A month long composition study of historic art. Each Image was done in 30 minutes, the left is the original and the right is the rough reproduction. As a challenge these images were done without the use of measurement tools, layers, masks or other special aids.
Recently felt inspired to do some sort of fantasy/scifi styled environment with moody lighting so the last day or two have been coming up with this lantern concept. Its taken inspiration from things like wooden African masks, Aztec architecture as well as eastern Buddhist culture. The polished golden copper type surface seen in a number of Buddhist statues have a historical yet in some way even futuristic appeal. Its no surprise its a favourite material for Steampunk and other technology focused genera as well. As for masks, well they are just cool.
A design I have remembered since I was a kid were mask cameras from the PC game Thief II The Metal Age.
Ya ok, it pretty old and doesn’t look very nice by today’s standards but its a interesting concept. There were even solid copper versions that were mounted on the walls to alert the guards. With these things as a starting bases for ideas I think it will result in some pretty neat stuff.
This week I will be going through a Gnomon Workshop done by Nate Stephens that is focused on environment sculpting. Really looking forward to it. Usually there isn’t much that’s new however the way I see it a small investment even for just a single useful tip can save an incredible amount of time and cause an overall improvement in quality for years to come.
Yesterday I got sidetracked while blocking some environments by Cycles volumetric light scatter material. It was just last month that a new workstation came in allowing for 12 render threads so I wanted to test the material for practicality in rendering a video. Short answer… nope. Reading through the render engines documentation and spending nearly a whole day changing settings and preview rendering resulted disappointing. It seems even in the latest version noise is still far too big of a problem. Setting volumetric to 8192 samples with various speed enhancements with light paths, branched tracing and algorithm types still gave mediocre results that took all night to render. And we are talking a Linux machine built for rendering.
Thinking about this problem we should wonder why this information cant be baked into a voxel grid. It makes me wish I was more knowledgeable on light physics, there must be some way that could work.
Apart from that I also came on this recently. Have never been a fan of Cycles rough glossy shader but never knew it was this bad. Discovering that Blenders “Linear” shading isn’t really linear at all will definitely change the way I handle shading network from now on. According to the article putting your gloss info through a math node set to “Power” with a value of 2.2 helps this problem somewhat.