Here goes my first experiment 🙂
I have been doing a series of test projects in Formation with Octane Render, a GPU based render, to check the possibilities of using it in production.
One of these little projects was a comparison between a real image of a Tequila bottle on a table took in the studio, and a series of renders of the cg bottle version using Octane Render and Mental Ray under similar lighting conditions.
The workstation I used was an Intel Xeon CPU W3570 3.2 Ghz with 8 cores, and the graphics card a GeForce GTS 240 with 96 cuda cores.
This is not a fair comparison, was made mainly to learn the work flow and possibilities to use in real projects.
For a very affordable budget you can get a geforce-gtx-760 with 1152 cuda cores that will reduce massively the render times.
What most impressed me was the ease and speed of getting a “cool photorealistic look” in the render. Right now I’m looking forward to continue with more tests and waiting get the right project to use it in production.
The starting point: some pictures with a Canon 550D.
HDR image for the lighting. Canon 550D with a Sigma 8 mm lens.
3D scene setup in Maya: the main light source was the HDR picture. For Octane that was the only light source. For MR I had added a directional light to create the shadow and emit photons for caustics.
This is the result.
Octane Render. 5 hours. Trace 32. 15000 samples. Path Trace.
Mental Ray. 6.5 minutes. 128 reflection and refraction. Caustics.
Mental Ray. 6.09 minutes. 128 reflection and refraction. No Caustics.
Mental Ray. 4.42 minutes. 128 reflection and refraction. No Caustics.
Mental Ray. 5.15 minutes. 128 reflection and refraction. Caustics.
Octane Render. 3 hours. Trace 64. 15000 samples. Path Trace
Here a comparison of the renders side to side. From left to right: real picture, Octane render, Mental Ray with caustics, Mental ray no caustics.
Other shots from Octane Render. They were made with a low number of samples so we got a lot of noise.