When Lightwave 2018 was released with the new render engine, my first thought was: "why?".
Cycles could've been integrated instead while keeping the old (still good!) render engine around in Layout. Or AMD ProRender. Both have more developers and Cycles is a more mature render engine at any rate. Both have permissive licenses which allow for this.
The few lightwave devs left could have focused more energy on actually improving the software in key areas that are just screaming out for attention given to it. And they could have ridden the Cycles gravy train for free, as well have GPU rendering.
But no. Instead a whole new CPU renderer was developed with no hope for a GPU version in sight. A waste of time and resources in re-inventing the wheel. Creating a new battlefront while losing ground in other major fronts was a strategic blunder of epic proportions, in my opinion.
But, hey: water under the bridge. Hindsight is 20/20, right?
No point in repeating that I think, now they know how it has been perceived in terms of speed issues, and they can choose to do nothing about speed enhancements for the CPU, and do nothing to get a GPU version going...if that is how they choose do do, they will surely shoot themself in the foot..I honestly do not think it can survive without adressing those two issues.
They also have new technology to wrestle with in terms of performing slow, shape primitives, which are very nice if you want infinite detail on landscapes for instance (not restricted only to that) but It becomes unbearable to work with when you have to wait for minutes to have the shape primitive iterate to anything you actually can see materialize..if you are working with procedural textures, put that in relation to seconds when you have dense subpatch landscape with the same procedural texture..so new technology, but also implemented as being very slow to get decent feedback from.
Same with VDB fluids...nice technology and stuff that I can not do directly either in blender for some force stuff, but also lacking a lot...and very slow to use, and very hard to get decent OpenGL presentation.
Same with fiberfx, new primitive type..and the results may be very nice once rendered, but so darn slow I can not bear to use it at all ..and I do not spend any time on it anymore since cycles does it so much faster..both cpu and gpu for that matter.
So it was a dramatic change they did for 2018-19, and they need to survive adapt and overcome, and that they do not do by having the new technology performing as it does in it´s current state.
All the new tech is exciting sort of, VDB tools, more realism in materials and lighting, and in volumetrics, I think I prefer the volumetrics best if I have to choose from the new technology, even though I was a bit reluctant against it
in some cases, and desptite they screwed up perfectly good workflow for hv´s and some functions, it does deliver more realistic volumetrics and with some functions that was much harder to acheive with old hypervoxels..so they gained in some areas there and lost in some others.
I still haven´t evaluated the render speed differences when it comes to volumetric items with procedural textures such as clouds, comparing gpu cycles with lightwave´s volume item and cpu, I have some test scenes for cycles that I may need to check, thoughthe scenes has to be at the same scale for getting the proper comparison evaluation.
Then you got Octane volumetrics which I haven´t had the chance to try that much, VDB files didn´t work properly, though I need to work a bit with some of the stuff Lino Grandi showcased for volumetric clouds and octane in blender and compare the speed and ease of setting up with Lightwave´s native volumetrics.