[Steve Bentley]

We’ve had the mesh get “lost” in C4d (similar to how in the past you couldn’t replace a image with one of the same name; C4d just wouldn’t refresh.
So often we’ll move the mesh files and then re point C4D to the RF files.

[Steve Bentley]

Thanks for having a look Cassius.
I though it might be a bit of the ol gimbal thing.
I wasn’t clear on what you meant by turn off the POI on export. First I didn’t know you could do that, and second I’m not sure it matters – the POI doesn’t come into C4D (if it did I think it would have to generate a target tag and object). But it could be adding the POI to the global rotation and that’s probably where my problem begins.

Oh wait a minute – do you mean use a single node cam instead of a 2 node cam? Ya that might be the ticket!

[Steve Bentley]

Its that decimal position. When you have edges that can fall perfectly on a single pixel no antialiasing is needed. But you are telling that soft square to be partly on one pixel and partly on another so it has to antialias. This happens with other shapes too, its just not as noticeable. If you have a really sharp image and you place it at .5 pixels instead of whole pixels it will get slightly softer.
Scaling a non vector shape will cause the same “issue” (its not really an issue its just antialiasing)
This can actually help some cases to reduce flicker by intentionally putting the image at a decimal pixel position.
Once things move of course all bets are off because unless you are careful the object will always be falling on a non perfect pixels.
Video is actually pretty soft and you only see how soft it is when you can get a vector shape to line up to perfect pixels as the square on the right has.

[Steve Bentley]

I think it needs a few noises of different scales. And then (you can see the long streak of the spec highlight) use an anisotropic specular highlight. Widen it out and soften it.
There is also some reflection there but its really really subtle and softened.

[Steve Bentley]

We did one of these and we searched and searched.
I think the issue is when you do time lapse you never know where the plant will end up, so it goes out of frame or pops at the last second to full open. Then there’s the fascination with the bloom and that’s all anyone ever focuses on, so the stem and base get left out (also because the time it takes to grow the stalk is soooo long compared to how long it takes the bud to open, you have to choose a different intervelometer time depending on which is more important).
So its rare to find blooming with the full plant in view. Even rarer to find it on blue screen (leaves are green after all)
In the end we 3D animated a bunch of master plants and rendered them from different angles and then used those to populate the 3D track of the 2D footage.

[Steve Bentley]

Are you interpreting the footage to preserve the RGB? (depth channel only) And are you using at least a 16bit comp and 16bit rendered elements? This give the depth channel so much more head room and you can crush and expand your depth channel to stretch that focus area. You can use the depthBuffer/curves setting in LensCare or pre curve your footage in a precomp before lens care sees it to expand the area of focus.

Can you do a Position pass in Maya? These tend to work better than depth passes because you can alter the scale of the scene for that pass, thereby expanding what is in focus.

You can also use a shorter lens and put the camera closer – shorter lenses have longer focus areas than longer lenses.
Or you can use a higher f stop in your camera settings to increase what is in focus.

These last three are really just changing the falloff of the depth gradient, which you can do yourself in AE.

[Steve Bentley]

It doesn’t look like there’s a lot of blue sky there to key (but I didn’t bother to see how much was in the Raw file to pull out – since the upload was compressed). So you might be better to do a roto of your “skyline” since its a locked off shot. Then put the sky in you want and crush the levels of everything else down. The tree is an issue, so you will have to crush the levels to pull a matte and combine that with your roto. Keep in mind as K mentioned above, those objects tend to be darker than the sky so forget the matte and you could crush the levels just a little and then multiply/blend the tree etc right over the sky with no matte at all.

Notice that parking sign. Unless you have a car head light on a sign at night, its usually in silhouette against the glow of the sky – and all skys glow unless you are waaaaaay outside of town. So you might not need a pulled matte at all.

Cinema tends to use a blue cast to trick you into thinking its night (the opening scenes of Jaws are day for night and blue blue blue). So you can use sodium vapour pools color blended into the shot over a general blue cast and this gives a classic warm/blue coloring we expect in night shoots.
If you really want to get tricky, they tend to water the streets down for night shots to get interesting reflections. Again this is something we have come to expect in film so the more of those kinds of things you can add the more you fool the viewer into accepting the old tropes.

The white objects are an issue as they are going to get muddy as you try and darken them enough. Make a roto and use the roto as a multiplied element for a silhouette of those objects. There is very little detail in night time, er, details.

Shadows at night are actually more contrasty than the ones in the daytime (the blue daytime sky adds some fill to lighten them up) So don’t be afraid of contrast as you darken down.

You can also add some noise over all. CCD’s of todays camera get pretty noisy in the dark and high ISO films used for night shooting are the same (Amadeus used only practical lighting on film – the opening scenes are very grainy) and we’ve come to expect it (just like the totally unrealistic blue cast from the moon). The noise will also help tie the fake sky into the real footage.
Adding some offscreen flares (calling Mr. Abrams, Mr. JJ Abrams…) can create a fog that will also tie different elements together. It also helps even out the black levels which can really scream “fake” in Day for Night.

[Steve Bentley]

You didn’t mention which ram preview you are doing – the spacebar one or the zero key one.

Check the preview pallet that the loop button is on or off (your choice) and you might try hitting the reset button (shown in jpg) as this will write a new prefs file.
If you are using the new AE2018, try purging all your caches as this seems to fix many weird new “features”.

[Steve Bentley]

You can have multiple valid tex folders (you can find the list in your prefs) as long as the textures from one project are in a valid tex folder, then when you import a c4d file into a new one all should be good.
C4D normally asks you if you want to copy the image you are applying to a material into one of the tex folders. If you say no, it’s somewhere on your drive and will get used for that project only – it can get “lost” when you open the project again. So its a good idea to set those tex folders up and say yes when C4d asks where you want to put it.
If all of that has been done, I’ve found when an image goes missing, if I go find it for one texture, save the file as something else and then open the file back up, C4d will often find the others. Its like that missing image give c4d a hint as to where to look.

[Steve Bentley]

Is there just a decimal missing there? If the OP had said “when I switch my comp to 594 fps…” I’d think 600 fps, but because of the 59.94 does he actually mean 60?