[Steve Bentley]

First the important stuff: Hi Adam, how are you doing?

John, are you perhaps thinking about the UV map that is used to warp the image in the 2D projection system?

Whether the texture map is baked or not is probably irrelevant to the projection since you have to render the footage anyway and baked or not that will turn out the same.
Can you describe the project a little more (or is there another thread that I can go to where you have done that?)

And thinking about it another way – are you actually rendering footage that will be projected or are you just generating the UV’s for the projector warp?

[Steve Bentley]

Depending on your alpha (straight or premult) that edge may or may not show in your picture viewer inside C4D, but may not show in your final comp.
Straight alphas have to generate stuff outside the matte (and that may be what you are seeing) so that the alpha can cut the image off clean and you don’t get what amounts to a double matte line – one from the content edge and one from the alpha edge. This extra stuff won’t show in the final comp as the alpha cuts it off.

What format of file are you rendering to? What depth?

What color is the object you are using as a catcher? Are you using the built in catcher? or the after market one that we used to use before C4D realized we needed one built in? (R17 I think).

[Steve Bentley]

Thanks for that Andrew – you’re a star!
Now why wouldn’t Nvidia tell us about that tool? Do you know how many hours we’ve spent on the phone with them? Perhaps they weren’t happy that we used to do Ruby animations for ATI.
Smooth as silk now!

Cheers

[Steve Bentley]

3D is tricky when using a non ortho view – your brain can convince you that the box you are moving around is set on a plane when in fact it’s wayyyy off.

Depending on the age of your setup and what you were using before, there can be two exporters in AE for use with C4D. On our system they are labeled: CINEMA 4D exporter and Maxon Cinema 4D exporter. One is from Cineware and one is from an older version of what came with C4Dlite. I prefer the first one.

Keep in mind that (depending on what you are doing) the Y in C4D works opposite to AE (higher numbers push things lower in the AE world). And that AE’s world origin in top left of frame and C4D’s is in the middle of the world.

Andrew is right – lets have a look at a simplified project (the way it is looks simple so just upload that – just the AE file should do it) and/or some ortho views.

[Steve Bentley]

Have you upgraded to AE2018 >15.00? Adobe has dropped support for some of the older Codecs (especially many QT codecs) so it may be that your source footage isn’t compatible any more.
Reverting to AE2018 15.00 will still allow you to use these older codecs – but you do have to put up with a few bugs that were fixed in 15.1.

[Steve Bentley]

It could also be a gfx card issue. We’ve got an AE setup with a GTX 970 and can’t find a way to make it playback smooth in AE even though it plays the rendered QT just fine on the same screen.

[Steve Bentley]

You could also just cheat it.
Try using a combination of bezier warp and displacement mapping and some precomposing.
Set up a bezier warp (depending on the shot a mesh warp can work too or the reshape effect – what you are trying to do is simulate the big curves the face that would distort any imagery projected on it). The bezier warp would be applied to the footage you want projected, but the beziers would be shaped to match the face. You can animate this warper but you might also need to track it. The easiest way is to track the mans face, then precompose the warped projected footage and have that precomp linked to the track of the face. The trick is to move the nested footage in the opposite direction of the track (use expressions to do this). That way the footage wont move with the mans head – the head moves within the projected beam – but the warp will. You may need to open two comp windows so you can animate the warp in the precomp while the track keeps it moving to match the face in the outer comp (it’s easier than jumping back and forth between comps).
You can also use some displacement to help sell the illusion too. Use the brightness of the face (you may have to crush the levels a bit to get enough contrast) to shift the precomped warp around a bit so that subtle distortions are shifted on every frame.
Projected imagery on an uneven surface can be so wacky that you can get something that looks right even though it might not be optically/mathematically correct.
You may need to roto the head so that the projected footage is masked and doesn’t extend beyond the face. You can also use this roto as a blurred element to help smear the projected footage at the edges (use another displacement map or a radial blur set to zoom and the center of the zoom linked to the track point of the head). If you had shot this, the projected footage would have hit the edge of the face and the imagery would have distorted into a smear – 100 pixels might have been on the front-facing cheek, but it might only take 5 pixels to span the side-facing cheek to ear distance.

[Steve Bentley]

If the statue element is moving you will need to track that to get a point you can “glue” to. If the face you want to map onto the statue is moving you may need to stabilize that so that you have a static point to glue to the statue’s static point.
The bigger issue is any camera angle change. If the camera is moving around the statue then the shapes will change – sides of noses will be visible at different times, eyes will disappear behind bridge of nose etc. Do you have some sample movies for us to see of both elements?

If its a 3D track you are after (the camera’s angle to the statue changes) then you will need to match move the statue footage with a 3D tracker. This is all very doable inside AE but the hard part is generating the matching face element with the same 3D move – not hard at all if the face is a 3D generated face, but much harder if its a real human head that is shot (think: motion control shot)

[Steve Bentley]

I want to check what you are saying in your post: It sounds like you want to use the png as a shadow and place it over your solid, but then you say you put the solid on top and use multiply. So something’s not clear.

Do you want the solid to act as the color of the shadow and then use the shadow element’s alpha act as a window to pass the solid through? If so the solid should be above the main element and below the shadow element with the alpha. The solid’s blend mode can be multiply and have a track matte passing the solid through the matte of the shadow element.
Or
If you want the shadow element to be the shadow on top of the solid, then the shadow element should have the multiply blend mode and not the solid.
Keep in mind that some blend modes don’t work correctly when there is nothing below them. So if you have an alpha’d element that is using a screen or multiply blend mode, and there are no layers below that, (even though you might have a light colored comp) it may not look right because there is no other layer to do the math on.

The other thing this might be is that you need two versions of the alpha element. One that is visible and one that is acting as a track matte.
You can have a layer act as a track matte AND still have that layer turned on with a blend mode other than normal. However keep in mind that the alpha will allow that layer to step on whatever is passing through the alpha because the layer with the alpha is on top.

[Steve Bentley]

Some of those orphans that stream off the orange and take on a life of their own are classic! Very nice.
Watch out for the reflections from ambient light (and you blocking the light) bouncing off the front glass. (we can see you!)
I saw the green screen versions – once you have your matte (the checkerboard) version you don’t ever need to make a green screen version. That will only degrade the edges more – the matte is the matte. just archive the footage with an alpha channel for use later.
Re salt: I might not have been clear enough – add the salt (and yes, more than what’s in your salt shaker) to some water and then fill the tank to about 1/3. Then put water with no salt the rest of the way up the tank. Its better if the salt water can be colder than the water above it and if you can rig a system where some thin polyetheline plastic can be put on top of the salt water, then the clean water is added on top and then slowly pull the plastic out – this helps the 2 waters not mix ( I sound like a Fremen). You can also try a little camphor in the upper water (pre mix just a pinch) this will lower the surface tension of the upper water and help prevent mixing too.
You can alter “gravity” by either cooling or warming your ink. Warm will tend to make it rise back up in the tank and cold will make it fall faster.
You can also squirt clean water with a second syringe and that can add a “front” for the ink to curl around. If you’ve ever see time lapse of clouds in the mountains, you know that there is no difference between liquid and air.