With the rising costs of memory and GPUs expected next year, I believe it’s crucial for D5 Render to focus on performance optimization and FPS improvements.
I’ve noticed that separating 3D models into layers and turning off those not visible in a scene helps a little, but it’s still not enough. Additionally, using scatter or brush tools for natural models like trees remains heavy on performance, even though they are supposed to be lighter than placing each model individually or using low-poly assets.
The ideal solution would be to load only what the camera sees in the scene. This would significantly reduce GPU and memory usage while also improving render times. For example, in a recent project, a single image render took 2 hours—even after optimizing the viewport as much as possible. Imagine the impact this has on workflow efficiency.
Since D5 is based on Unreal Engine, implementing technologies such as Culling and Nanite should be a priority. Without these optimizations, upgrading hardware will become increasingly difficult, especially with AI driving demand and reducing availability of GPUs and memory in the market.