital.com

[Steve Bentley]

Actually you are using two very different splines located in two very different places to make the sweep and the extrusion. I can’t help get the feeling that you are trying to achieve an effect for which there is a simpler solution.
Can you let me in on the purpose of the sweep? What happens in the animation?

A couple of things here. First, you have a very complex larger spline with a uniform spacing set to 28. You might be better off setting that to adaptive and bringing the number down from 5(the default) down to 1 or 2. that will make sure all the detail of the spline is captured and so you don’t get flat spots on some of the curves. A sure sign of low quality logo is facets on the edges, especially if it goes by close to camera. This is too complex a shape for only 28 knots on the bezier.

While there is nothing wrong with sweeping the way you are, there are a couple of gotchas that might crop up given your geometry. A sweep that comes back around to meet itself can sometimes have a kink at the end point because of the style of spline or because of the handles on that spline. That end point must have the same angle of handle as the start point or the sweep will come charging around on the wrong angle and the two tangents won’t match. With such a big profile compared to the spline shape you will have to watch out for that. And for the same reasons (big profile) when that profile goes around corners on the spline the sweep will often kink because the turning radius needed is smaller than the width of the profile. Usually these kinks (pie wedge shaped errors) are only seen on the inside of the shape or the ends so you might be ok here. A smaller profile width or a smoother more rounded spline is the answer here.

Finally if the outside wall of the sweep is in the same spot as the outside wall of the extrude and they overlap you will get flashing as the two surfaces compete for “who’s on top”.

I’ve fixed the spline to be the right shape, but I’m not sure where the sweep should live. I simply copied the big shape, selected the points I wanted, inverted the selection and then hit delete. Currently the sweep suffering from all of the above. (always they way!) You can move the shape in our out by moving the spline Z and you can adjust the “larger or smaller” than the extrude shape by either scaling your profile or moving the profile’s origin point with the axis modification tool. You can do all these things while the sweep is active so you can judge the effect.

Because the new spline is not in the spot as the non matching spline (in the Z) you may need to move it in and out.

I have a feeling this larger spline was pathed from an image and not drawn as a spline. From a motion graphics point of view, you should be warned that the logo will look “noisy” due to all the ins and outs of the outline. When you take something that was probably a brush stroke or sketchy and apply a hard edged structure to it (the extrude) you magnify the noisyness of the shape. Its kind of like drawing smoke with a marker – you have whispy stuff that you have to put a hard line to. It won’t have the same feel.
The logo may end up looking ragged. It might be better to stylize the shape and smooth some of that out. If you are going to make it reflective those little jaggies may end up making distracting pops and flashes of light instead of a smooth lovely gloss.

11582_11581sweepissuefix.c4d.zip

[Steve Bentley]

I’m not sure who the culprit is but there is no “United Nations of Y is up” kind of thing. Plexus is like this too, one of the first things you do is invert the Y.
At least its as easy as a negative sign, packages like formZ consider Z up.
It also has to do with the fact that AE has it’s origin at top left so when you increase the y value you travel down the composition window not up as positive values do in cinema.

[Steve Bentley]

can you upload the file as we’ll take a look?
Also remember that the origin of the profile (even if it’s offset from center) is what rides right along the spline.

[Steve Bentley]

There’s only 6 rings there – why go to the trouble of a cloner? But if you must – you might need a mograph selection tag to hit that lone blue texture.

[Steve Bentley]

Are you using the same spline used for the extrusion? Usually its just an issue with which way your profile object is oriented. If you use a circle (for example) as a profile for the sweep, depending on which way your spline runs, the initial sweep can generate a deflated garden hose looking thing instead of a tube. If you change the circle to an XY orientation instead of the default XZ it usually fixes that right up. For non primitive profiles you can generate that outline in another view (left vs front for instance) or change the axis orientation
with the axis mod tool.
One other thing – if your extrude has rounded endcaps, that can change the overall size of the object and your sweep won’t line up. Just check the constrain button in the caps tab and that will bring the size back to line up with your spline.

[Steve Bentley]

Can you tell me or show me what you will be projecting on? Half the battle is deciding what approach to take. Only experience can aid in this area. Happy to help.
While the idea of building the projection surface in 3D is tantalizing and 3d printing even more so (always looking for new ways to use mine), it can lead you down a painful blind alley.
The UV map that you can use in the calibrator is brilliant but it is meant to be used inside of MM only using MM’s mapping of MM generated effects onto the mesh of that object that is stored in MM. Its a great thing, especially the shadow generator. But I know a lot of people have then gone and done a 3D render in C4d with their model and then tried to apply that to the object in MM after calib. It can be done but you have to export every frame as a UV wrapping map and you have to bake the textures. There are a few hurdles on the MM side as well. They may have updated this to meld the warper and the UV calib – its the next logical step.

