Jump to content
  • Advertisement


  • Content Count

  • Joined

  • Last visited

Community Reputation

100 Neutral

About Seikilos

  • Rank
  1. Sorry, it wasn't meant as an offence. I don't want that someome is doing my work for me since I still insist on the fact that even the thread's title implies that I am searching for a library. Which actually can be interpreted as : Yes I want that someone does the work for me -> the library I am looking for. However it was definitely my fault not mentioning C++ as the target language. Oh additionally the lib may be closed source too. The issue is, that I'm lacking the exact term of the kind of work I need to do and for what I am searching the lib. This results in a somewhat shady requirements another problem is, that even we as the developer do not have a detailed specification of client's requirements on how exactly he wants the solution (and he refuses to give us them, but plese, do not drift to project managment issues now). That's why I am looking for a library that has basically already done 95% of clients whishes even if he doesn't know of them yet. You know the sooner the deadline the more attention to detail occurs on parts which were implicitly accepted before and now won't
  2. Sure. Sorry. It's C++ of course. I seldomly think at other languages when it comes to massive load real time applications
  3. Sorry, Java is definitely not in discussion
  4. @szecs: Sorry for the feeling breing ignored, I have read your explanations and also concidered them. Our main problem at hand is that our efforts of path extrusion are not as stable as the client want them to be (of course without paying for it) The initial question was about a lib and I have spent quite a time googling, the negative results are the direct cause for the thread here. I wanted to shift my labor from reinventing a probably well solved field to using a library that already has this functionality. Our solution has issues with outline detection and the resulting data is not easy to triangulate (direct effect) In summary what I need is a lib for vector drawing that is extrusion of paths and providing additional functionality like stipple pattern onto this extruded line. Ill-defined extruded vertices by our algorithm cause further issues with applying stipple patterns (which is dependent on correct and meaning-full extruded paths) Edit: There isn't much time of throwing all our algorithms away and try new once (that's why I am rather asking for a lib than for rendering tipps which may not fit in our context). This should not sound unthankful since I appreciate the hints as something to concider
  5. No, no access to the former tool but we know that it was based on TCL. There are even more requirements like stipple pattern on the resulting line which is a daunting task by itself
  6. It's not solved even though I have an approach atm :) It isn't like I am not able to solve it but I hoped to find a library that has already done it But thanks for some helpful tipps :)
  7. The client wants it to be. No matter how you turn it. The previous tool was able to do that (even on a entirely different base) and the new tool must be able to do at least as good as the old one. My current apporach is: Since I am able to detect two vertices which would produce a degenerated triangle, I do an intersection test with those and the original direction to determine a new intersection point. The generation algorithm would create two triangles on the same coordinate (since I made one coordinate from two this is of course replicated) To avoid double drawing on the same coords I need to drop one triangle which then interferes with my triangle drawing algorithm. *sigh*
  8. The "slow" factor is key issue here. We have of course cached drawing so no recalculating if not necessry but at some point when redrawing occurs, it always occurs on all such lines at the same time (of course only within view bounding rect) but a typical test case which we somehow 'must' pass is a recalc of several hundred polylines with dense packed path coordinates resulting in some >10.000 vertices for which intersection and reiterion must be done. Oh and of course this is not the only part of the draw, there textures and so on. It's like wanting a great AI but only giving 1% of the computation power to it
  9. @szecs: Your approach is currently similar to our current solution. The issue with it, as you pointed is a posssible intersection of your black line. We did some fallback paths for those kind of solution but it turned out that the client is twisting this damned polyline as long as he find some instability, e.g. intersections. @xissburg: Im sure this isnt stable too. Or there is an additional AND-ing done to create a correct description of the area. You could try to draw szecs first image and create a very thick outline and check whethere the inner edge does intersect or not
  10. Well the list of vertices is drawn by opengl as single triangles and is somehow restricted since the draw process is already under heavy usage where using additional stencil buffer is already a performance hit. Using voxel cloud and iso surface extraction on a large amount of data would not be accepted by the client. Even though I would love to use some third party lib since the details of this process are quite irrelevant for us atm and not to implement additional algorithms which create a new set of possible errors and testcases
  11. Yeah, you got the issue pretty well (without me requiring to draw it :D) And since there is also alpha involved I cannot simply hide some "not-so-good" situations :) Basically it must be perfectly generated
  12. It must be highly realtime since it is a mission critical here :)
  13. Yeah that's somehow wrong :) It is very complex and not even Inkscape does it good. I have already done that and Image this example above with much more points, like an spline rasterizer would generate a dense vertex list. Bow if you extrude these vertices by a certain length where where the "normals" of the direction start to intersect you end up with a degenerated vertex geometry. And to avoid reinventing this wheel I ask here for libs. There is already a significant amount of time spent for this task including several people. The complexity lies in the details. So thats why I look for a "path extrusion" library :)
  14. Ok, to be clear now with image: Left is the input data, I get vertices and a width. E.g user clicks three times onto a canvas and defines a width. On the right side the green dots are the required output. To draw it I need the real coordinates of the vertices extruded (is this the right word) about the factor "width" .
  15. xissm thanks. Didnt know the name of that :) But did you understand me? I was not talking about mesh creation or tesselation of shapes. Your google search did not produce the hoped results @Kwizatz: It is in 2D and I need to locate points in a planar xy layer which are determined by cordinate and width. Just imagine a 2d drawing tool for vectors (e.g 2d poly-line). You place n points and define a width. And this rendering is quite complex for many situations :) Edit: Just to be sure, I looked into CGAL for instance and I found nice polygon generation from points, like I describe some area and CGAL calculates the vertices but I describe a path and width which need to be transformed into some polygonal representation [Edited by - Seikilos on November 4, 2009 10:04:01 AM]
  • Advertisement

Important Information

By using GameDev.net, you agree to our community Guidelines, Terms of Use, and Privacy Policy.

GameDev.net is your game development community. Create an account for your GameDev Portfolio and participate in the largest developer community in the games industry.

Sign me up!