Leadwerks Dev Blog

Testing Werkspace/Blogspot integration

2010-03-01T10:58:00.000-08:00

Testing...

Blog Moved

2009-11-22T13:02:00.000-08:00

This blog has moved to this site.

2009-11-19T13:40:00.001-08:00

Leadwerks Engine 2.3 will be released Monday at noon Pacific time.

Lua Script in Leadwerks Engine 2.3

2009-11-16T15:40:00.000-08:00

2009-11-16T09:04:00.000-08:00

Because ideas on their own count for so little, publishers want games, not ideas. Ideas are cheap, a dime a dozen; everyone in the game industry has ideas. Recognize that your "great idea" is probably not that great, not that original, and not that interesting to others. Virtually everyone thinks their game ideas are extraordinarily good, and everyone is wrong almost all the time.

This is hard for beginners to accept, partly because it’s easy to come up with a few ideas, so it’s nice to think that you only need to come up with a great idea to make a lot of money. No, there’s a lot of work in making a successful game, beginning with generating LOTS of ideas. The more ideas you have, the more likely you’ll have a few really good ones that can become really good games.

There’s a “pyramid” of game design (see illustration) that goes like this:
• Lots of people get ideas
• Fewer successfully go from general idea to a specific game idea
• Fewer yet produce a prototype
• Fewer yet produce a decently playable prototype
• Very few produce a completely designed game
• And very, very few produce a really good complete game

Read more.

2009-11-13T17:16:00.000-08:00

Some time ago I commissioned the creation of a character to serve as a sort of mascot, and to use for whatever I liked in Leadwerks Engine. Here he is:

I'm not sure exactly what I will do with him yet, but I think it looks fantastic. You'll probably see him around more in the future.

Model was made by Ilya Kachura.

2009-11-05T21:16:00.000-08:00

Check out this guy's blog.

2009-11-02T09:40:00.000-08:00

Disclaimer: I tried Ubuntu once and hated it. I don't think Linux will ever become popular, because the developers won't take responsibility for it, and push problems off on the user.

That said, Linux looks more like the Windows I like than Windows does.

2009-11-01T22:53:00.000-08:00

Nice video here:

Windows 7

2009-11-01T13:08:00.001-08:00

This is my opinion of Windows 7. When I tried Vista, I immediately disliked it, but used it for eight months. At the end of that period I still disliked it for all the same reasons.

Nerd discovers the outside world; marvels at graphical quality

2009-10-29T11:43:00.000-07:00

2009-10-29T11:00:00.000-07:00

There's some interesting information on scripting available in the wiki.

2009-10-25T18:42:00.000-07:00

Well, the roads are completely done, and are identical to those in Crysis. This was not something I originally planned on including in 2.3, but you will get it anyways.

2009-10-24T10:04:00.000-07:00

2009-10-23T18:36:00.000-07:00

This is what I was talking about. I just got the texture mapping working. (That was hard!)

Here you can see the road is split along the terrain edges, and aligns PERFECTLY to the terrain, as if it was cut right out of the terrain. There is no need to have a slightly raised surface, and there is never any z-fighting.

This was written entirely in Lua (except for the surface slice routine)! The script updates one segment at a time each loop, so you can edit without waiting for the road to build. It's pretty fast, but a little too slow to update completely every time you move a road. With the staggered updating, it takes 1-2 seconds to build roads. They took this same approach in Crysis Sandbox.

Here you can see the texture mapping. The UV mapping has to be performed after the mesh is constructed. The mesh in the shot above originally had no UV coordinates, and I had to figure out how to map them! I had a few failed attempts, but finally figured out how. There's no alpha blending on the edges at the moment, but I assure you the road here is completely flush with the terrain.

You can read about one of the problems I ran into and solved here.

2009-10-23T08:16:00.001-07:00

I am unsatisfied with how roads look and act, and must devote more time to make them perfect.

