Naughty Alien

..but my grid generator IS actually Navmesh lets see how nav mesh works, right? Basically, a "navigation mesh" is a logical graph of convex regions that defines where a computer-controlled character is allowed to move. Often these regions are a lower-resolution version of the model of the floor on which the game's characters walk. The character can move anywhere inside the physical boundaries defined by the regions, and the accompanying graph can be used with various algorithms, such as the A* algorithm, to solve paths which can, in turn, be used by a AI algorithms to help decide how the character should react to its environment. For example, consider a room that has some obstacles in it as in the image below. The navigation mesh for this room might look like the left-most image below. We break up the floor of the room into a set of convex regions that border each other. These regions are organized as a graph, as shown in the right-most image below, which can be used in search algorithms such as A* or Breadth-First to very quickly find ideal paths from one region to any other region. For example, the path that results of a Breadth-First search from A to L would probably be {A, B, C, D, O, F, L}. The whole idea here is that we've taken the very complex problem of moving a character across a complex level, and we've reduced it to the generally very simple problem of moving a character in a straight line between two touching convex regions. However, the main problem with navigation meshes is that they're hard to construct. You'll usually build the polgyons manually in a 3D editor, or generate the mesh automaticaly from the game models themselves, and then you still have to build the graph. GRAPH, thats key And thats what grid generator does in tutorial i have exposed, generating graph with nodes over nav mesh instead just convex regions border each other. Why? Because grid generator output will be not just walkable areas but also NON walkable areas obscured by level geometry loaded later and also eventually updated in runtime. All those data are ready to use for search algo with no checking but result path..its fast..very very fast..so, it is actually navmesh, with more data filled in to it, rather than just consider convex regions. Most important, Convex regions connected to each other, may be VERY messy because u cant get always nicely balanced triangulation on custom geometry, there is NO WAY, without loosing edges. and your convex regions in navmesh depend on it (means navigation itself), on other side, my grid generator will 'equalize' entire navmesh area and deliver balanced graph with precalculated data accurately set for exact nav mesh, necessary for very fast search(what raw navmesh convex regions use too anyway).. P.S. Grid generator also doesnt require any additional work once initiated(only 1 code line) while for raw navmesh you have to manually set some quads. And i still do believe, pivots(grid generator using it) are faster than quads, so performance cant be worse with expansion..

..grid generator exposed here works almost identically like navmeshes, only difference is nodes generation itself based on navmesh itself(navmesh use its edges / vertices, while im using nicely balanced nodes)

..hi guys..in some near future, Hidden Dawn will enter its final circle of development for PC, therefore, it will be time for some testing on various PC configurations, and I will be in need for some additional testing, i cant do on my own..having said that, I would like to know who eventually will be interested in such thing, provided this: -full name, address, contact phone will be required (and checked) -NDA under Publisher specific conditions -no payment will be made for such work, however, name will be exposed nicely in game credit list related to task given and free copy of game will be delivered This is something i am thinking about, since i am unable to do this on my own on such wide range of hardware PC has, and i am looking for people who are more in to serious use of LE rather than hobbyists. Eventually I will also contact Josh for recommendation over folks I could work with(if he has any) and rest will be my personal opinion . Final word over this will be, of course publisher so it may happen that even if i want to do testing here, they will stop me and i have to let them know about this, so this is not final decision, but i would like to have list of people interested to do beta2 testing for me, under such conditions i have listed. Thank you all in advance guys.

..hhehe...well..I think Josh had some T-shirts and coffe mugs I dont know whats happen now I hope it they re appear, Tshirt will be something usable actually

hi guys...these lessons using CPU only...however, similar algo i have plugged in to GPU trough OpenCL, so in my game actual thing is executed on GPU(calculation and path returned) and its blowing away comparisons i have mentioned here..its just supercomputing fast ..but this lessons Im presenting here, are VERY fast having in mind its CPU..last night i have added new grouping algo over grid so A-star will look for bigger searcheable blocks and cut off what is not important(means cut off larger portions of nodes), so speed up is insane, and its still main CPU..now on my 7600GT rig im getting under 1 ms over 35000 nodes..with such speed is entirely usable in games with no problems at all..eventuall after i release lessons, if people likes it, i may consider lesson of porting thing for GPU execution..will see..T-Shirt+coffe

..tested just now on 8600GT rig, 30 000 nodes..cant get any value..it appears to be under 1 ms, whats excellent

..i have tested on my 7600GT low end test rig (5 years old machine), 20 000 nodes and returned path took about 2.1ms

..well..actually its related to something called gimbal lock thats why such definitions are made..ill make small normalizing function and post it back, so u can use it.

