| 00:20.51 | *** join/#brlcad CIA-2 (~CIA@flapjack.navi.cx) |
| 01:06.40 | *** join/#brlcad cad185 (~181043d9@bz.bzflag.bz) |
| 01:27.46 | *** join/#brlcad macTRON (~18cfd6ee@bz.bzflag.bz) |
| 01:29.44 | macTRON | Is this thing on? |
| 01:30.16 | macTRON | Um . . . I have a question or two about brlcad |
| 01:44.23 | *** join/#brlcad PhantomBantam (~phantomba@dialup-4.238.67.205.Dial1.Providence1.Level3.net) |
| 01:59.16 | learner | PhantomBantam, what do you mean? |
| 01:59.42 | learner | you provide a datafile that consists of height values |
| 02:00.13 | learner | it's basically what most packages would call a "height field" used for modeling terrain, for example |
| 02:00.51 | PhantomBantam | So, all that file is is plain text file with numbers? |
| 02:01.22 | learner | no, it's a binary data file that it's expecting |
| 02:02.01 | learner | do you have an existing data set? |
| 02:02.10 | learner | or are you looking for something auto-generated? |
| 02:02.19 | PhantomBantam | No. |
| 02:02.38 | PhantomBantam | Yes, I am looking for something autogenerated. |
| 02:02.39 | learner | it's not too difficult to take an existing binary data, like an image, and convert it |
| 02:02.51 | learner | ahh, okay -- probably even easier |
| 02:04.14 | learner | lemme see if we have a prog that already does that |
| 02:04.17 | PhantomBantam | What kind of image? |
| 02:04.24 | PhantomBantam | Okay. |
| 02:41.00 | *** join/#brlcad CIA-3 (~CIA@flapjack.navi.cx) |
| 02:50.24 | learner | PhantomBantam, sorry for the delay, got sidetracked fixing a bug |
| 02:51.09 | learner | I don't see a random dataset generator.. the closest is a procedural database, but that generates a spline terrain field |
| 02:51.22 | learner | (src/proc-db/terrain) |
| 02:57.35 | PhantomBantam | Okay, I'll give it a shot. |
| 02:57.46 | PhantomBantam | thanks |
| 02:57.48 | learner | that said, let's say you have a png image of terrain data |
| 02:58.28 | learner | like, http://ftp.brlcad.org/tmp/terrain.png |
| 02:59.08 | learner | to make that into a 3D height field in mged, you have to convert the file into the format needed which is network unsigned short |
| 02:59.44 | learner | png-pix terrain.png | pix-bw terrain.bw |
| 03:00.13 | learner | that converts the png into a raw black and white image datafile |
| 03:00.30 | learner | then to convert it to dsp data, you can use brl-cad's cv command: |
| 03:00.50 | learner | cv uc nus terrain.bw terrain.data |
| 03:01.07 | learner | that terrain.data is now ready for import into mged as a dsp |
| 03:02.17 | learner | at the mged command prompt type the "in" command, and it will prompt you |
| 03:02.38 | learner | mged> in |
| 03:02.56 | learner | Enter name of solid: terrain |
| 03:03.02 | learner | Enter solid type: dsp |
| 03:03.18 | learner | Take data from file or database binary object [f|o]: f |
| 03:03.29 | learner | Enter name of file/object: terrain.data |
| 03:03.34 | learner | Enter width..: 512 |
| 03:03.38 | learner | Enter height...: 512 |
| 03:03.46 | learner | Interpolation: 1 |
| 03:04.13 | learner | Cut direction [ad|lR|Lr]: ad (doesn't matter much most of the time) |
| 03:04.33 | learner | Cell size: 1000 (1 meter per grid cell) |
| 03:05.02 | learner | Unit elevation: 500 (smaller is flat terrain, larger is mountainous) |
| 03:05.36 | learner | e terrain and you should see it show up in all it's glory |
| 03:07.34 | PhantomBantam | I'll try. |
| 03:28.25 | *** join/#brlcad PhantomBantam (~phantomba@dialup-4.238.67.177.Dial1.Providence1.Level3.net) |
| 03:28.52 | PhantomBantam | Sorry, I was disconnected for some reason. |
| 03:29.06 | learner | no problem |
| 03:29.08 | learner | any luck? |
| 03:31.09 | learner | let me know if you need me to repost any of the walkthrough again |
| 03:32.54 | PhantomBantam | No, it's still on the screen. |
| 03:36.18 | PhantomBantam | $ png-pix ~/desktop/terrain.png |
| 03:36.19 | PhantomBantam | -bash: png-pix: command not found |
