Last night I decided I wanted to try creating depth of field in mental ray with a camera than to do it with the Photoshop filter in post. I’ll admit, the results are stunning.. however the length of time to reach the final product is far too long. Yes my computer I used at work is a decent (though not entirely meant for) CAD workstation, it took 2 hours, 40 minutes for a draft 1280×720 image of simple spheres. So I tried again overnight but bumped the settings to medium-high quality at 1920×1080 with a variety of materials such as car paint, wet concrete, concrete curbs, glass, frosted glass, asphalt, matte plastic, satined chrome, and rough metal. It took 14 hours to get through 20% of the image. The third image is a draft version of the complex render. It finished in 2 hours but I needed to set the quality to sub-par and the resolution to 960×540.
Unacceptable for this environment but a definite possibility if a client wants a single great image and I have more than a single night to render – say a whole weekend. I do however plan on using this technique on personal renders (if I ever finish them 100%). The tutorial I used is at the end of the post.