..okay, recently i have noted you folks about navigation grid generator used for path search..so today i have finished optimized A-Star algo, using that grid for search. Algo can be easy switch to behave as Djikstra if you want that. Its incredibly fast thanks to navigation grid because loads of parenting and calculations are done during grid generation and stored as a data already so path search algo doesn't waste time on unnecessary calculus..here are some small shots, both in action and connected...so now just text/description left to do.. ..once tutorial description(pdf text) done, somebody owe me coffe and LE T-Shirt

..well..you can always make small method/function to normalize such thing and you are set..provide input in range 0-360 and normalize that in range exposure 0-180<->-180-0..

..Lumooja, PS3 is way over low-end PC..that thing has capability of supercomputing speed, more powerful than brand new PC's...its GPU is yes, weaker than new PC GPU's new days (PS3 has modified nVidia 7xxx card model)..its about use..if you rely only on GPU to do things like you do with LE on PC, then yes, you are right, but keep in mind that such use of PS3 is WRONG, because you have sitting 6 cores capable of doing just about anything, from postprocessing tasks up to common things like loading meshes and textures..not to mention that for instance, AI of any sort will be faster than any PC newdays because its executed on SPU's too, instead of main CPU on PC's..PS3 deffered renderer (Phyre) is outstanding and support 120 real time lights at once (in sight) with no slowdown, and thats not a joke..PS3 is really..different..

..well..music composer who did all soundtracks for Dreamfall, doing it for Hidden Dawn too, just minor license issues regarding soundtracks need to be solved (publisher request again) so..we will see..game grown over my expectations with really a lot of characters spreaded over 40 levels..I cant say that story is that strong as DreamFall, but it has very unusual ending..I have plenty of stuff to show and so far i show only small bits because of very strict NDA..ETA depend now on publisher, after changes requested are done(if you remember it was some characters modifications, then whole set of frozen world is changed and its really looking crazy good now..ill post some screens soon)...i just hope they will not drag me until PS3 port is done, because i really wanna let it go out..I will eventually need some beta2 testers , so i was thinking to engage some folks from here but i dont know how smart is it regarding 'leaking' issues..i will see..

..90% of PC version is done..some design issues(according to publisher) are taken back, as well as some voiceovers(35 minutes) since language are not correct according to them(in sense of expressions, not some bad words or something)..I can just say that most of 3rd person adventure games lovers (games like Dreamfall) will be very nicely surprised with visuals and storyline..right now im porting to PS3 and at same time fixing issues mentioned..i hope my lil game will make ya all proud to have LE

..im wondering , how your development rigs look like, so feel free to post it here .. since i have started thread, ill start with mine..here is my main development rig, i have few more but its just testing.. ..here it is..PS3 Tool unit..

..thanks Aggor..i appreciate your interest as well as other folks..i hope this will help people to enjoy Pathfinding, rather than having a nightmares..please, keep in mind that this lesson will cover: -Grid creation -Grid alignement on to custom level geometry -Grid hierarchy building (parent to each child node distance included too) -Grid formatting in order to get rid from unnecessary nodes so search can be much more efficient, while parent-childs structure must stay intact -Grid walk ability determination so we know exactly where we can walk or not. Grid walk ability can be updated in runtime but its not covered in this lesson and its not part of process grid generator should do, because its expensive to do such thing (grid i show on pics has 10 000 nodes at start, after formatting end up around 2700)..however, grid walk ability will be updated very quickly with algo explained in lesson 2, so this one is JUST and ONLY creation of nice grid area suitable and ready to use for pretty much any search algo worth salt..lesson2 is search algos and grid updates...

..physics can determine walkable node as non walkable, eg. piece of rock fall down and block path..however, its expensive to keep refresh whole node structure to check that so, such update happen only if result path contain node what is blocked so it will be reported back to grid as a non walkable...that algo is part of lesson 2..