| 03:36.19 | PhantomBantam | Computer:~ PB$ /usr/brlcad/bin/png-pix ~/desktop/terrain.png |
| 03:36.19 | PhantomBantam | ?????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????}}? |
| 03:37.07 | PhantomBantam | That's what I get when I try to use the png-pix. |
| 03:37.24 | learner | er, how do you normally run mged? |
| 03:37.43 | learner | you type /usr/brlcad/bin/mged ? |
| 03:38.09 | PhantomBantam | I use x11 |
| 03:38.20 | learner | you should add /usr/brlcad/bin to your path, i.e. add this line to your ~/.bash_profile: export PATH=/usr/brlcad/bin:$PATH |
| 03:38.32 | learner | assuming you use bash |
| 03:39.07 | learner | then you'll be able to type just pix-png, mged, etc |
| 03:39.33 | learner | that said, the command you needed was "png-pix terrain.png | pix-bw terrain.bw" |
| 03:39.37 | PhantomBantam | Oh, Okay. |
| 03:39.57 | learner | the output was being piped to pix-bw instead of specifying a temp file |
| 03:40.34 | learner | otherwise, that becomes two steps: png-pix terrain.png > terrain.pix |
| 03:40.41 | learner | pix-bw terrain.pix terrain.bw |
| 03:54.01 | PhantomBantam | I tried adding that line, and it still doesn't work |
| 03:54.45 | learner | it doesn't take effect until you either make a new shell/xterm/whatever or resource the file manually |
| 03:54.52 | learner | source ~/.bash_profile |
| 04:01.39 | PhantomBantam | I still get the question marks. |
| 04:02.01 | learner | yes, png-pix outputs data to stdout |
| 04:02.10 | learner | so you have to redirect it to a file or a pipe, like I showed |
| 04:02.20 | learner | 23:39:33: learner: that said, the command you needed was "png-pix terrain.png | pix-bw terrain.bw" |
| 04:02.37 | learner | 23:40:34: learner: otherwise, that becomes two steps: png-pix terrain.png > terrain.pix |
| 04:02.37 | learner | 23:40:42: learner: pix-bw terrain.pix terrain.bw |
| 04:03.18 | learner | make sense? |
| 04:04.20 | PhantomBantam | But then it says can't open "terrain.bw" |
| 04:04.48 | learner | which did you do? |
| 04:04.59 | learner | using the pipe or the two step |
| 04:05.23 | PhantomBantam | The pipe. |
| 04:06.03 | learner | ack, my bad.. typo |
| 04:06.13 | learner | twice even |
| 04:06.21 | learner | pix-bw sends to stdout as well |
| 04:06.38 | learner | so it's: png-pix terrain.png | pix-bw > terrain.bw |
| 04:06.51 | learner | sorry about that |
| 04:08.53 | PhantomBantam | Yes! It worked. |
| 04:09.48 | learner | you can pretty much feed any image into it like that |
| 04:09.57 | learner | though the default values assume a 512x512 image |
| 04:11.32 | learner | actually, never mind.. any image size should work.. pix-bw doesn't care about dimensions, neither does cv |
| 04:21.52 | PhantomBantam | I tried, but got:mapped file open failed |
| 04:21.52 | PhantomBantam | mapped file open failed |
| 04:21.52 | PhantomBantam | cannot find data for DSP |
| 04:22.15 | learner | what's your in line? |
| 04:22.33 | learner | you did do the cv step right? |
| 04:22.42 | PhantomBantam | yes |
| 04:23.35 | learner | hit the up arrow until you see the "in" line that you wrote |
| 04:23.51 | learner | should look something like: in terrain dsp f .... |
| 04:24.00 | learner | paste that here |
| 04:24.08 | PhantomBantam | in terrain.s dsp f ~/desktop/terrain.data 512 512 1 ad 1000 500 |
| 04:24.33 | learner | hmm, might be the ~ |
| 04:24.42 | learner | make it relative |
| 04:24.45 | learner | or absolute |
| 04:25.00 | learner | pwd |
| 04:26.45 | PhantomBantam | Hm? |
| 04:27.01 | learner | what does pwd say? |
| 04:27.24 | PhantomBantam | Users/PB |
| 04:27.45 | learner | <PROTECTED> |
| 04:27.54 | PhantomBantam | It works. |
| 04:28.07 | PhantomBantam | I just got rid of the ~/ |
| 04:28.20 | learner | yeah, that makes it relative |
| 04:28.23 | PhantomBantam | Wow, looks great. |
