Rick

I would be making this a texture that I put on a plane and draw in the transparent world. Then I rotate the plane which gives a similar result. This is for 3D unit selection for my game. I don't need a GIF. I'm more interested in how to make the 2D image of a ring like the one I've posted.

So how did you do something like this in Gimp? I only need a static image as I'll use it as a texture to a plane and rotate the plane to give the "animation". Also it seems like we aren't on the same page. I'm specifically interested in a ring that has this style where it starts dark at one spot in the circle and gets lighter/more transparent as it goes around. So not sure what you are talking about with 3D renders and 2D renders. I just want a 2D image that has a ring like this on it.

That first video doesn't really look all good or real. I would think destroying CSG solids would be able to produce some pretty realistic results on predefined blast types. A bullet or dynamite could probably be reproduced fairly well when done against say a wall, ground, or pillar. Adding the idea of physics to where things would collapse I would think would be difficult. Putting holes or blasting out chunks would be about setting tolerances and forces for the CSG solids and then applying some carve out logic with that given data to reflect what an impact defined by some force would look like against the "toughness" of the CSG solid (doable for someone like Josh who has experience with CSG). When the force hits, run this logic against it, make some particles and generated debris and blast it out with a force. But doing that blast all around a small concrete hut and having the hut eventually cave in under the weight I would think would be difficult to automatically do I would think. Would PC's be able to generate this kind of thing RT fast enough? Everything from PhysX looks like it breaks apart as wood. Even the concrete walls in one of the videos.

http://dl.dropbox.com/u/1293842/Circle.jpg The above image is the idea. Doesn't anyone know how to get this technique with paint.net? I want to use this idea for my unit selection ring but have a hard time describing it to google for how to do it. It starts out solid and dark and gradually gets lighter in color or more transparent. Looking for some technique that will create a "perfect" ring and tone of colors and transparency value because doing it by hand is most likely not on option for me as I'd butcher it.

Won't the shadows pretty much still have to be baked though? Isn't that a limitation of the hardware?

Another small update on GvB. Got stealth working and added a selection ring. I did some code cleanup and separation also as most of the code was all in 1 lua file and was becoming a pain to manage.

They should return to their original state once you stop.

One way to do this is shown in my Pi-Main http://www.leadwerks.com/werkspace/files/file/72-pi-main-v01/ This basically creates a variable in the main script that you can check against in your model scripts. I'm sure there are other ways to do this also. Perhaps checking if fw is nil or not as I seem to recall that variable doesn't get created until the main game script runs? But I do agree it would have been nice if there were 2 separate Updates() that we could work with or something along that line to separate from design mode vs run-time mode.

Flexman, do you own 3D World Studio because it seems like it'll be just like that for the most part. There are prefabs in that also. If you haven't used that I would suggest getting it. Even the free download would help get a leg up on what it's going to be like in the new LE3 editor.

I'll gladly use other people's code as long as it fits nice with my code. The pathfinding script I found for Lua is a perfect example of the kind of plug-n-play stuff that works well. But does that game project still require some kind of compiler? I think that was Josh's point. The editor is the compiler for Lua programs so no external tools required.

Yeah, you can do that also. Just note you'll have to tell the AI scripts who the "player" is. If you look here -> http://www.leadwerks.com/werkspace/files/category/6-scripts/ and look for things that start with Pi-, this is the route I was taking some time back. It was all script attachment based with one variable I think in the main script to tell if we were in design mode or gameplay mode since everything you put in Update() will run no matter what mode you are in and sometimes you wouldn't want that. Like in your example you probably wouldn't want your AI running around when in design mode. To let the AI scripts "know" who the player is you could use the link system (although that might get to be a pain to have to link the player to each enemy) or you could create a custom property that holds the name of the player object you have. You can see some examples I've done in the past with Pi- scripts where a small little demo is entirely done in model scripts (or what I call Thingoids)

About your 3rd question here is what I've been doing. 1) Create a model that represents a character (or whatever) 2) Create a lua script for that model and in that scrip add variables specific to it: function class:CreateObject(model) local object=self.super:CreateObject(model) object.health = 100 3) Then in my main lua game file is where I create these models (although you could place them at design time also). I then either loop through a list I've stored the models in or if I pick with the mouse the entity to get the model which then you can use: objecttable[model].health = 20 So once you get a variable pointing to a model you can use it as a key to the objecttable table to get those model specific variables.

I've had this idea for this game for many years. I remember about 2 years ago I played around with that idea. I had an archer class that could place an ice trap on a tile and when an enemy would step on that tile it would freeze them in place. I might do something similar but as a different ability since I don't have an archer I have stealth working now for the assassin, which I think is pretty sweet!

