Talk:Volume ray casting

I don't think that the Paper by Krueger et al. should be the only external reference to this article.

merge suggection

Oppose. See Talk:Ray casting about merge.

Basic algorithm

The description makes it seem as if this was the only way to do it. There are, however, instances of this algorithm that don't do any shading (only classification/coloring).

Furthermore, it makes sense to distinguish between pre- and post-classification and explain their respective advantages/problems. (See e.g. this master thesis)

Similarly, back-to-front compositing is not the only compositing method. Front-to-back compositing offers optimization possibilities of early out alpha tests. (See e.g. GPU Gems Chapter 39)

The advanced algorithms should also mention preintegration methods. Spynacker (talk) 10:26, 3 November 2015 (UTC)

On second thought, most of this should be explained on the general Volume rendering page but should be referenced/mentioned here. --Spynacker (talk) 11:11, 3 November 2015 (UTC)

Advanced adaptive algorithms

I see no citation for this part and it lacks an explanation what the samples adapt to. Spynacker (talk) 10:12, 3 November 2015 (UTC)

Ray-casting vs Ray-tracing

The ray-casting article makes the distinction between ray-casting and ray-tracing. This article's lead does not make this distinction and continues the confusion of the two. I see the distinction is made when the application and context are clear, but perhaps this should be addressed further? BlitzGreg (talk) 11:10, 10 December 2015 (UTC)

I could not find any stark distinctions inside ray-casting, it now states "It is essentially the same as ray tracing" (and the ray tracing article states: Roth invented the term "ray casting" before hearing of “ray tracing”, but they are essentially the same.)

I think it makes sense to think of Volume ray casting and SDF raymarching as different kinds of broader ray tracing term. They all trace rays, and they are all NOT rasterization (which would be the opposite of ray tracing if I had to pick a single one)... The distinctions in the article could still be kept, but they would be between Volume ray tracing and geometric-primitive ray-tracing (or polygonal-mesh ray-tracing? or something like that...) teadrinker (talk) 03:37, 2 December 2020 (UTC)

Ray Marching should not direct here

I think in computer graphics today, ray marching is generally thought of as the traversal of a space defined by signed distance function (SDF), using bounding spheres. (a method very popular on the community shadertoy.com)

This commonplace definition is very different from the one described on this page. Ray marching as understood by people at shadertoy for instance, can spawn secondary rays (but does not have to).

Note that the usage of the term "ray marching" to describe this method was popularized along with GPU hardware that supported fragment shaders. In early papers/literature, the algorithm might initially have been called other names, like sphere tracing etc.

teadrinker (talk) 20:26, 29 July 2020 (UTC)

@Teadrinker: Sphere tracing is one of several methods for ray marching, so it's not a synonym. Ray marching often uses iterative methods to find intersections between rays and implicit surfaces, so this appears to be different from the algorithm that is described here. Jarble (talk) 21:09, 22 November 2020 (UTC)

@Jarble: Agree, I tried to clarify by stating these 2 kinds teadrinker (talk) 19:07, 1 December 2020 (UTC)

@Lithopsian: I saw that the clarification was reverted... I still feel this is a great page for Volume ray marching, but the description of the alogrithm is too specific to also cover the typical sdf raymarching case. So if people come to wikipedia looking for ray marching in the sphere tracing / sdf context, they will just be confused (and this seem to be the most common use of the term if I just do a google search). Note that I think it would be more logical if people just called this sphere tracing, and it was categorized as a acceleration / space traversal technique for ray tracing, but unfortunately that is not how it turned out. teadrinker (talk) 20:56, 1 December 2020 (UTC)

These definitions seem fine to me. teadrinker (talk) 21:40, 1 December 2020 (UTC)

And yet they state quite explicitly that sphere tracing is just one of many types of ray marching. You don't seem to have grasped that yet, perhaps because it is the only one you're familiar with. Lithopsian (talk) 21:45, 1 December 2020 (UTC)

Kudos for discussing instead of edit-warring. The clarification was neither an article nor a disambiguation page. As an article, it did not attempt to define Ray marching, which is a fundamental problem. A normal Wikipedia article should be about one thing. You may be able to write an article about a single concept of "ray marching", I remain to be convinced that there are really two different things here. But you didn't do that, so it got reverted. There are also disambiguation (dab) pages, which are a list of wikilinks for a term where it is ambiguous which article that term would most commonly indicate. Dab pages have a defined format and are tagged as such, which you didn't do, but no matter. More importantly, they are lists of Wikipedia articles, not lists of everything you think the term might indicate. There is no Wikipedia article for sphere tracing, although it may be notable enough to have one, so no dab page. You could even find an existing article that discusses sphere tracing and either create a redirect to it or just link the dab page to it. I didn't find such an article, but maybe you could shoe-horn a section into an article somewhere. If and when you have two articles for an ambiguous term, usually one of them will be a more common usage than the other. This would be the primary topic and you would redirect to it (possibly with a WP:HATNOTE for anyone that wanted the other article). So potentially you could just re-target ray marching to different article if you feel that would be a more common usage, but again I didn't find anywhere suitable. You can get valuable clues about a term by looking at its usage in existing Wikipedia articles, and also be warned off if it is linked from a lot of places. Breaking all those wikilinks, or even "fixing" them, can be a very bad idea. For every internal wikilink, there could be a hundred external urls linked to the page, so you'd better be very sure that they all want to end up at your new target. Again, I remain to be convinced that there really are two different things here. In particular, I disagree with your contention that ray-marching is most commonly considered to be traversal of a volume using an SDF, however familiar that may be to you from one context. If you feel that ray-marching is either different from what is described here, or at least also refers to something different, I would suggest attempting to write an article about it, perhaps at Draft:Ray marching. You can then get feedback about whether the topic is notable, whether the article is a valid description of the topic, whether it should stay with this name, and technical things like references, layout, etc. There is an articles for creation method that will formally advertise a draft for review. Alternatively, you could go to Wikipedia talk:WikiProject Computing and discuss the whole thing first with people who know more about it than me, and probably more than you. It is fairly active so you should be able to get a few opinions.

@Lithopsian: Thanks for this detailed explanation, I opted to create Draft:Ray marching as suggested. I agree that sphere tracing is just one of many types of ray marching, that's is kind of why I started this discussion in the first place :) teadrinker (talk) 02:32, 2 December 2020 (UTC)

Good luck. I don't really have the expertise to contribute, but I've tagged the redirect so people are aware that there is a draft in progress. Lithopsian (talk) 20:53, 2 December 2020 (UTC)

Raymarching doesn't use the algorithm described in this page. I'm all for giving ray marching a separate page.HiPeeps124816 (talk) 19:35, 31 July 2021 (UTC)

References