| 04:29.33 | learner | you can test out different heights by running "kill terrain" to delete it, then up arrow to the "in" line and change that last number (e.g. try 1000 instead of 500) |
| 04:29.53 | PhantomBantam | Okay, thanks a ton. |
| 04:29.58 | learner | small numbers will make flattened, gently rolling terrain |
| 04:30.04 | learner | large numbers will make mountains |
| 04:30.30 | learner | changing the second to last number will change how big/small it is |
| 04:32.58 | learner | it's one of the fastest and most efficient terrain representations.. you can go pretty darn big and still raytrace relatively quickly |
| 04:33.05 | learner | only limited by your machine's memory |
| 04:40.29 | PhantomBantam | Great. |
| 04:43.17 | PhantomBantam | Ineed to go now. Thanks a ton. |
| 04:55.37 | *** join/#brlcad PhantomBantam (~phantomba@dialup-4.238.67.177.Dial1.Providence1.Level3.net) |
| 04:55.40 | *** part/#brlcad PhantomBantam (~phantomba@dialup-4.238.67.177.Dial1.Providence1.Level3.net) |
| 10:09.26 | *** join/#brlcad ibot (ibot@apt.bot.TimRiker.active.supporter.pdpc) |
| 10:09.26 | *** topic/#brlcad is http://brlcad.org/ || BRL-CAD is now Open Source! || Screenshots: http://sourceforge.net/project/screenshots.php?group_id=105292 || http://brlcad.org/images/mged.jpg || Release 7.2.0 is posted (20050305) |
| 12:09.09 | *** join/#brlcad CIA-3 (~CIA@flapjack.navi.cx) |
| 14:41.34 | *** join/#brlcad narnia (~terrylr@johann.blauedonau.com) |
| 16:02.04 | learner | cia is still not working.. |
| 16:02.15 | learner | slew of commits and no gratification :) |
| 16:06.26 | narnia | no joy here either. |
| 16:07.43 | learner | did you actually start writing a compiler? |
| 16:07.56 | learner | or something else |
| 16:08.03 | learner | or was that just a joke |
| 16:08.31 | learner | if it wasn't, gamasutra has one of the best walkthrough's I've seen |
| 16:09.11 | learner | that's probably several months of effort to get right, though |
| 16:09.53 | narnia | i am rewriting the parser and working through a compiler design. several people have recommended the 'dragon book', but that book is not a possibility. |
| 16:11.39 | narnia | there is only so much patching possible with the old nist design. the more patches the uglier it becomes. |
| 16:11.41 | learner | it is probably "the" book |
| 16:12.29 | learner | but certainly not necessary unless you have a really complicated language (which express is not comparatively) |
| 16:12.30 | narnia | yes, the 'dragon book' is 'the' book on compiler design. |
| 16:13.00 | narnia | well, i sure would not say that. |
| 16:13.26 | narnia | the local 'where' clauses and global rules. |
| 16:15.10 | learner | why are you going that route, though? was there actually a showstopper? |
| 16:15.49 | learner | I would have defaulted back to the other c++ based parser next before starting fresh |
| 16:17.24 | learner | just curious |
| 16:18.25 | learner | woo hoo!.. libwdb now has support for writing binary data objects! |
| 16:19.20 | narnia | showstoppers are: 0) not having 10303 part 4x or 5xx series. 1) from the old nist scl source code readme file " The |
| 16:19.20 | narnia | following subclasses of STEPaggregate are stubs and have yet to be |
| 16:19.20 | narnia | implemented: Array, Bag, Set, List. |
| 16:19.20 | narnia | " |
| 16:20.41 | narnia | i have no idea what many entity attributes (variables) are even used for. |
| 16:20.57 | learner | now sure what the latter means, but perhaps it simply wasn't necessary |
| 16:21.29 | learner | isn't the 5xx series just for specific validate tests? what's 4x? |
| 16:21.31 | narnia | then there are all rhe exceptions to the rules that are introduced in 10303-11ed2, 10303-21-2002, etc. |
| 16:22.21 | narnia | Integrated generic resources (Parts 40-49) |
| 16:22.32 | narnia | Application interpreted constructs (Parts 500 - 599) |
| 16:23.02 | learner | okay |
| 16:23.11 | learner | why are those needed?? |
