I wonder why many people create 3D Animation in 3D software then export passes into comp software but HF creates 3D Animation while compositing without creating passes?
The dedicated 3D software has animation and render features beyond Hitfilm's capacity. For example, dedicated 3D software can render a "depth pass," which is a greyscale pass based on object distance from camera. A depth pass can be used later to generate things like lens blur. Now, Hitfilm's inbuilt camera has built-in depth of field blur, so why would I want a depth pass? First, Hitfilm's DoF is a box blur, which is fast to calculate, but it's not actually "correct," since a lens's blur "shape" is determined by the shape of the aperture. So, instead, manipulation of the depth map and use of a lens blur effect generates better results. Secondly, 3D programs don't deal well with video elements or types of generated media Hitfilm does very well with. Third, Hitfilm can point-rig models, but doesn't support bones and IK, so 3D programs can animate in ways Hitfilm can't. Fourth, the new normal/bump maps and Cook-Torrance shader in Hitfilm greatly increases the power of the render engine, but there are other types of maps (Reflections, Transparency, etc) Hitfilm still lacks as well as other types of physical shaders or Illumination features (subsurface scattering) Hitfilm doesn't have. So there's a lot of things these dedicated 3D programs can do Hitfilm can't (and shouldn't).
That's fine, actually. Hitfilm strives to be a bit of all things--editing, compositing, VFX and 3D model animation are all different tasks. Hitfilm does pretty well at all, but the compositing/VFX part of Hitfilm's toolset is it's most powerful tools. Hitfilm is really a compositor that just happens to have editing and 3D animation built in. It doesn't excel at these tasks, but it's a very unique attempt at all in one software that let's you do a little of everything.
Another reason for different passes is more control over the final image. If I have specular highlights on a seperate pass, and I decide objects A and B are fine, but C is too shiny, I can mask out the specular on C. Doing it all in one pass instead means that to change highlights on C I have to tweak C's materials, and maybe scene lighting and rerender the whole thing out again! So there are advantages to comping passes (and one can do that--render passes--in Hitfilm if one wants).
Additionally, pre CGI VFX just required comping elements and passes, and some of the workflow carried over to digital. For example, in the famous Star Destroyer/Blockade Runner sequence in the first Star Wars, each ship would be multiple passes. The lights on each ship would be their own passes, for example. Even rendering in Hitfilm, one can get better results from layering things. Simon Jones did a tutorial on adding lights to 3D models, which basically requires multiple layers. The bottom pass is a lit and textured model. The lights are a copy of the model unilluminated, but textured in pure black, EXCEPT the lights, with the lighting pass comped in an ADD blend. This is better than just having the lights being an emissive color, since you can add glows to the lights separately from the base layer.
So there are advantages and disadvantages to both methods, and doing an "all in one" render vs a "multi-pass" render depends on the shot and individual artist.
Three examples from my own work. This first one is a "one-pass" render--the models are simple, there's nothing going on with lighting that's tricky... This was a good choice for a single render.
This next shot is a "one-pass" render, but there are several copies of each model: the base models are a single layer, window lights are another, warp engine glows are a third, impulse engine glows a fourth and blinking running lights are a fifth. Look near the end of the shot between the Klingon shields collapsing and the Klingon exploding and you'll see the Klingon's engines, rear torpedo bay and window lights flickering and dimming. To get those elements to ficker independently I needed separate model layers, but I could still put this shot together in one comp shot.
Now, this third shot is very much in-progress (my new computer can show up in the mail, thanks!), so there's a lot to fix and adjust. However, the background hallway is a Camera Projection using 15 screens and 22 lights. The Daleks have a different lighting rig (the Daleks didn't need all the lights the hallway did), and are their own pass. The Dalek's cast shadows are a different pass. So I can adjust their opacity separately. The weapons and explosion are seperate passes (and the explosion needs work).
Now... Here's the tricky bit: the viewscreens on the hallway walls are their own pass. Look at the front-right screen as the second Dalek passes and you'll see that screen is showing the main shot with video feedback... This is why the shot required multiple passes... So that I can render the shot and paste the render into the wallscreen layers, re-render. Repaste into the wallscreen. Re-render. Repaste. That way I can render out those required passes without having to re-render everything over and over and over. Same with the explosion. I can adjust that and change that and tweak that without having to render everything else! Because a 15-screen projection rig with 20+ lights renders really really slowly! For this particular shot, multiple render passes and compositing isn't just the smartest way, it's the ONLY way!
Wow tks so much Triem!
Does Hitfilm 4 has render-passes feature?
Not in the sense of automatic exports. You kinda have to manually set it up. I cannot be more specific today, but I hope to get a tutorial out soon that discusses this.