The original learning contract was constructed around experiments related to the quality of lighting an environment. It was built to produce work that would look amazing in theory despite the original thesis point was based on economy and average consumer products. I tried in the past to achieve an economic rig by producing a quality rig and stripping down. It might make more sense to do the opposite to see what I get. In order to understand the economy side of environmental lighting I would try and produce a rig that could compete with physical sun and sky.
The aim for this week was to try and produce a light rig that would be sufficient to rival the mental ray physical sun and sky. The process would have been solely base on starting on getting an economic render than bringing up the quality. I would also look into how shaders could impact this.
Firstly I started off with a base render of nothing but physical sun and sky and recorded a time of approximately 2 minutes. The renders were all done on the same QANTM iMac hardware.
Fig. 1, Physical Sun and Sky render
Here is how the following lighting rigs were then constructed then timed:
Fig. 2, Week 8 result
- - Week 8 environment with all of its lights and bounced lights came in at approximately 8 minutes
- - Bounced lights were stripped were replaced with Final gather, this brought the render time to 11 minutes
- - Other unnecessary lights were stripped leaving only major lights (ie fil lights). Roughly 6 minutes
- - 3 point light system with target lights as fill lights. Still roughly at 5 minutes.
- Having the directional light on its own with final gather approximated the same time as physical sun and sky.
From here I also looked into shaders and the impact they could give. I went through a handful of shaders, including ciesky, portal light, photometric light etc. It was a bit hard to understand what I was doing as I didn’t fully know the potential of these particular shaders. After a while I was hit with an answer.
The reason all of the light rigs were taking so long is because I was using multiple Mayan lights which were apparently more expensive. For every light that is created, light information is multiplied and therefore having less is better.
Physical sun and sky uses only one directional light and therefore already trumps Mayan light methods.
My experiment with shaders seemed a much more sensible route to take. With my knowledge from previous weeks I was able to get some results with shaders but still got the same render time.
LEARNING TO FAIL
It hit me that if I was building a light rig that had a dependency on shaders that I wouldn’t have much customisation over traits light shadow colour. Therefore this method was no better than Physical sun and sky. Mental ray had a system that had been developed and engineered for years. The sky node and the sun node work too well together as well as other shaders in the mental ray network. I wouldn’t that my efforts had been foolish but I grew an understanding on physical sun and sky worked.
Overall I probably have not really travelled far but have gone through a long journey to understand the value or Maya/Mental ray light systems.