Gavin Greenwalt
Well-known member
- Joined
- Dec 29, 2006
- Messages
- 7,614
- Reaction score
- 0
- Points
- 36
- Age
- 40
Keep in mind that a digital efx-heavy movie will be quite expensive to produce entirely at resolutions higher than 6K.
Not as bad as you would think (not as bad as I thought it would be when I started running tests). Adaptive sampling means we're already rendering some shots at 32x sampling for a single pixel. If instead we reduced sampling to 4x and rendered at a higher resolution you would get improved resolution but only a little bit more aliasing. Most scenes are largely low detail. You wouldn't notice the large expanses of shading with lower sampling. In some regards rendering small is actually more render intensive at times because a single pixel pixel encompasses far more detail. We've started rendering some passes in 4k for 1080p just because it made things quicker and easier.
That though only applies to 3D side of things. Compositing would indeed be a pain. Hopefully GPUs and solid state drives will remove that impediment in the next 2-3 years. On the 3D side rendering is going to be sped up by 10-100x in the next 1-2 years. We're about to go through a very dramatic shift in VFX in a very short time frame. Much of which will probably be visible before the Medium Format Epics ship.
On the other end of the spectrum. I'm comping a 4k shot right now which will have a 1080p deliverable but I think target web. The 4k might be slow and clunky on the rendering side, but it makes keying and tracking much easier than an SD or HD source. I'm not even sure tracking would have been possible in SD, I've had to rely on small wrinkles in the greenscreen. I would personally say the extra render time is less than the extra artist time on working with a lower resolution source. And unlike artist time, render farms are cheap.
We just need to drop the price of REDRockets so that it's financially sensible to put it in more computers and rely less on RAIDs.