With the 3D printed object (I assumed to use as a test) this does work but you can’t go right from 3D print to full scale without repeating the entire process and in some cases redoing your content. The problem is that, at full size, the width of the lens is usually insignificant and can be considered a single point and that’s how the math is done (in fact it needs to be a single point to make the math doable). But when your lens is about the size of the object you are projecting on, the matrix of the projector chip can “see” more faces than it would at full size because the lens is so wide compared to the model – it’s literally seeing around corners. So you can use it to test or present to client but you will have to remap once you get full size and probably tweak like crazy.

But while all of the above are cool and good for people without content creation chops or facilities (because you can do everything inside of Mad mapper – its just it will look like the next guy’s work who also worked exclusively inside mad mapper), I would recommend going the classic route.
Yes, build a model in 3d but not for what you think. Build it with the warping grid in mind. If you have geometry that has quads that you can overlay over the real scene those can help you align the warp grid to the faces and that will keep you from guessing as to what’s parallel and whats not.
Lets go back to my column example. If you try and map the MM quad mesh to a real marble column, you won’t have any reference as to how thin to make the quads as you round to the edges of the column (see image below). You can do this with math (its a simple cosin function) but I’m sure you had enough of that in highschool. So the model is not used to project content on to (at least not yet) but to help you know where to put your quads. The hard part (sometimes) is modeling in nice clean regular sized quads.

Or you could use a trick. Just model anyway you want any way you want. Then project a perfect grid (or the checkerboard) on to your model with a long lensed camera inside c4d using the projection mapping tag. (long lensed so the camera, now acting as projector, doesn’t add its own distortion) The camera can be centered on the model (squared up) or can be at your POV spot. Both are acceptable but it does produce different results – depending on the effect you are after in the end. Use this image of your model with checkeboard mapped onto it (ignoring the actual 3D mesh it was built from) as your mesh warping base in MM for alignment.
Its not as perfect as modelling in quads but you can always tweak afterward.

With the projector now mapped to the surface you can use a render of your model (or a photo taken from the POV of your building) as your ref in AE or Cinema to animate to. You can use the 3D model to generate shadows etc or silhouettes inside windows – think about it like a flat plane even though its geometry. Remember if you project on to your model it will get distorted twice – rule of thumb – what ever comes out of the c4d render thats what it will look like on the building or whatever your surface is. I prefer using the 3D model for the above mentioned things, enhancing the geometry of the building that is already there and making large lights swoop around that aren’t really there but the CG shadows seemingly generated by the building’s own architecture are very persuasive, but also using the surfaces on the buildings as planes and adding my own depth within these areas. For this content you can model in 3D but just use a flat plane as a canvas and then comp in this element to whereever you want it. Wizzing a glowing trail across the building face is all done in the warper – just wizz a trail around in AE within the boundaries of the building facade – MM will look after the perspective correction. Just think about depth as you do – so you could arc it toward camera in AE then shoot it back into where the doorway is – the combo of the warp done by mm and the depth in the AE render is almost mind bending in real life.

One other artsty tip that helps with faking perspective – light colors recede and dark colors come forward in the human brain (its because there is so much aerial diffusion at the horizon that it looks lighter, so we assume anything lighter must be farther away because we all know just how far that horizon is before you fall off the flat earth.

11578_column.jpg.zip

[Steve Bentley]

Whoops hit send too soon. No example needed but thanks for the offer. The problem with using the multi shader (and I worded that poorly about it being on the cloner only, you are quite right) is the sheer number of versions of the textures I would have to include and that would be a nightmare to wade through to assign. There also isn’t a good way for the effector falloff “shapes” to steer that except for a selection tag, but again I would have a line of tags as long as my arm.
I think I’ll head back into xpresso/coffee/python land to make this happen. “One texture to rule them all!” If you hear me burbling “my precious” you will know it didn’t go well.

[Steve Bentley]

You might still run into a few problems because the cylinder is so long compared to the width of the vessel. So it’s conveivable that one will get generated while its already sticking outside the walls. It’s unlikely, but possible.

I don’t know what this should look like but for what it is the divisions of the cyclinder seems high for this use. So you could bring it down to say 18 or 24 divisions instead of 36 before you poly them. But even with that reduction, the onscreen playback (without rendering) will only go so far before the system can’t keep up with all thats going. These should render really fast you so you can render a movie to check them.

[Steve Bentley]

Ok easy fix. When you are doing collision detection (or frankly anything particle related) you should always polygonalize your models to be emitted and surfaces. If you convert those cylinders to poly’s no more leaks. The system needs discreet geometry to do the math on and the parametric shapes are kinda fluid. You can also sometimes turn off render perfect but the cylinder doesn’t have one of those check boxes I don’t think.