Hey Midnightrider,
The first thing that needs to be differentiated is the use of options/effects (2d and post) and volumetric (3D) creations. But, yes, it is not as simple as that alone.
As things might change, I went through the most possible scenarios, as I normally avoid that theme as much as possible in 3D directly. Glow is a hard to justify “quality” if DOF blur is there or not (if the effect is not exaggerated!), in that way it is an remarkable post effect—except when it is needed in reflections or refraction etc.
The Standard Rendering (vs the e.g., the Physical) has no option (Effects Tab) to calculate anything that is performed after the main rendering. As Patrick mentioned, this is a post effect. Glow by itself has more than this as limitation, for example objects behind “glass” will not render with glow. Think of it as information that needs to be directly and pure in the view of the camera. If that pixel has to be a glow effect than the post effect will use it, other wise it is gone. Gone means here - based on the depth of field it is not pure information anymore, it is mixed and lost the trigger function to get “glowed”. (This is perhaps not a real technical information, like one would expect in a “Technical Director Handbook”, but this is how I would describe it.
Alpha and Depth of field is supported since a while, but Alpha should be 100% transparent—otherwise the results might not satisfy.
The Z-Mapp or Z-Pass, is the pass that you get as Depth-path in Multi-pass rendering. It turns the distance among objects and camera into gray values. The scale of it is based here how the camera is set up. Based on these values, After Effects can to a certain degree render depth based effects. To a certain degree means here, each pixel is a value and not an picture element anymore, hence no use of Anti Aliasing here. It can have only one distance. Anti Aliasing would deliver a value between two objects, and the quality goes south based on that. A typical beginner mistake to try to get away with it. Objects have then a weird blurry outline around them. Not nice nor wanted.
If you think about the single value rule for these maps, any fog, glow or “whatever” volumetric information can’t (!) be included. It is not a given to have the luxury of a depth compositing information available with it. (It is technically possible to do this and standard for some advanced studios, but the file size explodes with this kind of data, so if you can’t effort a three digit number of MB per frame and pass, skip it. Perhaps later. Perhaps you can imagine how slightly refractive glass can work in this, like 20 objects in a row, like bottles or such, each changes the object behind it as well, each time a different blur-factor. As 3D application try to simulate reality, it is always to se in conjunction with render times. Practical light can do that a billion times perhaps (refracting, bouncing etc) in an 1/48th of a second (standard film/movie exposure time for example. To get that 100% simulated no computer in an standard budget is able to do that). So, having said that, objects have a point in space and that becomes blurred. Nice and easy. Glow is a volumetric (at least it should be…, effect. Depth of field is a spatial effect. In that way the glow needs to be rendered and blurred differently with each “increment” of distance to the camera. Volumetric effects are not easy on the hardware, but to blur something blurry—with a lot information behind it (other objects, perhaps with glow again) a render-engine night mare -time wise. So, the given 2D simulations are there to make everyone as much as possible happy. But of course the blur of it should be different in the reflection of it, as the distance is different, etc, etc. There comes the artist into the game to decide how to “fake it”, and how to use compositing to support a sane render time.
Please keep in mind that any post effect in CINEMA 4D has no effect inside of the scene, hence—no reflection of such effects, no refraction nor shadows, etc.
Similar to that is the “Glow” Option (vs Object Glow) it works on “hot” values or Object Buffer, but as well here, there are more limitations than real advantages. One needs to know the needs the scene, the target etc to decide how useful that is.
So, end of story? There are some workarounds, but I haven’t seen anything that solves the problem and avoids producing new problems. So, if needed, you have to pay for it with render time, will say—>Physical Renderer, also here, the glow will not show in reflections, based on its timing in the render pipeline.
The other way is to take a copy of that object, and use the Mesh>Create Tools>SmoothShift, to create a slightly larger object, use a luminance channel on it and in the transparency a fresnel to gain soft borders. This needs to be an extra pass and “added” to the renderings later on. It works in reflections and refractions. BUT: Refractions and Reflections have no Depth of Field either, so that is perhaps as well not the way to go.
If you like to have it, the physical renderer takes the “Glow” option (RenderSettings) not the Object Glow or the Material Glow even through glass like objects.
It is a tricky work and needs attention to detail, will say project/case specific. I have no other answer here.
I’m happy to look into specify cases, if a scene file is provided.
All the best
Sassi