Timeline for Toon/cel shading with variable line width?
Current License: CC BY-SA 3.0
21 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Nov 15, 2018 at 14:58 | answer | added | SnoepGames | timeline score: 0 | |
| Dec 6, 2012 at 15:53 | vote | accept | Engineer | ||
| Dec 4, 2012 at 19:03 | comment | added | Engineer | @Gajoo, As I was saying in the question, it's not just about the silhouette. Note the intrusive lines behind the girl's shoulder in the first image. Those are a function of the flipped-normals approach, and act as suggestive contours / creases. I need same. | |
| Dec 4, 2012 at 18:57 | comment | added | Ali1S232 | what about using same algorithm as the one generating red outlines, and then thinning/growing lines based on their depth? | |
| Dec 4, 2012 at 15:02 | comment | added | snake5 | 25 million vertices (linear reads/writes) vs. 23 million texture samples (rather poorly cached memory reads + decompression of render target data) @ 1280x720. Limiting/reducing vertex count is a lot easier, by the way. Especially these days, when big monitors (1920x1080+) & multimonitor setups are getting very popular. | |
| Dec 4, 2012 at 14:33 | history | edited | Engineer | CC BY-SA 3.0 | deleted 176 characters in body |
| Dec 4, 2012 at 14:26 | comment | added | Engineer | @snake5 You're honestly suggesting that I run through 150 million vertices per render update (about 25 million with view culling), in the GS? I'm not really buying that, but thanks for the input. Refer back to what I stated above about vertex counts and complexity. Even with your 10x figure, the fragment shader would at least draw even, and probably do a hell of a lot better. | |
| Dec 4, 2012 at 14:07 | comment | added | snake5 | Varying line thickness is a feature of the geometric approach. Can't efficiently get it any other way. Since you want that, I don't see pixel shaders as being the "more efficient method": number of pixels to be searched = (line_thickness * 2 + 1) ^ 2 - 1. That would basically mean that your post-process shader will be a lot (guesstimate: 10x) slower if max. line width equals 2. Just drop pre-judging and try the geometry shader approach. | |
| Dec 4, 2012 at 13:54 | history | edited | Engineer | CC BY-SA 3.0 | added 1948 characters in body |
| Dec 4, 2012 at 13:49 | answer | added | Engineer | timeline score: 4 | |
| Dec 4, 2012 at 12:14 | history | edited | Engineer | CC BY-SA 3.0 | deleted 43 characters in body |
| Dec 4, 2012 at 11:25 | history | edited | Engineer | CC BY-SA 3.0 | added 42 characters in body |
| Dec 4, 2012 at 4:01 | history | tweeted | twitter.com/#!/StackGameDev/status/275812134573711360 | ||
| Dec 3, 2012 at 15:12 | history | edited | Engineer | edited tags | |
| Dec 3, 2012 at 14:53 | history | edited | Engineer | CC BY-SA 3.0 | added 111 characters in body |
| Dec 3, 2012 at 14:37 | history | edited | Engineer | CC BY-SA 3.0 | added 3 characters in body |
| Dec 3, 2012 at 14:22 | history | edited | Engineer | CC BY-SA 3.0 | added 123 characters in body |
| Dec 3, 2012 at 14:03 | history | edited | Engineer | CC BY-SA 3.0 | added 43 characters in body |
| Dec 3, 2012 at 13:57 | history | edited | Engineer | CC BY-SA 3.0 | edited title |
| Dec 3, 2012 at 13:50 | history | edited | Engineer | CC BY-SA 3.0 | edited title |
| Dec 3, 2012 at 13:33 | history | asked | Engineer | CC BY-SA 3.0 |