In product photography for the web, one works with two resolutions: 1) Camera Resolution, which is fixed, generally around 20Mp for a DSLR and 2) Screen Resolution which is generally around 100dpi for a desktop LCD monitor.
Whether you are photographing a blank sheet of copy paper or a roomful of paisley woolen scarves, the image size from the camera is always the same: around 20Mp(1). Visualize renders, however, are a different story, and my problem is that my "image size" increases linearly with image complexity. In the example below I've had to create a reasonable good looking stainless steel M6x1 Socket Cap Screw:
It's an improvement on the threadless socket cap "plugs" that you often see.
When I first imported this "model", I had to "Retain Structure" so that I could treat different "Parts" of the screw with different appearances, of which there are only three. As a result, this screw has no less than 361,538 polygons and takes a few minutes to hybrid render 1500 passes with my GTX1080ti card. The problem is that my final image is a repair kit that includes about 100 of these screws, detailed below:
If my polygons weren't out of control with only one screw, they are now. With nearly 43.4 million polygons, my rendering has seized up to less than a single pass per second and takes a couple of hours to render something useful. I understand that 43.4 million is somewhere near 100 screws X 361,538 polygons per screw....but much of this is useless information to my final image; it's cool to have the detail there for when I zoom in but the web graphic is static and neither rotates or zooms. Is there a way to control "complexity" prior to a render so that I don't waste time rendering "useless", sub-millimeter detail? Can I force a lower polygon count?
Final image, actual size, 600px x 600px, 159kb.....can barely even see those threads!
(1) About 20Mp for a crop sensor, or APSC, camera shooting RAW