| 16:24.18 | learner | sorry for all the questions, you do know the scl a heck of a lot better than I at this point, but for basic geometry parser those don't seem that critical |
| 16:25.15 | learner | useful, maybe .. critical.. dunno |
| 16:25.17 | narnia | example: 10303-042e2 defines the entity trimmed_curve. i have nothing that tells me how that trimmed_curve behaves. |
| 16:26.02 | narnia | all geometry is defined in part 4x and 5xx. |
| 16:26.26 | narnia | ap203 just includes them. |
| 16:27.02 | narnia | i have absolutely nothing to go by that says how the geometry entities interact. |
| 16:27.18 | learner | sure, but why do you need to know how they behave? a trimmed curve is going to have some values that pretty much convert 1-1 with a nurb .. it's the app's arena to make it behave |
| 16:28.18 | learner | for a converter, behavior is limited to "is this thing valid and representable" |
| 16:28.48 | learner | if I were modeling with step as the core data format, that wouldn't be the case, but for a converter.. |
| 16:29.33 | learner | there's already going to be a fair amount of simplification/omission or unsupported features even |
| 16:34.04 | narnia | i am not understanding what you are are saying. when i asked about the conversion polyspin said that it should either be a bag of triangles or a nm something. |
| 16:35.10 | learner | yes, that was specific to the various tesselations you were asking about |
| 16:35.43 | learner | bots are best for triangle formats, nmg for polygons |
| 16:36.24 | learner | splines trimmed or otherwise can either be tesselated into a triangulated format or brought in as a nurb directly |
| 16:37.41 | learner | that's more knowledge of brl-cad's abilities, though.. not the step's representation's "ability" |
| 16:39.02 | learner | a sphere is a good example, given that it can be represented in about a half-dozen ways |
| 16:39.26 | narnia | out of the numerous .stp files i have been using i have yet to find a true polygon. |
| 16:40.14 | learner | spline spheres and actual sphere primitives in step formate would come in as a brl-cad sphere, a tesselated one would come in as a bot if it's triangulated, nmg otherwise |
| 16:42.00 | learner | those mappings do get trickier, of course, but they pretty much just have to be looked at on a 1 by 1 basis to figure out the easiest mapping |
| 16:45.12 | narnia | example of what i am dealing with. |
| 16:45.17 | narnia | #9120 = POLYLINE('PolyLine2',(#9113,#9118,#9117,#9116)); |
| 16:45.17 | narnia | #9130 = POLYLINE('PolyLine2',(#9110,#9113,#9112,#9117)); |
| 16:45.17 | narnia | #9140 = POLYLINE('PolyLine3',(#9118,#9115,#9110,#9111)); |
| 16:45.17 | narnia | #9150 = POLYLINE('PolyLine4',(#9112,#9111,#9116,#9115)); |
| 16:46.26 | narnia | the above polyline entities define 4 lines between the cartesian_points given as arguments. |
| 16:46.45 | learner | so 9113 is a cartesian point for example? |
| 16:46.58 | narnia | the four polylines taken together form a cube. |
| 16:47.13 | learner | a cube or a square? |
| 16:47.27 | narnia | 9113 is an entity instance name: #9113 = CARTESIAN_POINT('CartPt4',(0.0,0.0,2.0)); |
| 16:47.36 | narnia | a cube |
| 16:48.12 | learner | hmm.. how do four lines make a 3d cube? |
| 16:48.48 | narnia | #9110 = CARTESIAN_POINT('CartPt1',(0.0,0.0,0.0)); |
| 16:48.48 | narnia | #9111 = CARTESIAN_POINT('CartPt2',(2.0,0.0,0.0)); |
| 16:48.48 | narnia | #9112 = CARTESIAN_POINT('CartPt3',(2.0,0.0,2.0)); |
| 16:48.48 | narnia | #9113 = CARTESIAN_POINT('CartPt4',(0.0,0.0,2.0)); |
| 16:48.48 | narnia | #9115 = CARTESIAN_POINT('CartPt5',(0.0,2.0,0.0)); |
| 16:48.49 | narnia | #9116 = CARTESIAN_POINT('CartPt6',(2.0,2.0,0.0)); |
| 16:48.51 | narnia | #9117 = CARTESIAN_POINT('CartPt7',(2.0,2.0,2.0)); |
| 16:48.53 | narnia | #9118 = CARTESIAN_POINT('CartPt8',(0.0,2.0,2.0)); |
| 16:48.55 | narnia | #9120 = POLYLINE('PolyLine2',(#9113,#9118,#9117,#9116)); |
| 16:48.55 | learner | ahh, I think I get it.. polyline is basically a face |
