Would this help? https://www.fxguide.com/fxfeatured/face-it-will-gemini-man/ [www.fxguide.com]

Oh yes, very impressive work right there. I feel like that's a whole other type of development though. For now, I'll put it on the maybe pile

What do you think of adding something like this to your scales workflow?

yes, thats definitely something in the works !

You mentioned issues rendering a large number of scales. Have you considered turning them into primitives? https://www.sidefx.com/docs/houdini/model/primitives.html [www.sidefx.com]

Hey Bennel,

what do you mean exactly ?

I mean turning them into lightweight geometry.

Right, I see, but that happens after, so that won't speed up the geo creation process

Oh. I hope you'll come up with something.

Thanks ! I'm looking intto opencl and using that to speed things up I already have the opencl speed gain with the smoothing part of the sculpt tool. If I can get my head wrapped around the whole opencl thing, I might be able to speed a lot of other things up, intersections, finding nearpoints, etc..

Performance isn't bad btw, it just drops to about 10-15 fps with a lot of scales. I'd like it to be as smooth as possible

How's it going with the opencl?

Hi Bennell,

haven't looked into yet lately, I'm doing some commissioned work at the moment

Would this paper help you in developing your HDA when you're back to working on it?

Perhaps, I'm a little out of it at the moment, Let's see when I get back into it

What about using subdivision surfaces for smaller scales and render them as scales using displacement at rendertime?

Maybe, yes, thing is, I do my rendering in Octane, so I'm not familiar with Mantra..

I read about rendering subdivision surfaces with displacement from the RepTile paper I gave you and a paper about procedural mosaic arrangement. It describes a vector shader used for geometric tiles. Would perturbing normals be possible for your HDA?

I have thought about shaders / Octane OSL because doing things like this on the GPU would give a lot of possibilities and speed improvements but it is a different approach entirely. So another one for the maybe pile

Yes yes yes, super cool. I had a moment of inspiration an decided to try a completely new approach. Instead of creating the geo and then finding the intersections ( which works but is too slow for my liking) I am now actually growing the scales which gives very nice and beautiful results and on top of that it is much faster !! Yaaaaaaaaay

Great! Can you show that?