Saturday, 20 September 2014

Owl: Final Comped Render (Still Image)

Tweaking and optimising the rendering has been a very time consuming process. I think it's important that the image itself holds up to scrutiny which means finding ways to incorporate the bells and whistles needed to make a convincing CG shot.  I do have a good recipe now, so my workflow is going to speed up dramatically.  While I'm not ready to show off animation just yet (because I'd rather it viewed within a final render, rather than disparate rendered elements) I am ready to demonstrate what a final render will look like. This is 95% close to the final image as it's going to get, so if you don't like it, you better stop reading this blog and start paying attention to some other animation! :)

The final composite is made up of a number of render layers, spread over a two different scenes. The scene in focus is a flat 2D plane with an animated sequence applied on top. The out of focus shot is another separate render. It's unconventional but it gets the job done. 

Generally, I keep specular and reflections within the same beauty pass, mostly because I've spent the time tweaking them within Maya that I never really have to adjust them anyway, and also because keeping track of reflections with lots of objects and layers in a scene can be tricky sometimes. In this case, I prefer ease of rendering. I do have a layer for shadow, as it's easier to mess around with the opacity and softness in post-prodcution. Otherwise, the beauty pass is very simple but with a few added effects. The only thing missing from this shot is Ambient Occlusion, which will be present in other scenes.

Motion Blur: With Mental Ray's Rasterizer, motion blur is a piece of cake to add. It's brilliant, with some limitations depending on the scene of course. You do have to sacrifice some image quality and I've found the edges aren't the cleanest (anti-aliasing) compared to scanline rendering, but it's the quickest and easiest way of adding motion blur. You can get the quality to a very high standard, but then you lose the benefits of the quick render. This is one of the things I've gone back and added to all the animated scenes which didn't have it in the first place. 

Depth of Field: DOF has been massive pain and has caused a few days worth of headaches. The problem is it's fundamental to the way I initially visualised the animation, so I can't sacrifice it. True DOF is easy to implement but very difficult to render out in a reasonable time. For a small time operation like myself, it's practically impossible to produce a 5 minute animation with high-quality DOF. 

Camera DOF AND 3D motion blur in an animation? You might as well aspire to build a giant render server on the Moon. Not that it being on the Moon would help, but that's certainly is how far out of reach it feels to be able even think about rendering an animation with both. Z-Depth isn't really a viable alternative in this instance either. The only way I've found of doing both (and someone please correct me if there is another way)  is to render every object requiring blurring onto a separate layer and apply lens blur in post-production, compensating the blur manually based on the objects position the scene. It's cumbersome and hard to keep track of, but it works and achieves roughly the same effect. I'm operating under the suggestion that my final 'camera' inside the vignette is set to infinity focus so I can get away with not blurring objects in the distance. 

Not a lot of posts recently, but I hope this post sheds some light on what I've been doing. I should be able to get a sneak peak of the animation uploaded soon.

EDIT - Some other stuff that will added: a film grain layer and a subtle telecine wobble applied to  the final comped animation.