2009-10-22T12:56:00.001-07:00

Get your credit card out

2009-10-21T14:30:00.000-07:00

...and head over to www.dexsoft-games.com to buy the Northern vegetation pack. I don't care if you don't even need it, just give them money so they will continue making incredible media. This is the artistic savior independent developers have been waiting for.

http://forum.leadwerks.com/viewtopic.php?f=32&t=5509

2009-10-18T20:36:00.001-07:00

Here's a really cool video. This is running in Leadwerks Editor.

2009-10-14T07:30:00.000-07:00

I tried my own vehicle implementation yesterday, and the results are really good so far. This will allow more control over the vehicles, and we won't be forced to wait for any third-party solutions.

2009-10-09T23:55:00.001-07:00

I want to release this November 1. It might be late, but if it is, that's okay.

Logic Problem

2009-10-05T12:18:00.000-07:00

I am working out the advanced logic and interaction by applying it to different problems. Messaging the way the original Quake engine was set up is pretty straightforward. Half-Life 2's input-output system is interesting, but requires an additional mechanism beyond the existing keyvalues system. Lua allows some really powerful stuff. You could enter a string in the property editor, and the script itself could execute this string! But if it's confusing for me, it's going to be really confusing for an artist.

It's easy to make a button open a door:

-Set the door to be the target of the button.
-When the button is pressed send a message "Activate" to the door.
-The door receives this message and interprets it to mean that it should open.

How would you do the following in your favorite game engine, without scripting?:

Buttons 1, 2, and 3 can be switched on and off. Only when buttons 1, 2, and 3 are all turned on, a door opens.

Your ideas?

OpenGL 3

2009-10-04T14:26:00.000-07:00

I finally found a good reason to upgrade the renderer to OpenGL3. This will allows MSAA with deferred lighting. It might possibly also yield some performance gains, but that is speculation. This will occur some time in 2010.

2009-10-04T10:14:00.000-07:00

Here is some information on ATI's new cards. Benchmarks reveal they are twice as fast as the GEForce 285! And the GEForce 285 is much faster than the hardware I originally designed the engine for, the GEForce 8 series. I need to get my hands on this stuff so I can see where the bottlenecks are. I don't have to require an ATI 5870, but it would be a good idea to design things around that so that the engine scales with the hardware and can do more.

NVidia seems to be focusing on computational stuff. I'm really not sure how I feel about buying a GPU that's designed to run C++ code. I am sure there are some competing variables that would cause a tradeoff when you are designing hardware for graphics versus general computation. I like NVidia hardware, but are they losing focus?

S.T.A.L.K.E.R. learns from Leadwerks

2009-10-03T12:11:00.000-07:00

GSC Game World, 2009. See slide 29.
Leadwerks 2008. See page 4.

I've picked up quite a few ideas from them, too!

2009-10-02T10:36:00.001-07:00

Uniform buffers fixed?

2009-09-29T13:02:00.000-07:00

It will be interesting to see how uniform buffers perform versus a float texture. NVidia says the bug I reported is fixed for the next drivers.

2009-09-29T09:20:00.000-07:00

This is just creepy:

Awesome new gmod movie

2009-09-28T21:05:00.000-07:00

2009-09-22T13:48:00.001-07:00

2009-09-21T17:17:00.000-07:00

Tyler and I have been experimenting with bloom and iris adjustment. HDR does make a small difference, but it's not really an effect in and of itself. It's just a buffer format, and it can preserve the bloom color better so it doesn't wash out to pure white. Basically, I wanted really dark shadows for outdoor areas, and really bright sky coming in windows. I think this is that indistinguishable quality that makes Crysis and Far Cry 2 look real.

This is pretty interesting stuff:

2009-09-19T20:48:00.000-07:00

Underwater caustics!:

2009-09-19T09:24:00.000-07:00

I replaced the water texture with an animated normal map. Sync and copy framewerk.mod. The tool I used to generate the normal map can be found here. Thanks to Erik for pointing this out to me.

