While I certainly agree with you regarding the ancillary functions of lighting, camera, mapping and material handling, there might be a way to ignore all of that.
You could possible just create an ambient occlusion lit scene and render it with the camera and light positioning of the current viewport. Some examples:
You could offer the option of overlaying with wireframe:
This would have the advantage of people able to 'visualize' their design, which is really important when "sketching" in 3D. Many times I've noticed a fillet is too small, or a detail too large ONLY after I've exported and rendered the MoI3D model.
Hi Chipp, yes I've thought pretty close along those same lines before that I would like to have a simple kind of rendering like that built in directly to MoI at some point here. Something that could give you a result without any lighting setup like you're describing. I'd probably at least give a try at writing the renderer myself though.
@You "...Something that could give you a result without any lighting setup like you're describing. I'd probably at least give a try at writing the renderer myself though..."
Sounds very interesting...and in perfect Moi's unique style!
- Point incidence gradient to create the cell diffuse look
- camera incidence gradient for the rim light (self illumination)
- multiply by an occlusion as self-illumination.
I'm sure Michael can do something pretty cool, current MoI viewport is allready super sexy.
I wonder why RenderMan is suggested, since we have Max's Cycles solution for a long time now?!
I could also imagine once Michael would implement Rendering Options in MoI people would ask then
for more rendering features, "stealing" Michaels time for implementing Modeling Tools for MoI... ;-)
FWIW, I like Michael's idea of just having an OpenGL shaded viewport he codes himself. That way he can add AO, lighting modes, etc based on the video card available (I suspect the OpenGL api takes care of most of this). That way we can have realtime views.
Hi Chipp, well the kind of renderer that I was referring to would likely be an offline type renderer where it goes off and crunches for a minute or something, not a realtime viewport display one.
That generally makes it easier to focus on quality instead of totally on speed at the expense of quality. So I'll be able to do things like generate a custom mesh for the render that's fitted to the render image screen space so you won't see any polygons. There are some pieces in place to do some of that now actually, the PDF/AI exporter in V3 does that for the shaded background image.
The OpenGL API doesn't is actually pretty low level, it doesn't really take care of any of those sorts of things for you it involves a whole bunch of custom shader programs and work unfortunately.
That's not to say I wouldn't want an integrated renderer to be pretty fast, but realtime is a different thing yet than that.
that is a very nice looking rendering. Do I understand correctly that it was done in MoI? At least you show it in the MoI viewport.
Unfortunately I don't understand the following:
Here's a little shader test, using no lights.
- Point incidence gradient to create the cell diffuse look
- camera incidence gradient for the rim light (self illumination)
- multiply by an occlusion as self-illumination.
Sorry for the confusion, it's actually a 'fake' screen.
It's rendered in Modo, using easy (fast) computing method : only gradients and an ambient occlusion. (no raytrace shadow, no GI, etc)
NPR.jpg