| 16:48.57 | narnia | #9130 = POLYLINE('PolyLine2',(#9110,#9113,#9112,#9117)); |
| 16:48.59 | narnia | #9140 = POLYLINE('PolyLine3',(#9118,#9115,#9110,#9111)); |
| 16:49.01 | narnia | #9150 = POLYLINE('PolyLine4',(#9112,#9111,#9116,#9115)); |
| 16:49.08 | learner | multiple lines, not just one |
| 16:49.14 | learner | 4 to be exact |
| 16:49.21 | narnia | well, no. that is what i thought originally. |
| 16:50.28 | narnia | hang on i have a povray model that shows what i am attempting to convey. |
| 16:50.55 | learner | i don't see how a single polyline can be anything else but line segments from point to point |
| 16:51.11 | learner | the only question I'd have is whether it was a closed loop or an open loop |
| 16:51.19 | learner | from the points, it looks open |
| 16:53.06 | learner | yeah, that makes more sense .. they're just connected line segments, and presumably there's another entity that ties them together |
| 16:56.28 | narnia | well no, there is no further entity that ties them together. |
| 16:57.04 | narnia | http://204.238.34.206/cube1-polyline1.jpg |
| 16:57.08 | learner | then it's not solid geometry |
| 16:57.10 | narnia | http://204.238.34.206/cube1-polyline2.jpg |
| 16:57.15 | learner | it's just an outline of a cube |
| 16:57.16 | narnia | http://204.238.34.206/cube1-polyline2.jpg |
| 16:57.23 | narnia | http://204.238.34.206/cube1-polyline3.jpg |
| 16:57.31 | narnia | http://204.238.34.206/cube1-polyline4.jpg |
| 16:57.36 | learner | something you might find output from a sketch editor |
| 16:57.54 | narnia | http://204.238.34.206/cube1.jpg |
| 16:58.16 | learner | yep |
| 16:58.23 | learner | i just made the same thing in mged |
| 16:58.59 | narnia | i agree, but according to way 10303 sees it it is a cube. |
| 16:59.03 | learner | that is not a cube, though, at least from a modeling perspective |
| 16:59.10 | learner | it is simply 4 line segments |
| 16:59.21 | narnia | i totally agree. |
| 16:59.34 | learner | i recall seeing primitive datatypes, though |
| 16:59.39 | narnia | but to step that is a cube. |
| 17:00.31 | learner | what makes you say it's a cube, nothing from the little bit of express you pasted would indicate that |
| 17:01.26 | learner | there's nothing that even indicates that they're any different from any random polyline I can think up |
| 17:01.35 | narnia | i agree. |
| 17:02.14 | learner | well, that would imply that it isn't then.. nothing to the contrary contradicts it |
| 17:02.56 | learner | dumping out 1000 points on a sphere's surface and 100 polylines that randomly connects them all doesn't make a sphere |
| 17:03.17 | learner | even non-random connections like this cube |
| 17:04.03 | learner | presented with what you've shown, there's nothing to import/convert yet |
| 17:04.10 | narnia | okay, using the above 'cube' example. how should that be mapped to brl-cad? |
| 17:04.48 | learner | there's no solid geometry yet |
| 17:05.22 | learner | so there is no mapping yet for at least anything we'd care about |
| 17:05.23 | narnia | you are saying there is nothing to map? |
| 17:05.54 | learner | exactly, like I said.. there is nothing that says those four poly lines have _anything_ to do with each other even |
| 17:06.08 | learner | I can't just assume they do because they're in the in file |
| 17:06.18 | learner | there could be thousands of polylines like those |
| 17:07.00 | narnia | well, in one of the .stp files i have been using there are 1000's of polylines like that. |
| 17:08.03 | learner | i mean add one more #9151 = POLYLINE('PolylineFoo', (#9111, #9112, #9113, #9114)); and it's no longer a cube |
| 17:08.20 | narnia | agree |
| 17:08.52 | learner | there has to be another entity that ties the polylines together or there's no information about connectivity yet |
| 17:10.53 | learner | some editors will let you make non-solid-geometry like a segmented line that has no solid geometry value |