2009-09-18T22:57:00.001-07:00

Amazing!

Cuda Fails

2009-09-17T10:24:00.000-07:00

This story itself isn't news, but it is consistent with the feedback I have received from other developers. My prediction? All these massively parallel processing models are going to fail. The GPU is the only parallel processor in history that has been a success, and that is largely because the programmer doesn't know it is a parallel processor.

2009-09-16T12:36:00.000-07:00

This is interesting:

Float texture
Render time=7
Entity draw time=2.3
FPS=108

Uniform buffer
Render time = 3
Entity draw time = 0.14
FPS = 104.7

Non-instanced
Render time = 5
Entity draw time = 1.9
FPS = 81

Uniform array
Render time = 4
Entity draw time = 0.12
FPS = 112

There's a texture buffer method which would probably be best for uploading matrix data, but it is an OpenGL3 thing. My best estimate is this would yield a 3.7% performance improvement.

This is the first time I have run into any situation where OpenGL3 made a difference. I also found NVidia's uniform buffer extension broke in their recent drivers. I am no longer using this technique, but it would have been the fastest. In fact, the evaluation kit will display visual errors because it is using LE 2.22, which did rely on uniform buffers.

The best technique would be to use NVidia's bindless graphics, but it requires GEForce 9+ hardware, and I am not going to do that.

The result? The way I was already doing this is the best way until the engine requirements are raised to OpenGL 3. This matter is resolved.

2009-09-15T19:57:00.001-07:00

Gained a 10-15% performance improvement today with directional lights.

Also noticing in my test scene:

Render time = 7 millisecs.
Entity culling / prep time = 1 msecs (good!)
Terrain cull / prep time = 0.2 msecs (very good!)
Entity render time = 2.5 msecs...why?
Terrain render time = 0.4 msecs

So now to identify where the rest of the usage is. I thought the entity culling time would be the most expensive operation, but apparently not.

=====================

The culprit is the float texture I use to upload matrices for instanced rendering.

2009-09-13T20:51:00.001-07:00

2009-09-12T11:40:00.000-07:00

The networking API has been cleaned up and is ready to use. A separate netwerks.dll is provided so you can run it on a separate thread if you want. I also added an optional cubeface parameter to the SetColorBuffer() command, for rendering to cubemaps. No idea if it works yet or not, but the headers and documentation is updated, at least. Sync!

2009-09-12T00:16:00.000-07:00

Here's a first look at some waypoint AI. It was really fun to code this because I could just hit Ctrl+S and see the behavior immediately. Thanks to "Red Oktober" for the model:

2009-09-11T08:29:00.000-07:00

Fixed a small bug in the BlitzMax Framewerk module. Sync and copy files.

2009-09-07T22:48:00.000-07:00

Here's an image from before Crysis was released that I found very inspiring. I don't recognize this map from the game.

And here is another amazing image:

Posted here in the forum.

2009-09-07T10:34:00.001-07:00

This is that problem I was talking about yesterday with Lua compiled files.

2009-09-05T15:35:00.001-07:00

The module and DLL are updated for BlitzMax 1.34. Sync.

Ramblings

2009-09-04T21:58:00.000-07:00

My position is that publicly traded companies should be illegal. It shifts the focus of the business away from long-term sustainability towards short-term speculation, causing economic instability. It diffuses responsibility, as any decision can be justified for the good of hundreds or thousands of shareholders. A privately owned company is maintained for the purpose of long-term sustainability and the improvement of its products, and thus the good of the customer. A publicly traded company is a bunch of monkeys driving a bus off a cliff trying to make a fast buck and get out. Valve Corporation has my respect for their decision to remain private.

Incidentally, in today's uncertain economy, independent game development may actually be a wise career choice. Your business is worldwide, so you will be unaffected by whatever local turmoil your region may be experiencing. You can't be laid off or fired, and if you do a good job the sky is the limit, as far as your income goes. Not everyone has a skill they can make money with on a worldwide basis.

