Comments on: 3D rendering as functional reactive programming http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming Inspirations & experiments, mainly about denotative/functional programming in Haskell Sat, 26 Sep 2020 21:06:12 +0000 hourly 1 http://wordpress.org/?v=4.1.17 By: conal http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-316 Tue, 20 Jan 2009 19:28:13 +0000 http://conal.net/blog/?p=75#comment-316 @newsham – Yes, that’s the idea: futures are opaque. Events are partially transparent. For fancier rendering, using transmission through 3D media (air, jello, etc), the (per-ray) model is analogous to (continuous-time) behaviors with integration.

]]> By: newsham http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-315 Tue, 20 Jan 2009 18:59:43 +0000 http://conal.net/blog/?p=75#comment-315 So (stretching a metaphor) the future occludes the more distant future in FRP?

]]> By: andreas http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-314 Tue, 20 Jan 2009 18:43:01 +0000 http://conal.net/blog/?p=75#comment-314 Hi conal. i am really excited about your interesting post. So i was thinking about a small example where you don’t need the color but just the depth value like ShadowMapping. Semantically something like

type ShadowMap = Image Depth
renderShadow :: Geometry -> ShadowMap

And a Light that uses the image (creating a dependency).

shadowPointLight :: Color -> R^3 -> ShadowMap -> Light

What are your thoughts on that? Could that also be described ?

]]> By: conal http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-313 Sat, 17 Jan 2009 17:25:39 +0000 http://conal.net/blog/?p=75#comment-313

In practice, points also need tangents and bitangents for certain types of lighting. What are your thoughts on that?

Thanks, Peter. I’d forgotten that normals aren’t enough for all lighting models. An easy solution is to throw away less information. In place of location and normal, we might have location and first derivative (full, i.e., both partial derivatives), or a full (infinite) derivative tower.

]]> By: Peter Verswyvelen http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-312 Sat, 17 Jan 2009 00:53:34 +0000 http://conal.net/blog/?p=75#comment-312 A bit off topic. You write:

“Each point on these surfaces has a location, a normal (perpendicular direction), and material properties (determining how light is reflected by and transmitted through the surface at the point).”

In practice, points also need tangents and bitangents for certain types of lighting. What are your thoughts on that? See e.g. http://www.terathon.com/code/tangent.html

]]> By: conal http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-311 Wed, 14 Jan 2009 05:23:31 +0000 http://conal.net/blog/?p=75#comment-311 Hi Artyom. Thanks for the feedback. Yes, I’m talking about Reactive as an approach to FRP that is built around the notion of future values. Also about FieldTrip, though I haven’t yet applied my new semantic ideas in the library.

About Reactive and real applications, I think it’s getting quite close. There’s been a persistent troubling bug that now appears to be rooted in a concurrency bug in the GHC’s run-time system. That bug has been fixed, and once I can get a new stable version, I’ll better know how close we are. I’ve also been exploring changes to Reactive’s programming model that I think will make FRP more well-suited to space-time modular user-interaction. You can find hints about these changes in some of my recent blog posts.

]]> By: Artyom Shalkhakov http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-310 Wed, 14 Jan 2009 04:48:08 +0000 http://conal.net/blog/?p=75#comment-310 An enjoyable post, thanks. The functional approach seems very elegant.

I haven’t looked into Reactive and FieldTrip, are describing them?

PS when will Reactive be ready for usage in real applications?

]]> By: conal http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-309 Wed, 14 Jan 2009 03:36:24 +0000 http://conal.net/blog/?p=75#comment-309 Hi Michael. I love your idea for extending the FRP abstractions in ways that account for reflections, diffraction etc, and then seeing what these extended abstractions suggest for FRP. Thanks!

]]> By: Michael Robinson http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-308 Tue, 13 Jan 2009 20:02:47 +0000 http://conal.net/blog/?p=75#comment-308 This is a clever idea, and an enjoyable post! You could extend it to permit accounting for more complicated phenomena, like reflections or diffraction by changing your Time type to be a rooted tree, which of course has a notion of partial ordering rather than a total ordering as in Ord. Metaphorically, the root of such a tree lies in the Image surface, each edge is marked with a distance/propagation time, and each node in the tree is on a reflective portion of your Geometry. It seems this might mesh well with your notion of interactive FRP.

]]> By: Dougal Stanton http://conal.net/blog/posts/3d-rendering-as-functional-reactive-programming#comment-307 Mon, 12 Jan 2009 10:22:41 +0000 http://conal.net/blog/?p=75#comment-307 I’m currently reading Steven Pinker’s The Stuff of Thought. He devotes lengthy sections to the discussion of repurposing basic metaphors for more abstract domains. One of the simplest he describes is using metaphors of space to describe time (time as a journey, time as a flowing river, looking into the past or the future). It’s interesting that you’re basically doing the opposite here: you already had the temporal metaphors (well, functions) worked out and you’re reusing them to describe three-dimensional space.

]]>