| 17:11.23 | learner | so I could see a case where something might export that to a file, but there's "nothing there" yet |
| 17:11.48 | learner | you either warn and go on, abort, or ignore it |
| 17:12.01 | learner | i'd ignore that particular geometry until something connects them together |
| 17:13.10 | learner | stash all the polylines in memory, but if they're not used to make something (either a 2d sketch or a 3d surface/object/shape) then dump them |
| 17:13.47 | narnia | even if the .stp used something like :: GEOMETRIC_CURVE_SET('GeomCurSet1',(#9120,#9130,#9140,#9150)); |
| 17:13.49 | learner | you could technically stash them in the brl-cad db, but not in a way useful to anything but the converter |
| 17:14.31 | learner | now that's totally different.. that ties the polylines together |
| 17:14.59 | narnia | which would 'join' the 4 polylines into a labeled curve set. it still does not convey a solid. |
| 17:15.19 | learner | that there makes a geometric set, which happens to be a closed surface |
| 17:16.15 | learner | I'd still hope there's another entity that references that entity or it's name so that I know it's supposed to be a cube |
| 17:16.26 | narnia | i agree it is a closed surface |
| 17:16.36 | learner | so that it can directly import as a arb8 |
| 17:17.32 | learner | otherwise you'll have to tesellate that surface and bring it in as an nmg (6 unoriented polygons) |
| 17:18.51 | narnia | what does arb stand for again? |
| 17:22.55 | narnia | the point i have been attempting to make is that in many cases there is no way of knowing what the 'object' is until the entire .stp file has been parsed and the 'object' is 'assembled' in memory. only then, in many cases, do i know what the 'object' 'looks' like. |
| 17:25.25 | learner | arb is an arbitrary regular polyhedron with as many as 8 vertices |
| 17:25.57 | learner | basically all of the polyhedron you can make out of the faces of a cube |
| 17:26.34 | learner | a cube, a wedge, a 4-faced pyramid, a 5-faced pyramid, etc |
| 17:27.26 | learner | i don't doubt that you won't be able to do anything until the entire tree is assembled in memory |
| 17:27.32 | learner | that was what I presumed libexpress actually did |
| 17:27.48 | narnia | well no. |
| 17:27.53 | learner | then I can iterate over the geometric types, and convert |
| 17:28.01 | narnia | libexpress parses the schema. |
| 17:28.30 | learner | which gives me something that will eat ap203 for example |
| 17:28.34 | narnia | i.e. 10303-203-aim-long.exp |
| 17:28.42 | narnia | well no. |
| 17:29.17 | learner | eh, i just said what you did and you say "well no."? :) |
| 17:29.33 | learner | that is ap203 |
| 17:29.53 | learner | the express schema for it |
| 17:29.54 | narnia | hang on brb. |
| 17:35.18 | learner | I don't think we need to get caught up in the exact semantics, but feel free to enlighten me. from what I gathered at glance, libexpress parses a schema you provide like ap203. that then effectively gives you the means to parse ap203 step files and read the data as entities. you get all the entities into memory (either manually or assisted) and then have to decide what/how to convert based on all of the entity types |
| 17:37.42 | learner | that doesn't "solve the entire problem", but once all the entity types are read into memory I have enough to convert -- i could iterate over all of the GEOMETRIC_* entity types, for example, and convert accordingly |
| 17:40.02 | learner | of course, having to recursively dereference subentities all the way down the heirarchy to get the actual data like in the cube example.. I'd end up with a 6-sided polygon that is even detectable (at least that example is, forms a nice closed surface) and import directly as either an arb8 or an nmg |
| 17:43.30 | narnia | here is the current design: |
| 17:44.34 | narnia | parse a part21 .stp file. based on the part21 header using the appropriate ap parser parse the data sections of the part21 .stp file |