Economic recession can be a good thing. It allows the economy to restructure so that the firms producing real goods can prosper, and the unsustainable firms can be weeded out. Unfortunately, the last year of government actions have rewarded the unproductive firms (like General Motors) at the expense of the productive ones. Sure, people would lose jobs if GM ceased to exist, but it would open up the market for a new car company to take its place.

Want to fix the economy? Get rid of the income tax. They don't wanna fix the economy that bad.

Trees for Leadwerks Engine developers

2009-09-04T12:51:00.000-07:00

I've worked out a deal with Michael Bethke whereby Leadwerks Engine developers will get free use of an excellent pack of Crysis-quality trees and vegetation. The files are all in .max format, and I need a capable artist to export and prepare the files. This includes the following:
-Paint vertices for leaf sway.
-Rescale and adjust meshes.
-Delete double-sided leaf planes, if necessary.
-Export .gmf files for multiple LOD levels (LODs are already made).
-Create material files.
-Create .phy collision shapes.

The meshes are all merged together, and I don't know how to adjust them, so I will leave that to you. If anyone with the technical knowledge feels like taking this on, let me know. It will be a lot of work, but we'll get a great set of vegetation for everyone to use with the engine.

2009-09-02T19:45:00.000-07:00

I thought this image was oddly beautiful. It's by no means realistic, but I like it.

SDK Update Tool

2009-09-02T10:30:00.001-07:00

This was all Erik Nitschke's idea, and mostly his work, but I am really pleased with how it turned out. Run the updater to install the SDK once; I can then update single files, and you can update your SDK installation by running the updater again. Only the updated files will be downloaded. This will allow me to make small fixes to the engine more easily, and we won't have to download big installers anymore.

Character Artist

2009-08-30T20:23:00.000-07:00

I'm on the lookout for a character artist to make one animated skinned model. I am starting with AI. We will need something to work with, and it needs to look really professional. You might ask why bother, because it is just placeholder art. People will replace it with their own character when they make a game, right? Wrong.

This character will be used and reused in nearly every project, and will become a symbol of this engine. People want a playable game they can modify, and good-looking art assets are a part of that experience. Therefore it does matter what kind of media are used in the examples. That's what I think.

So if you have any artists in mind, or any concepts or ideas of what this should look like, speak up. I have the following conditions in mind:

He should function as both a protagonist and enemy, so no zombies or aliens.

He should hold and use a gun of some kind (for AI and testing bullets).

What do you think?

2009-08-30T14:42:00.000-07:00

I am running an experiment with Google Adwords on the forum, using a modification provided by Lithium Studios. It will be interesting to see the results after a few days.

2009-08-28T11:38:00.001-07:00

I picked up a small contracting job I will be busy with this weekend. Version 2.27 is nearly ready, and will be released next week.

2009-08-24T13:24:00.000-07:00

Here's a creepy gameplay idea: Make your enemies invisible, and only detectable by the shadows they cast.

2009-08-12T19:15:00.000-07:00

I was at Fry's Electronics and saw an ATI 4890 for $220. Curious to see the specs, I picked up the box and was astounded to see this card has 800 stream processors. Amazingly, I saw another card with an AGP interface that also had 800 stream processors and a gig of video memory, for only $100. AGP was introduced in 1997. Talk about upgradability!

On further investigation, it seems the GEForce 295 has a bandwidth speed of 223 gb/s, while the ATI 4890 is about half that.

I predict the number of stream processors will reach into the thousands soon, and will effectively make the number of polys/pixels irrelevant, even for expensive shaders. However, other constraints will come into play more and more, particularly texture bandwidth and CPU speeds.

2009-08-12T14:49:00.000-07:00

I did some facial morphing stuff with shaders and a model courtesy of Mike Rusby. The results didn't blow me away, but it works. I think the facial poses need more expression, because at this point it just looks like a few isolated vertices moving, rather than a human expression.