..obstacle between parent-children are calculated too so what left is only walkable points. Lesson 2 will cover search algos over this grid generation. Auto grid just need to generate set of waynodes, build hierarchy between them, store distance values between each parent and its children and clean up grid from unnecessary nodes while leaving only walkable nodes and those non walkable because of obstructions based on game level..

..these are just tests shots over tutorial lesson 1 related to automatic navigation grid generation..so i just wanted to test thing over some random geometry and see how is it work, align, linking child-parent nodes, etc..and so far is very nice..hopefully tomorrow will release whole lesson.. spiral shape geometry quarter of torus full torus and hole nicely ignored while structure of parent-child is intact terrain pyramids..i know useles, but it was nice to see parenting and all variables very nicely done even on such ugly geometry.. i hope you like it, im happy to see it working over pretty much anything

..hey Macklebee, im not sure have you received my PM. Tomorrow Grid lesson is done, so i wanted to know, is it english your native language, because i need someone to take a look over lesson text and correct my spelling errors, send me back so i can pack everuthing with images and code in to lesson..

..another thing is, that almost nobody here, want to produce simple game actually, and instead, want to make a 'Crysis 10' since LE renderer is nice..and thats why 99% of stuff just fall in to trash bin of fail experiments..

..i hope engine will not grow in to unity/shiva form of engine, just for sake of hobbyists satisfaction..biggest benefit for me is ability to have freedom in a way to code i want or to create tools suit my own needs..

..i have just finished reading all reviews from Devmaster, related to LE..i mean..i just dont understand, why some people writing so wrong things about something, that they obviously dont know?? I mean, look at guy 'Chillingham' over there .. i was really objective and trying to understand source of his informations, but i really couldnt find any sane or proper source of informations what will guide me to make such statements as he did..he claim engine is poor?? I mean, what does it mean poor? I see it here, running on 7600GT at 25 FPS and chewing 250K polys with 38 characters moving around..is that guy ever tried Bmax to claim such really wrong statements, i mean, only thingy i saw faster on tests performed, than Bmax was C++ and difference was minor..what is his base for claim that there is no commercial games in dev..heck, i know 3 guys here working on commercial game, and me myself, running final cycle on hidden Dawn and already designing new game(3d pnc adventure)..im really wondering, why people writing such nonsense ? is it personal or something?? I just dont understand..im wondering what you think about it folks, but be objective, really..

.it is announced by sony that PS4 will be entirely compatible with current one (few changes were involved in to GPU and some for sake of backward compatibility), so basically, everything you develop for PS3 will work just fine on PS4..Sony announced also upcoming editor for PS3/Phyre, really neat looking one, similar to Unreal...as for Xbox..they are not moving anywhere soon too..they did announced that in development is an X-Engine, designed for Xbox360(it seems that PS3 nailed it nicely with Phyre/Editor), so M$ doing pretty much same thing now for Xbox360.. @Rekindled Phoenix ..method to port for PSP and Xbox360 is basically Phyre...by the end of this year Phyre will natively run on PSP as a one of rendering targets, and Xbox 360 port is also matter of 'rescaling' Phyre's use of SPU's in favor of more shader pipelines(Xbox 360 has 48 Unified Shader Pipelines, while PS3 has 24 pixel pipelines, and 8 vertex pipelines). PSP Translator may be interested for folks here, since, people could easy run that on their PSP's they have, their applications..but we will see how it goes..

hello HeadClot, and welcome to LE world ..ill try to answer your questions, from the perspective i see 1) You will do just fine with LE, if you take one step at the time, regarding learning curve, so, there are some tutorials, and most important, people here are very friendly so you will not be alone, no matter where you stuck..code is not heavy, but keep in mind, it depend on complexity of your project and what you trying to do..whatever case may be, some learning curve is necessary, but dont worry about it, you will enjoy it a lot 2) Blender is just fine, long as you can export FBX..3dsmax, Maya, XSI, etc..you can use pretty much everything I hope this helps a bit..I highly recommend LE..you will not regret it(long as your scope of development is PC only, at this moment)

..my low end test rig is 7600GT and everything i set, works just fine on it..it helped me a lot to carefully design things and code so everything more than 7600 simply just fly..