What do I get if I move from DAZ Studio 4.12.1.117 (approx. midsummer 2020), to the very latest version 4.14.0.8.. (just released yesterday)? Here are the things I noticed in the Changelogs…
* Various Quicktime (3 to 4) upgrades, and dForce and iRay updates. DAZ Studio now runs iRay 2020.1.1.
Does this speed up rendering, as has happened in the recent past? No, it seems not. The official iRay devs’ blog talks of “lots of RTX related stability improvements” for iRay 2020.0.2 (May 2020), “many many fixes (features, general stability and memory usage)” for 2020.1.0 final (mid August 2020), and the most recent 2020.1.1 has no details mentioned on the blog. One imagines that if there was a speed boost for 2020.1.1, it would have been mentioned there. Thus it seems to be stability and memory improvements for iRay, rather than speed boosts. However, 2020.1.1 does support the new NVIDIA 30-series cards, so if you’re one of the lucky few who have one then yes… you get faster rendering in iRay.
* iRay render settings now include Environment > Matte Fog > Matte Fog parameters.
This appears to render the scene with a very uniform white haze, the application of which can greatly speed up render times (according to forum tests). I imagine the use-case here may be that three renders of the same scene/camera are made, and the two foggy renders are then masked and feathered and then blended in Photoshop. One would thus get a ‘ground fog’ layer and a ‘distant depth-haze’ backdrop layer which blended well with the straightforward scene, but with a relatively low overhead in additional render-time compared to using volumetrics? Just my guess.
* “Basic OpenGL” is now labelled “Viewport”, and “Intermediate OpenGL” is now labelled “Multi-pass OpenGL”. DAZ Studio now requires OpenGL 4.1 as a minimum.
A year ago I updated my main desktop from OpenGL 4.2 to 4.5, and the Xeon workstation runs 4.4, so that’s fine.
* There’s also the new Filament in the latest DAZ Studio, which is Google’s Android/mobile-friendly open-source real-time renderer.
At present 4.14.x is Windows 64-bit only (not compatible with the latest Mac OS, Big Sir) and obviously not yet a suitable replacement for OpenGL, though they’re promoting Filament for “faster viewport” rendering. But… what was so slow about the instant real-time OpenGL? Nothing. There are a few people who are occasionally heard implying that OpenGL takes a while to “render” for them, but I’ve never known what they’re talking about. Even saving out an 8k render takes only a second. It’s instant, in real-time, and it always has been. And anyway Filament still doesn’t look anything like your final iRay render, so it’s not “what you see is what you get” like iClone or Blender’s Eevee. The test renders I’ve seen look terrible. Filament is something to ignore and come back to in a few years, I’d say, when all the wrinkles have been ironed out and it’s fully cooked. But by then we may all have $600 real-time iRay ray-tracing cards, and a real-time WYSIWYG viewport.
* The specs say NVIDIA graphics-cards drivers are a ‘must’…
…but I’ve seen that before. They’re not, unless of course you have the relevant NVIDIA card, in which case… yes. But CPU for iRay is quite possible. I saw no mention in the Changelogs of any removal of iRay’s ability to use the CPUs for rendering. With the aid of the wonderful Scene Optimizer I can render a non-complex scene quite nicely via CPU on the non-NVIDIA desktop, and on the Xeon workstation I can even get a real-time iRay Viewport running on CPUs only.