2009-08-10T07:14:00.000-07:00

2.27 will be out this week. The version 2.3 with Lua scripting and a new editor will take a while before it is ready. It's just a matter of using it to make things and discovering problems as I go. Be patient, and when it is ready it will be tested under many conditions and have lots of example scripts.

2009-08-07T17:59:00.000-07:00

Masterxilo posted a video showing off his implementation of ragdolls and terrain decals:

2009-08-07T14:13:00.000-07:00

Here's an interesting email I got from someone today:

Most users involved in the market are either artists or designers, they're not programmers. Honestly, i have never seen a programmer to start and organize a project. They're just hired as freelancers, or team members. The game itself, is in the artist/designer mind. I'm an artist myself, and i really suffer with the lack of solutions for the indie market, involving a decent asset pipeline, as well game logic construction.

Programmers are important. You can't make a game without them. But I don't think they are the driving creative force that makes a game. From my own experience, it can be very difficult to come up with a concept and make it into reality, especially when you are using low-quality filler media. I would much rather have a mod team come to me with some artwork and a concept, and then try to deliver what they want, and tell them which of their crazy ideas won't work.

That's why I asked people to post links to mod pages. I am interested in how they are able to come up with their concepts. They often have fantastic artwork and web page design. The concept sets the theme and mood for a game, and without that it just isn't inspiriing.

Best Mods?

2009-08-07T10:39:00.000-07:00

Post your favorite game mods here.

I'll start:
Garry's Mod

Why I use Newton

2009-08-04T08:44:00.001-07:00

These are the things you don't read about or find out unless you are on the ground actually using a physics SDK.

2009-08-04T06:46:00.000-07:00

Lua doesn't necessarily make writing code easier; Some of the things it does are little weird and I'm not used to it. The difference, I think, is reusability. I have a simple GUI running. If I were writing this in C++ or BlitzMax, it would only be useful to 50% of the users. Because its written in Lua, everyone can just drop the script into their program and run it.

With programming languages, importing someone else's code is often too much hassle. With Lua there are no worries about header files, or what other parts of the program may be doing. You don't have to worry about the code not working in six months because of some change in the design of your program. Lua allows collaboration in a way programming languages don't.

My GUI is not going to do everything everyone wants. I want to keep it simple. Since the basic framework is there, users can extend and improve it, and whatever improvements they come up with can easily be shared with everyone. Up until now, code collaboration has for all practical purposes been impossible. I know you can share C++ or BlitzMax code, but if something is so difficult no one even wants to do it, is it really supported? Lua is such a huge step forward.

Framewerk for C++ now available

2009-08-03T13:12:00.000-07:00

Mika Heinonen has posted his framewerk code for C++. This allows you to render all effects including SSAO, godrays, DOF, water, and more with simple commands. Check it out here.

2009-08-03T10:59:00.000-07:00

I don't think programmers are the people who are going to make games. I think the people who make games are going to be artists who can do a little bit of scripting; their soul has not yet been crushed by the tedium of programming, and they still have fresh ideas and ambition. I think the difference is "modders" are focused on making their game, and are only willing to put up with whatever technical nonsense is required to make their idea reality. For many programmers, obscure technical knowledge is the point of programming; It's just interesting to them, for some reason I can't fathom.

This realization gives me pause. Noobs may suck at programming, but I have a feeling a lot of them may be better at getting games done.

Firefox Wins

2009-08-02T09:44:00.000-07:00

IE8 can't download code from github, even in compatibility mode. FireFox is the only working browser I can find with a somewhat standard-looking interface. Google Chrome loses because of its "homemade" GUI and lack of per-site cookie control.

I don't particularly like Firefox, but it's the least bad of a lousy bunch. My favorite browsers are still Internet Explorer 6 and Netscape Navigator. I don't use IE6 because I assume it has horrible security holes that had to be fixed with subsequent versions, and of course Netscape has been gone for years.