They are just 2 very different things with different skills required. Understanding how networking works is generally not that hard (if you are technical to start with), just like understanding how modeling works isn't hard either (again if you are technical to start with). Actually doing the 2 though is when things can get tricky. When coding networking for a game a good portion of your code (pending you are using a library to do all the socket stuff) is coming up with "tricks" to mask lag. If you are coming from more of a hardware perspective you probably do things to make sure the network is working in top shape. However there will always be lag and you probably just accept that for the most part and tell people there isn't much you can do about it so they have to wait for 2 seconds to open a file on this shared network drive. From a gaming perspective you have to do something about it as waiting even 250ms is often not acceptable. So you have to code smoke and mirrors to give the illusion that there isn't any wait time. That's really the biggest part about making network games. That and security as that can get pretty deep. Figuring out how to send data back and forth is generally the easy part. I think normally when people start with networking games they go through some stages. You can send player input to the server and have both client and server run the simulation. This will lead to characters being out of sync. You then realize you have to find a way to keep them in sync, which normally means having the server (since it's the "boss") send back corrected locations to the client. You then realize you can't just move the client to these corrected values because it'll "jump" around on the screen so instead you have to smooth out the corrections over multiple frames on the client side. You also realize you can't send keystrokes to the server and wait for the server to come back with where you are in the game world because that lag is to much (locally when you test it'll be fine because your ping will be like 10ms or so). I think I read anywhere over 33ms and humans can notice the lag from key press to seeing the action being done on their screen. So now you have to send the command off to the server but also do the command locally so as to avoid this input lag. Then if you get into shooting or any sorts that opens up another can of worms because by the time you press your shoot button the server's version of the world is different than yours. Some common ways to handle this is storing off snapshots of the world for the last x ms on the server so when the client sends a shoot msg you can "rewind" the server world to see where everything was at the time (by using the ping time of the sent command as well) to see if the client hit something. If you have ever played an online RTS some will mask lag by running an animation locally of the unit you said to do something. The animation time is to mask your command being sent to the server and waiting for the reply. AOE did this. When you would give a command your little guy would salute and say "yes sir" which took like 1/2 a second and it was there to mask lag. So just some info on what to expect when trying to make a network game.

My suggestion would be not to start with a networked game. It adds complexity which when you are just starting out is hard to deal with. My suggestion would be to create a very basic single player game first so you get the ideas behind many concepts that go into making games before diving into networking.

I didn't yet. Between getting stealth working for my game and hanging with the family I didn't have enough time this weekend. I also wanted to use my board but scale it and still have the tiles the same size for the test, but with my new Win 7 I don't have my 3DWS key anymore and that's what I used for the board and texture scaling. I have an email in with Josh to see if he could get me that but haven't heard back from him yet.

Your AI logic can be anywhere. C++, Lua, BMax, whatever. There is no right/wrong way to handle it sorta. Generally with complex AI you'd want the fastest means to run it so Lua generally won't be the best choice but if your AI is simple enough then Lua is fine. Don't want to get into another language war but in general that's the idea. I would say if you are starting out just use whatever language you feel the most comfortable with.

Man, I'm pretty sure I have a mild case of dyslexia It seems to work now, thanks.

Can't seem to get this working. Here is the mat file I'm using with the above code. The result looks just normal. The model doesn't look transparent at all. texture0="Orc_D2.dds" texture1="Orc_N2.dds" alpha=blend shader="abstract::mesh_diffuse_bumpmap_skin.vert","abstract::mesh_diffuse_bumpmap_specular.frag" shadowshader="abstract::mesh_shadow_skin.vert" I'm saving the texture as DXT5(Alpha) so guessing it has that Alpha layer. Has anyone done something like this before with the framework and if so if they could share all the details about how they got it to work that would help.

So my actual texture has to be semi-transparent as well? That can't be controlled in code via some alpha value to EntityColor() or anything?

So you can't use the same material then as what you would use in the main world for the model huh? That sucks. What should the material look like for this to work?

I've never actually been able to get semi-transparent models to work in LE. What exactly is required to get a model to be semi-transparent? 1) Does anything in the material or texture need to be different from a model that I would draw in the main world? 2) How do you set the level of transparency to this model? 3) Is the level of transparency specific to each model or the entire transparent world? What I'm currently doing is the following but the model has 0 transparency. It looks normal. function object:LoadTransparentModel() -- Change to transparent layer SetWorld(fw.transparency.world) -- Load assassin model that will be transparent object.transparentModel = LoadModel("abstract::fantasy_bad_assassin.gmf") object.transparentModel:SetPosition(object.model:GetPosition(1), 1) object.model:Hide() EntityColor(object.transparentModel, Vec4(1, 1, 1, .3)) -- Go back into main layer SetWorld(fw.main.world) end

That looks cool! Yeah that's sort of the idea I'm going for except more abilities per piece, less move restrictions, and "real-time" instead of turn based. I want/will bring internet play to this also and actually would love to make it pure network play because then I wouldn't have to make AI, but I suppose people prefer playing against AI at first to learn the game. If Josh is able to get networking in LE3 and have it working in Lua and multiple platforms that would be perfect. If he doesn't then I want to branch the code off and find a Lua networking library for Windows, at least, and see if I can get it on Steam and integrate it with steam id's because I'd love to have global rankings of players based on their steam id. I'm picturing having a Play Random button that gets you in the game against a live player quickly and handles rankings. Actually those screenshots are showing me that I should make some different arenas instead of having just the black background.

Thanks Pixel. Hoping to get the abilities in somewhat working condition this weekend and then it's on to the AI which I'm sure is going to be tough. Not really even sure where to begin with that in this type of game. I guess I'll have to pick a strategy for the AI to follow and try not to make it that obvious to the player somehow

A short video showing some updates: