paramecij

Nicely done. I'm basically doing the same thing but with a Kinect sensor. I wasn't aware of the Asus Xtion, just googled it up... seems to be quite similar in price and functionality, or are there any major differences between the two? I've previously used a Wiimote inside LE2 (heavily inspired by stuff Johnny Lee did), but have now switched to using a Kinect for motion sensing.. The sky is the limit with this, indeed - but still, I've found out nothing beats keyboard & mouse in speed, accuracy and most importantly, long term use. I can't imagine myself playing long sessions while having to move around a room or waiving my hand inside the editor for an hour or two. A good example of where this could be used to nicely augment gameplay and immersion would be a top down game like GTA1/2. I remember when I was younger and played the first two GTAs, I often caught myself leaning my head to the sides of the monitor trying to see behind buildings. I guess it was a natural reaction I've since unlearned because I know better. Oh and If anyone is lucky enough to have a digital projector, google jDome, and if you're half a craftsman you can build your own for as little as free if you have suitable materials at home.

That's also my experience with .x files. I've found the best intermediary format to be .SMD (halflife2) for importing it into blender/3dmax/UU3D/etc before converting to gmf

To prevent fast moving objects from missing collisions, use: void SweptCollision(TBody body, int mode=1) But too fast moving objects can still fall through, so you might want to clamp the velocity in the update callback

File Name: crosshair pack1 File Submitter: paramecij File Submitted: 17 Oct 2011 File Category: Materials Resolution: 256x256 Here's the first part of my crosshair collection, it contains 64 64x64 crosshair images you can use in your games.. they are all solid white with an alpha channel SetBlend( BLEND_ALPHA ); // must enable this SetColor( 1.0f, 0.0f, 0.0f, 0.8f ); // color the crosshair red, and lower the opacity a bit DrawImage( crosshair, GraphicsWidth()/2 - 32, GraphicsHeight()/2 - 32 ); WARNING: may contain some offensive material B) Click here to download this file

Because you are using a pointer - PPROCESS_MEMORY_COUNTERS - the first P stands for Pointer (general ms naming convention), if you look-up the definition of it you'll probably see it's defined as a pointer *PPROCES_....

Using newton physics with LE: Using more textures on single surface: just use texture atlases, and sacrifice one channel of vertex color data as an offset into the atlas

I'm currently experimenting with such a "tool" I've made for LE2, which allows me to generate/modify LE terrain(not custom mesh) in code or editor, it's pretty easy to make. If you want to give this a try yourself, there's a lot of good material on this just google perlin noise to get started, or PM me if you have specific questions.. I could make a mini tutorial on this or I was thinking about releasing the "tool" once I update with some improvements and more features.. (see here ) It might be better to use tools like L3DT or EarthSculptor to generete a much more realistic terrain, and import that into LE (or modelling apps) .. you can also make good terrain meshes in blender/mudbox/3dcoat/etc (with caves, over-hangings, cliffs, etc)..

If you are trying to portray an historically accurate world, this is the correct physics model to use! Too bad they didn't have computers back then so they could put those theories into practice ... I would suggest methods already mentioned, but consider that you don't want to just stop the player cold, that's not fun especially if you are moving fast like in a vehicle. It's better if you can do it in a nice rubber kind of way instead of an invisible brick wall. Warn and slowly stop or turn the player back to the playing field, by checking if position is out off 'soft' bounds and throttling or clamping movement values on appropriate axes, then have a physics body for the outermost 'hard' bounds, that catches all other objects or those that break through the 'soft' limits. I prefer more visual and realistic restraints like impassable terrain because then I can expect it's probably the end of map, instead of going very far out into the desert only to find an invisible wall at some point.

I also would suggest just fitting the trunk of the character. If you go with just one body, make it fit the average pose volume of the characters body as tightly as possible, otherwise you'll get collisions when you shouldn't, doors will be problematic to go through, walking on slopes will look weird etc etc.. What you can do next, is a body for each limb or attachment (think rag-doll), if you just move these bodies according to bones, you can get more detailed collision data for you to act properly (and that's the tricky part), or if you just move the bones according to bodies you can get some nice rag-doll death animations, if you also throw in some IK system you can then take this to the next level.. Or if you just don't wan't the sword to go through walls etc, you can make a separate body for the sword, make it follow the swords movement, and when it collides, stop the current animation and blend into some kind of bounce back animation (with sound and particle effects). You can also render the sword on top of everything to give this illusion (I use this for my FPS guns and it works well, because they don't extend out much, for swords this might not be such a good option) If you want to perform body part hit detection using raycasts, instead of bodies you can use invisible low poly meshes.

I think he already did. Check out the switch model that came with the SDK, namely switch.frag and switch.lua

I don't know where to find good quality heightmaps (apart from GIS, DEM, .. data), I only know of tools that can make them (L3DT, World Machine2, GeoGen), but usually there's a learning curve and you have to invest some time for good results. How about an editor object that allows you to change some parameters, click apply and it auto-generates some terrain for you, or just does some noise filters over your existing terrain? ( I'm doing this in c++ now, not finished but already quite usable, it think it could be ported to LUA easily, see here)

Those platforms are all fine and dandy, but the real question is, will it be able to run on my Raspberry Pi?? ( http://www.raspberrypi.org )

1.) 3D Tools / modifiers / scene / scale tip: First I usually go in "Info / Scene" menu and check the 3D bounding box values (like height or any other dimension I know roughly how much it should be) and then I calculate the scale factor (or you can just see it's gonna be inches to meters conversion etc..) 2.A) Try checking out the scripts if they have something meaningful about animation frames in them.. 2.B ) Animation / Editor / If you click on Add button, then select the option to extract from existing animation and input the range of frames you wish. This creates another animation track/layer/whatever that you can then play in a loop ... also feel free to explore other options there, they'll come in handy

Thank you very much, I have some duplicated files now to get around this..

I thought so too, but it turns out CountChildren() returns just the number of immediate children of the bone (not the whole hierarchy) .. my models hierarchy is: Bip01 | Bip01_Pelvis | Bip01_Spine | Bip01_Spine1 | Bip01_L_Thigh | Bip01_R_Thigh ... CountChildren( FindChild( model,"Bip01" )); // returns 1 CountChildren( FindChild( model,"Bip01_Spine" )); // returns 3 CountChildren( GetChild( GetChild( FindChild( model,"Bip01" ), 1), 1)); // returns 3 (same as above)

That's true, except the editor will look for files ending with *dot3.dds to automatically load normal maps for terrain textures. Also the genmat.exe uses this convention...

Very nice! Incidentally, I have quite a collection of photo based textures for vegetation, that are looking for nice models to be put on ..maybe you could find them useful, or would you rather do your own?

I see trees these days are using 3000-5000 triangles for the main model, other small vegetation much less - as it's generally used in larger numbers.. If you mean that the plane is visible from both sides, just put this in your .mat file cullface=0 ..Can we perhaps see some work in progress pics already?

Obviously, as low as possible.. I can say 1000 polys for LOD1, but if you can make it look equally good with only 500, well... And the more LOD levels you have, the more performance scalable the vegetation is gonna be (ie. using lod levels 1-3 for high-end, 3-6 for low end systems). This totally depends on the model, but I usually half the poly count in each next LOD level (or even more). You can also use a different material for each LOD model. This way you can use simpler shaders, and maybe a smaller and more dense branch texture (more leafs to compensate for less geometry). It pays to have the vegetation as optimized as possible, since it's mostly used in very large numbers, the small savings add up...

..of which other 'guns' are you aware off?

You seem to *really* like Left4Dead. ... a lot! Anyway, I think this will give quite an edge to beginners and save a lot of time to more experienced users.. Nice work! Some questions though. Will it work seamlessly with terrain and custom geometry (ie. buildings that can be entered and are placed on terrain, a cave mesh that's placed to a terrain 'hole', etc). Do you plan to supplement this with a traditional waypoint system, so you could connect two adjacent nav meshes (two towns that are far apart), or would we need to roll-out our own? ... This is already usable for numerous types of games even in it's current form, but since you're tightly integrating it with the engine, I think it would make sense to really flesh it out and make the most of it...

What if I target an destination that is offset from player position (it would need to be determined if it's legal and sensible), then it would look like it's constantly trying to flank the player?

I don't know if I understand you correctly but you might try this.. DrawImage( texture, 0, HEIGHT, WIDTH, -HEIGHT);

You can get same or even better results if you just "paint" the flame by hand.. it's not hard to do - just an oval shape or two filled with a nice gradient.. And a couple of frames of animation isn't a lot of work either. Here's an example of what can be made in less than a minute.. But if you want more smooth animation (and more realistic movement) then I think it's best to use procedural generated flame, there are numerous texture generators that can do this, or even more specialized tools (strictly for flame effects or particles in general).. You can also generate some nice effects (with particles or other plug-ins) in your favorite modeling package (3dmax, maya, etc..), pull the camera way back, narrow the FOV, and render out frames with perfect alpha channels... Capturing real fire and masking it out can be tricky, best bet is to shoot it against a black background I would imagine, or one can just use one of the many stock video options (there's also some free ones). Don't do this, my friend damaged his camera this way... You can just zoom in on the flame instead - it's better because you get more DOF and a more flattened perspective (both of which are good for textures) -- EDIT -- Just for fun I also tried making one from photos.. with black it's easy to mask out, but the more transparent blueish tint is hard to preserve..

It's more of an hardware thing... http://www.microsoft.com/appliedsciences/AntiGhostingExplained.mspx