Cleaning up my hard drive

2009-08-01T12:49:00.001-07:00

Playstation 2 outselling Playstation 3

2009-08-01T08:41:00.000-07:00

Story here.

SDK 2.26 Released

2009-07-30T13:01:00.001-07:00

2.26
-Fixed bug where Sandbox lights couldn't be deleted.
-Included Material Editor missing from last version.
-Fixed entity scaling bug.
-Added missing "default.sbx" file.
-Added ClampTexture() command.
-GLShareContexts() is not automatic in BlitzMax. You must call this before creating a graphics window. The DLL calls this automatically when Graphics() is called.

SDK 2.25 released

2009-07-30T07:30:00.000-07:00

Leadwerks Engine SDK 2.25 is released. New features include support for up to 256 bones, changing of graphics resolution without restarting the engine, and GPU instancing on ATI cards. There are some flaws in the new release which we are working to identify and fix, so if you are working on a project right now you might want to wait for 2.26. Thanks to everyone who is helping me make the SDK better.

2009-07-29T11:31:00.000-07:00

I have disabled comments on this post because people seem to get upset when I talk about my experience with operating systems. This is only my experience, and other people might have found something entirely different. I was very entusiastic abot Windows Vista when I started with it, and despite my initial problems decided I would use the OS exclusively for more than six months. I don't really have anything against Microsoft, and I really like their hardware devices. That said, I am very happy to be back on Windows XP.

First of all, everything is fast. Very fast. It takes less than a minute to restart my computer. I don't have any benchmarks, but moving items around feels faster than Vista, and the GUI seems more responsive. A quick test revealed a 16% performance improvement over Windows Vista when running a Leadwerks Engine application. One reason I wanted to try Vista in the first place was DirectX 10. Apparently, the DX10 features in Crysis can be enabled in WindowsXP by editing a config file. I tried running STALKER: Clear Sky with the highest settings in Vista, but couldn't tell any difference with DX10.

Everything is back to a standard interface, with program buttons that look similar. The Windows Explorer toolbar and menu disappeared in Vista, and programs in general seemed to have no uniform look and feel. I don't remember if Winamp 2.95 wouldn't run, or if I just thought the program looked out of place in the Vista interface, but for some reason I installed Winamp 5 on Vista, and hated it enough to switch to Foobar2000. I am rocking Winamp 2.95 again on XP, and it doesn't look out of place on my desktop. I am still unhappy about the weird interface of Internet Explorer 8, but what can you do? I am happy to once again have a uniform visual style where all my programs look like they belong together. Additionally, the XP icons are easier to recognize. Once of my complaints when I first started using Vista was that the icons were all a shade of light blue, and were difficult to distinguish from one another. In general, I feel like my computer is mine again, and I can get work done more efficiently.

Sygate Personal Firewall works again. This is a program that would not run on Vista. I know there is some kind of built-in Windows Firewall, but I don't know how it works or what it does, since it has never asked me if I wanted to allow any program to access the network.

When I first booted up XP, I was hit with a wave of nostalgia for happier computing times. I don't know why, but Vista just felt like something I was comfortable with was taken away and replaced with something weird and alien. It reminded me of Halloween and Oktoberfest, which was around the time I stopped using XP. It was a time when I felt like my computer was something fun and cool.

I think Vista was slightly more stable, as Explorer has stopped responding twice under XP, and I had to open the task manager, kill the process, and run it again. I also like how Vista uses a thumbnail for image file icons. The wonderful Vista sidebar can be run in XP, but of course it is somewhat of a hack, and the CPU meter doesn't work.

Windows 7 is due out in September. Maybe this was a weird time to buy a copy of Windows XP, but I am not very receptive to the prospect of a new OS right now. This experience has taught me what I value in an operating system: Speed, low resource consumption, compatibility, and reliability. I think it is unlikely that Windows 7 will run faster than XP or use fewer system resources. There's not any features I can think of that would really entice me to try Windows 7. Sure, it has some kind of XP mode it can run programs in, but XP already has that, and doesn't require an extra click to turn it on. I just want the OS to stay out of my way.

2009-07-29T08:57:00.000-07:00

I'm still looking for a web browser I am happy with. Here are my findings:

Opera:
Fast but has an ugly interface. The name and logo are weird. The program feels weird, like it's the Linux of web browsers.

Firefox:
Stupid name, bloatware, consumes too much memory. I don't care about all the wonderful plugins it allows. I just want to get work done.

Internet Explorer 8:
Unusable. I've talked about this extensively.

Internet Explorer 7:
I don't like tabbed browsers and the new interface. I assume this version is no longer safe to use.

Internet Explorer 6:
This is my favorite. I love the interface...plain! However, I assume this version is no longer safe to use.

Maxthon:
Stupid name, security is unknown. A Chinese version of IE6 does not bode well.

Google Chrome:
Stupid name, ugly interface and logo. Doesn't allow control of cookies per-site, which is a deal-breaker for me.

My favorite browser of all time was probably Netscape Navigator. It had a great name, a cool theme, and the interface was a plain Windows interface.

I think the problem developers have is there is one small contingent of users who are very into their plugins and extra features, and their voices are very loud. Developers listen to these people and gear the application towards them, while the silent majority only care about speed, ease of use, and maybe the look and feel of the program.

Upgrading from Windows Vista to XP

2009-07-28T10:42:00.000-07:00

After using Windows Vista for seven months, here are my findings:

Half the time my external hard drive doesn't show up.

Takes my quad core with 4 gb RAM about 3-5 minutes to start up or shut down. The machine is rated 5.9 by Vista, the highest rating possible.

Forgets every folder's view settings so when I open it there are giant icons filling the screen.

When I change view settings, I have to right-click, go to the "Sort by..." menu, and then half the time "Type" isn't an available option. So I have to select the "More..." menu item, then scroll down a list of hundreds of sort options until I find the "Type" checkbox, check it, press okay. Then I have to right-click, go to the "Sort by..." menu, then click the "Type" menu item. Then I have to do it again next time I open the folder, because it forgot the settings I had.

I hate the "breadcrumb" bar and it usually takes more clicks and time to get where I want than the old "Up directory" button.

3D applications run about 15% slower than they do on Windows XP.

Uses a gig of RAM when nothing is running.

Some of the applications I use don't run. I have UAC disabled. My digital camera does not have drivers available for Vista. It's a few years old, but I should not have to buy a new camera to work with an OS.

There are some things I like about Vista:

Desktop gadgets are cool. I wish there were more.

The interface does look good.

Using image previews as the file icon is great.

However, after about 7 months of using Vista, I am ready to switch back to XP. I can't wait to see how fast this machine runs with Windows XP. WIth Vista, I generally feel like my computer is no longer mine. I want it back. I'm not a Microsoft hater. I just prefer Windows XP, after having tried both operating systems extensively. If all other factors were equal, I would of course prefer the newer version of Windows, so there has to be some significant motivation for me to revert to an older version.

Regarding Windows 7, I have not tried it and am not willing to invest the time in doing so. I think all the features they added in Vista really weigh the OS down, and I don't trust their ability to make a fast and sleek OS like XP after this. I also do not trust the word of the Windows 7 proponents, because these are the same people who said there was nothing wrong with Vista. Everyone said Vista would be faster, that SP1 would fix things, that an XP mode would make programs work, etc. So when I hear the exact same things said about Windows 7 by the same people, it means nothing to me.

GT300 specs revealed

2009-07-27T17:33:00.000-07:00

Kind of interesting read here. They have 512 stream processors! A GEForce 8800 has about 96-128, I think. The most interesting part of this story, to me, is it means the GEForce 8 and 9 series will drop further in price. I disagree completely with what Tim Sweeney says, though. I think both NVidia and Intel are chasing a market that isn't really there with all their super-parallel computing investments.

