Questions about iRay render speed

So I've found when rendering scenes that some shots can take a long time to begin rendering, but once they do they produce iterations very quickly. Other scenes can speed through the "prep" process very quickly but churn out render iterations very slowly. And of course there are scenes that are both quick to "prep" and quick to render, and scenes that are slow at both. I have a vague idea of what slows down both (textures, lighting, overall scene complexity) but I'm wondering if there is any information to be had on why this disparity exists between the time it takes to show the first render iteration and the rate at which additional iterations are produced. I've noticed even that in some scenes just changing the camera angle slightly can have a significant effect on the rendering speed.

Also as an add on question, I wonder if anyone knows why selecting "New scene" in the File menu takes a much longer time to complete than just deleting everything in a relatively simple scene.

Comments

  • Richard HaseltineRichard Haseltine Posts: 100,948

    Prep time is a matter of transferring the scene data to the render engine, so it depends on the amount of data. Render speed, however, while influenced by data (especially if there is more than the GPU can handle and it drops to CPU) is acutely influenced by how long a light path takes bouncing around the scene before achieving a final value for that iteration - a simple scene, in terms of data, can involve very long paths if there are a lot of reflective surfaces in an enclosed space. And of course there is also the issue, not intrinsically tied to either, of how many iterations are needed for a pixel to converge on a settled value.

  • Richard Haseltine said:

    Prep time is a matter of transferring the scene data to the render engine, so it depends on the amount of data. Render speed, however, while influenced by data (especially if there is more than the GPU can handle and it drops to CPU) is acutely influenced by how long a light path takes bouncing around the scene before achieving a final value for that iteration - a simple scene, in terms of data, can involve very long paths if there are a lot of reflective surfaces in an enclosed space. And of course there is also the issue, not intrinsically tied to either, of how many iterations are needed for a pixel to converge on a settled value.

    Thank you for that explanation 

Sign In or Register to comment.