| 17:45.20 | narnia | build a linked list of the entity instances found in the data section. |
| 17:45.53 | narnia | based on the part21 header information validate the data section. |
| 17:48.02 | learner | okay, so far it matches what I've expected |
| 17:50.09 | narnia | for ap203 there are several 'presentation' types. |
| 17:50.14 | narnia | advanced_boundary_representation |
| 17:50.28 | narnia | <PROTECTED> |
| 17:50.39 | narnia | <PROTECTED> |
| 17:50.49 | narnia | <PROTECTED> |
| 17:51.00 | narnia | <PROTECTED> |
| 17:51.21 | learner | yes, those are the specifics that polyspin responded to.. they map rather well to the types he mentioned |
| 17:51.54 | narnia | depending on the 'presentation' type create an in memory model of the 'object'. |
| 17:53.34 | learner | now in the prior example, how does the GEOMETRIC_CURVE_SET relate to presentation types? |
| 17:54.03 | learner | is it tagged as one of those or implicitly belongs to one of those or unrelated? |
| 17:58.09 | narnia | it would either be advanced_brep_shape_representation, edge_based_wireframe_shape_representation, faceted_brep_shape_representation,geometrically_bounded_surface_shape_representation, geometrically_bounded_wireframe_shape_representation, manifold_surface_shape_representation, shell_based_wireframe_shape_representation, or one of the topological_representation_item |
| 17:59.22 | learner | so implicitly belongs to any of those.. is there a means that it would be tagged as one versus another? |
| 18:01.50 | narnia | generally, the last entity instance in the .stp file indicates which. example: #9200 = GEOMETRICALLY_BOUNDED_WIREFRAME_SHAPE_REPRESENTATION('gbwsr',(#9160),#9190); |
| 18:02.21 | narnia | #9810=GEOMETRICALLY_BOUNDED_WIREFRAME_SHAPE_REPRESENTATION('gbwsr',(#9800),#9828); |
| 18:02.32 | learner | ahh, beautiful |
| 18:02.44 | learner | that's the actual solid geometry |
| 18:02.56 | narnia | in the above example: 9800 is: |
| 18:03.11 | narnia | #9800 = GEOMETRIC_CURVE_SET('GeomCurSet1',(#9110,#9111,#9130,#9131, |
| 18:03.11 | narnia | #9562,#9566,#9571,#9576,#9581,#9153, #9586,#9158,#9159,#9591,#9173, |
| 18:03.11 | narnia | #9178,#9596,#9185, #9601,#9606,#9194,#9195,#9198,#9611, |
| 18:03.11 | narnia | #9616,#9211,#9215, #9621,#9227,#9626,#9631,#9238, |
| 18:03.11 | narnia | #9239,#9242,#9636,#9641, #9258,#9263,#9646,#9272, |
| 18:03.12 | narnia | #9651,#9656,#9283,#9284,#9287, #9661,#9666,#9303, |
| 18:03.14 | narnia | #9308,#9671,#9317,#9676, #9681,#9328,#9329,#9332,#9686,#9691, |
| 18:03.16 | narnia | #9348,#9353,#9696, #9362,#9701,#9706,#9373,#9374, |
| 18:03.18 | narnia | #9377,#9711,#9716,#9393, #9398,#9721,#9407,#9726, |
| 18:03.20 | narnia | #9413,#9414,#9450,#9451,#9460,#9461,#9470,#9471,#9480,#9481,#9490,#9491, |
| 18:03.22 | narnia | #9500,#9501,#9510,#9511,#9520,#9521,#9530,#9531,#9540,#9541,#9550,#9551)); |
| 18:03.24 | narnia | you get the point. |
| 18:03.49 | learner | actually, i'd say 9800 is not the geometry.. it's just the shape/collection |
| 18:03.50 | narnia | i have to 'drill' down to figure out what the object looks like. |
| 18:04.10 | learner | in the absence of 9810, I was presuming the contents |
| 18:04.18 | learner | same with the cube example |
| 18:04.35 | narnia | yes |
| 18:04.56 | learner | but having it tell me that it's one representation versus another makes it a geometric shape as opposed to a collection of values |
| 18:05.40 | learner | those represenation entities are what the converter would iterate over then |
| 18:05.57 | learner | recursively looking up their contents for conversion values |
| 18:06.24 | learner | build up an in-memory representation of the geometry |
| 18:10.00 | narnia | http://204.238.34.206/star1-p21.txt is a good example. i have been using this one for testing. |
| 18:18.27 | narnia | hang on i am trying to get a screenshot of the star1.p21 object. |
| 18:21.58 | narnia | opencascade shows http://204.238.34.206/star1-p21.jpg |
