Talk:Silhouette Edges

From Math Images

Jump to: navigation, search

The content of this page was written by Steve Cunningham. I am doing formatting and putting it on the Math Images site.

Nordhr 11:52 30 June 2011


Little suggestion: It might look nicer if the words were a clickable link, instead of having the URL written out, under Example for an Object.
--Chanj 21:37, 4 August 2011 (UTC)

Thanks for the suggestion :). Do you like that better?

Nordhr 17:52 4 August 2011

Contents

Checklist

While there's a lot of good content here, I feel like the page needs a lot of work. The basic description is set up so that the algorithm/method will be the center of the discussion, yet we never see all of the steps of the algorithm worked out in detail. The section on Determining whether a surface facet normal points toward the viewer doesn't feel finished because it isn't crystal clear what you do with the information about whether or not it's pointed towards the normal. While it's implied earlier on, it needs to be repeated for clarity.

I reiterated what is done with the pointing toward/away information. --Nordhr

Overall, I think that this page needs significantly more text and examples before approval

I made a section on shadow volumes to give the page more context and content, and perhaps will make sections on some other things silhouette edges are used for --Nordhr

Messages to the Future

  • I don't have any messages to the future for this page; I feel it is complete.

References and footnotes

  • Images were all provided by Steve Cunningham in his email (which I said in their discussion section)
  • There are no direct quotes
  • All content was written from Steve Cunningham's knowledge, so I wrote this in the reference section.

Good Writing

Context

  • Since the page is about silhouette edges, I think it is pretty clear that the context is that we are showing how to find these edges.

There might be something more on "Why this is interesting". Perhaps the primary use of silhouette edges, at least as far as I know, is to help show structure in a model, especially if the colors are all similar. That is, it is a technique for helping to highlight geometric information in an image. Notice that a silhouette edge is a geometric edge, not an image edge. Image edges are found with tools as described in the Image Convolution page. Both geometric edges and image edges are techniques used in scientific visualization. SteveC

Quality of prose and page structuring

  • I think that Steve Cunningham did a very good job writing the prose (as I said on the last page).
  • This page references two helper pages (vectors and matrices).
  • This page actually keeps the intense math to a minimum. Instead it talks about concepts and what math would be done, without putting too much syntax.
  • I'm finding I'm a bit confused by the description of the algorithm in the basic description. This could be because you're not defining facet clearly. My confusion is coming from the fact that your first "if" clause seems like it's picking out faces that have normals pointing "towards" the viewer (what you mean by "towards" is unclear), but the second if clause that's within the first one says "If this surface normal points away from the viewer." Wouldn't you have already eliminated those, based on the first if clause? Basically, as someone unfamiliar with this content, I'm having a really hard time understanding this bit.
On Steve Cunningham's suggestion, I changed the word 'facet' to 'triangle' because the facets are just the triangles that make up the 3D shape. He also pointed out that the surface normal page directly describes how to computer the surface normal of a triangle, so I hope that is more clear now. I also added in another sentence to help explain what that sentence means: there is one triangle facing towards the viewer and another facing away. When two such triangles are next to each other, the edge they share becomes part of the silhouette edge. --Nordhr
  • Given that you talk about this algorithm in the beginning, I'd like a full walkthrough of it for an example somewhere on the page. This may take a while, but I think it's necessary.
  • This sentence: "

If you use the technique of constructing a grid for the function domain, as described in the discussion of surfaces, for each facet you can identify the adjacent grid triangles and thus the adjacent facets, and so you can compute the facet surface normals from either the geometric or analytic approach, as described above. "

is a long run-on/comma splice. That should be broken into at least three sentences. Because of the grammar problems, it is extremely hard to understand.
I see where that gets a bit wonky. I tried to fix it to sound better and make more sense. Was what I changed it to understandable to you? --Nordhr

Integration of Images and Text

  • All images are specifically referenced and explained in the text.
  • Images and text go hand in hand to explain concepts.

Examples, Calculations, Applications, Proofs

  • Examples of concepts are provided in the form of images on this page.

There might be some confusion in the Basic Description section by talking about "facets." I view a 3D object as being defined by a set of triangles (perhaps other shapes, but almost always triangles) in the surface. Every point in the surface lies in a triangle, and two adjacent triangles meet at a common edge. These triangles are the facets I talk about -- so you could replace "facet" with "triangle" in this description and have something easier to read. Then later on, "computing the surface normal of a facet" becomes "computing the surface normal of a triangle" and that is explicitly covered in the Surface Normals page. SteveC

  • Definitions are handled by linking to other pages, explanations, and images.

Mathematical Accuracy and Precision of Language

  • The page has been proofread by me and others; I believe it is free of errors.
  • I feel the page is worded very well, and there aren't too many symbols because the page is very conceptual.
  • This page links to other pages (i.e. Surface Normals) when needed.

In the "Determining whether a surface facet normal ..." section, the statement "if the angle between N and V' is positive;" should be "if the angle between N and V' is acute;" -- this is probably my error. But acute angle = positive dot product. SteveC
Yep, you're right. I fixed it now. -- Nordhr

Layout

  • Paragraphs are short and there are a lot of pictures throughout.
  • The example section is hidden to cut down on volume of visible text in the "A More Mathematical Explanation" section.
  • There is some whitespace next to the pseudocode, but I feel like that is necessary and makes it more readable. It is also next to the picture, so the whitespace is minimized.
  • Images are centered and don't disrupt text formatting.
  • All images are contained completely in the correct section.
  • The hide/show looks nice.
  • The page resizes very well. At very small widths the pseudocode has line breaks in the wrong place, but the indentation stays correct. At this width the image is already larger than the page, so I think it is a little extreme.
  • Something funny's going on with the formatting here: "(b) two tangent vectors at a point in the facet computed by partial

derivatives. "

Yep, I see what happened there. All fixed. --Nordhr
Personal tools