Real world example I encountered was an eye of a small fish, with actually modeled transparent liquid layer on top of the actual eyeball, which rendered black once I turned on the transparency. However, if you’re rendering geometry surfaces that are VERY close to each other, and shading is not right, this might be the thing that can fix it. Despite this is not a hidden parameter, you rarely need to change the default value of 0.001. Vm_raybias (Raytracing bias) – raytracing bias is the shortest distance between surfaces Mantra will actually trace. Pixel samples and primary ray samples can not be overriden on per-object basis. Displacement and object visibility overrides are exposed in object parameters by default. Local overrides can be done for indirect samples, ray limits, ray thresholds and motion blur. Local overrides of ray samples can help firing more samples only where it’s needed. That can be useful in scenes, where some materials produce larger amount of noise than others. Rendering properties existing locally on objects will override global parameters on ROP node. In Mantra, many rendering parameters can be set per object (by adding them through Edit Rendering Parameters dialogue, or using material stylesheets. Selection of specular models as you see it in Classic Shader Core interface Per-object overrides Having said that, phong doesn’t always generate less noisy result, so I recommend doing some trying out to see which spec model suits best your particular requirements. With Classic shader, you can switch between specular models directly from the user interface, Principled shader has this option buried inside it’s node network – look for network box called base_reflection_bsdf. Phong model tends to render less plausible look (also has quite a bit different response to material Roughness parameter), in many cases thou, the difference is negligible, while benefits can be significant. However, switching to Phong can significantly reduce noise of indirect rays in some cases, especially with refractions.
By default, shaders are set to GGX specular model. Mantra shaders, both Principled shader and Classic shader, allow for rendering a bunch of different specular models. pic files currently support larger than display data windows (.tif files do, but don’t allow negative pixel offsets). Vm_overscan lets you render camera overscan.
Since Houdini 15 this was sort of substituted by Quality sliders on Mantra ROP Sampling tab. Useful if you’re getting significantly more render noise in secondary rays. Vm_decoupleindirect (Enable Indirect Sample Limits) – allows separate sampling of indirect rays. In low contrast lighting situations it may be worth actually lowering this value for diffuse/volume rays to speed up calculations. For optimization sake consider raising importance levels only for rays where it really makes a visual difference (like mirror reflections). Note altering this value can significantly slow down rendering if too high. Ever had a 16k environment map that seemed to render lowres in reflections? Time to bump up light importance levels. Adjusting it basically changes the internal resolution of a texture applied to a light source. Vm_importancelevels (Light Importance Levels) – this is actually a light object parameter, not a ROP one. It usually happens once you try to render a larger amount of texture data or displacements. When render log says you hit the Sample Data Cache size you definitely want to bump this number up to speed up the rendering. Factory setting for this parameter is far too low and can impact rendering time badly! I recommend defaulting this one to 2048 instead of 512. Vm_samplecachesize (Sample Data Cache Size) – allows larger cache for rendering.
0 kommentar(er)
