Pod, if you'd had roughly read what I wrote, you'd know that this is only about the topic here. For bwrtz a very simple lighting may be enough, and with the Cheetah renderer (which I never used) I thought spots should reflect. I don't think this is about getting 60 h per frame down to 2 h, but more 5 minutes to 1. For bwrtz 'photorealistic' simple is not an option, so the tips here are only usable in such circumstances. He seems to look for tricks to cut even seconds from a frame which in an animation can amount to quiet some time. So, for example, dielectric may give better results, but a very simple cartoony glass material should be faster in this case. I actually don't believe that I would like the result, but each his own. As I said before, it's not important what I like or not but that bwrzt get's the results he is looking for.
For myself I don't always use hdr but mostly. And you can get photorealistic results with area lights etc.; only in Cheetah (as I wrote before) it's almost impossible without hdris to get a good result. So, while I try to optimize my scenes to a certain degree, quality is the most important thing for me. So I don't care much if a pbr takes a few minutes longer to render instead of another material if the result is not what I'm looking for. Like I already said quiet a few times, to a certain degree I stop optimizing and use planing instead, so I can render for a few hours without being unable to do my work.
I don't think Cheetah will ever get GPU acceleration. What's possible is full GPU-rendering with ProRender, which sadly, is still not developed far enough to be production ready. I don't know what AMD did in the last couple of years, but the software is still not very far advanced, quiet a lot of materials are a problem and so on. By the way, with very simple scenes as the topic of this thread, it can even be slower than cpu rendering (it takes more time to send the date to the gpu than rendering it out in such cases). The other downside is of course the simple fact that you'd need quiet good hardware. NVIDIA has simple so far more advanced in such things that AMD is simple no match. Even AMD's denoising isn't half as good (Intel denoising seems interesting but I don't know if that really has a future on MacOS).
> Network rendering
Network rendering would indeed be a good thing, but I don't know how big the demand for it actually is. Is it really worth it for Martin to open that can of worms? It's probably enough work to do it (programming render nodes and stuff), but even when it's done perfectly well, it still doesn't work for all people because of their network, the available (usually older) hardware that they can use as nodes and so on. I remember well when I tried it the first time in Lightwave and Vue, where it took me ages, to find what's wrong with my network to get it to function. But most people don't see it that way: They usually blame the developer. Back then, for me it wasn't worth it in the end. It was better to use the older computer for 'normal' work, like writing, accounting and so on, even some graphics (that could even be painting a texture), and using the top notch computer for heavy stuff only (that included 3d, some heavy graphics and games). I could have used the 3rd pc which I usually kept as a ready reserve system, but that would have gone to far.
Today, I have to admit, I do use it a lot. I'd be in a fix without it (and had to shell out big money to get top hardware).
But all this stuff about distributed and gpu accelerated rendering is a bit lost in this topic. Not because it doesn't help bwrtz in his task (I often digress myself) but because it's about something very, very specific that's actually of no interest for most Cheetah users. I don't think bwrtz uses much that wasn't available already in Cheetah 5. It should be quiet clear that I usually never would recommend what I did here because the resulting quality is only of use for unrealistic scenes.
Network and (e)gpu rendering on the other hand would be of interest for most of the people here. This stuff shouldn't be buried in this thread.