| 18:23.53 | narnia | how would you suggest that be mapped to brl-cad? |
| 18:29.20 | learner | looks like it'd convert in as a nurbs surface given all the trimmed curve entities |
| 18:30.00 | learner | alternatively, you could tesselate the curves and import in as a bot/nmg |
| 18:30.24 | learner | the latter wouldn't be ideal, but it would work easiest |
| 18:31.10 | learner | if you could detect that it's basicly one 2d shape, it best converts in as a sketch and an extrusion fully preserving |
| 18:31.55 | learner | it doesn't look like it is represented that way in the step file, though.. so probably nurbs |
| 18:33.21 | learner | the wireframe with curve models are going to be the hardest to decide on how to convert them -- probably just as easy to add a new wireframe primitive to brl-cad that fully captures it |
| 18:35.19 | learner | wireframe modeling provides no benefit over what we already have, it's just a different (less efficient) representation |
| 18:35.35 | learner | it could be added pretty easily, though |
| 18:36.04 | learner | it's basically an extension of the sketch primitive into 3D, just a heck of a lot harder to guarantee closure |
| 18:36.27 | narnia | i find that most of the .stp files i have seen could have been modeled in a more efficient manner. the star1 is a good example. |
| 18:37.05 | narnia | ap203 does have both a surface of revolution and surface of linear sweep. |
| 18:37.44 | narnia | the star1 should have been exported as a surface of linear sweep and be done with it. |
| 18:39.09 | learner | there is motivation to add both a wireframe primitive and/or a spline-based curvature boundary representation |
| 18:39.19 | learner | but that's somewhere down the line |
| 18:39.55 | narnia | do you have a better understanding of where i am coming from. |
| 18:40.12 | narnia | and what i am dealing with? |
| 18:42.44 | learner | csg would have you model that star as a cylinder with 20 objects (7 cylinders, 13 boxes) subtracted |
| 18:44.48 | learner | getting back to the original question, though, I haven't seen anything that warrants requiring a compiler .. you can get at the entities, those are pretty much all that's needed |
| 18:45.54 | learner | easy answer is to just tesselate that surface and bring it in as a bot -- that's pretty much what the iges converter does |
| 18:46.12 | learner | it's not 100% faithful, but then conversions almost never are |
| 18:46.28 | learner | for faithful, I'd take the two weeks to add that new primitive |
| 18:48.49 | narnia | the compiler is still needed at some point in the future. there is far too much glue that has to be written by hand at the moment. |
| 18:50.58 | learner | what is it doing for you though? |
| 18:52.09 | learner | you feed it a step file, it crunches it through to some optimal in-memory representation? |
| 18:53.40 | learner | there is not going to be a faithful representation no matter how you cut it without adding the union of all step-recognized geometry representations to brl-cad |
| 18:54.45 | learner | so you have to write the translation code.. that could go in a "compiler", i'd probably just call that a library -- or it can simply be the guts to the converter |
| 18:55.29 | learner | there's not much room for reuse without limiting the step to some subset in memory |
| 19:05.30 | narnia | i will think on it later. i need to go to farm&fleet and get some parts and gasoline for the lawn mower. my daughter is awake now so she can drive me. over cast day but still warm enough to get somethings done. also need to look at the water softner's timer and see why it stopped working. never an end to upkeep. |
| 19:06.06 | learner | alrighty, have a good one |
| 19:06.12 | narnia | always something demanding attention (repair). |
| 19:27.44 | *** join/#brlcad CIA-3 (~CIA@flapjack.navi.cx) |
| 21:25.04 | *** join/#brlcad CIA-3 (~CIA@flapjack.navi.cx) |