What could you do with one of these? Well, since the engine scales pretty well on any SM4 hardware, I think you could just turn the shadow distance and resolution up, add more vegetation with a higher view range, have more lights, and more post-processing effects at a better framerate. It would be interesting to see what one of these things could do.

The two most beautiful screenshots in Leadwerks history.

2009-07-24T15:14:00.001-07:00

I've been going through images to find suitable splash screens. These two stand out so far above the rest that I can't imagine including anything else.

First is Rich Digiovanni's power plant, made about a year ago. The detail and quality are awesome. Combined with SSAO and our lighting, the result is amazing:

I took this shot a few days ago while testing Hau Nghiep Phan's antialias filter. I think this is the first time I've really gotten outdoor scenery right. Rich made the textures here as well. I didn't think that much of this one at the time, but when I saw it later, it really struck me.

2009-07-24T08:00:00.000-07:00

Framewerk is done. All these effects work together correctly, and each can be toggled on and off:
-Bloom
-Distance fog
-Near and far DOF
-Antialias
-SSAO
-Refraction
-Volumetric light scattering
-Water

The code has been sent to Lumooja and is being converted to C++.

These images show the minimum and maximum settings:

2009-07-23T14:50:00.000-07:00

I still would like an animated water normal map.

2009-07-23T09:46:00.001-07:00

I need an animated normal map for water. Where can I get one?

HNPhan's revenge

2009-07-22T21:27:00.001-07:00

I am tending to the framework code. Just realized how nice HNPhan's edge antialias filter is!

2009-07-22T09:32:00.001-07:00

A somewhat abstract video today that came as a result of testing the DOF filter:

2009-07-21T12:49:00.001-07:00

2009-07-21T08:55:00.000-07:00

For the past few weeks, Noel Cower has been working on a rewrite of his Lua interface code for BlitzMax. You can read about the new module here. I think there is a lot of potential for this code to be used in many projects. I spent all evening trying to isolate a bug, and finally was able to produce a simple example, which I posted in the aforementioned thread. This is an example of how to report a bug. The person you are reporting to can't do anything until they have a simple code example that reliably produces the error.

2009-07-20T10:27:00.000-07:00

I am not one to be very picky about my web browser, but Internet Explorer 8 is so bad I can no longer use it. How can a company with so much money and so many employees fail to make a working web browser after 8 versions? PNG files cause the program to freeze. The "compatibility view" button is just a "press this button every time it appears or the page will be messed up" button. It also functions as an "erase all the text you just typed on the forum now that you notice the cursor is dancing around the page" feature.

I used to use FireFox, but I felt that program got bloated and condescending. At this point, FireFox's main feature, according to its proponents, seems to be that it's not IE. It uses a ton of memory and freezes up, at least when I tried it.

So now I am using Opera. It doesn't annoy me too much, and it's much faster than IE. I thought the slow speeds of the Leadwerks forum and Wiki pages were due to the server, but when I access the site in Opera it is really about five to seven times faster!

So long, IE and Firefox.

2009-07-19T20:01:00.000-07:00

So easy, even a programmer can do it.

2009-07-18T11:51:00.000-07:00

I've learned a lot about engine design, and the market for 3D authoring products. When I started out with this engine, the goal was to make something like Blitz3D, but with better graphics and physics. We successfully reached that goal. The uniform realtime lighting makes things as easy as Blitz3D, but lots of advanced graphics are supported with a very flexible system.

However, I also knew that only a few people ever managed to do anything of note with Blitz3D, because they just didn't know how to structure a game. The majority of our users program in C++. I realized this isn't necessarily because they like it, or because they are a good programmer...the average C++ programmer is about as skilled as the average BlitzMax, C#, or PureBasic programmer (but more helpless). The reason people use C++, for the most part, is because it is free. It's also worth pointing out that the language gets treated as a sort of script, using only LE commands. I'd guess the majority of C++ programmers on our forum don't know how to make a list of objects. So I feel they are more highly dependent on my code than I expected. Then there are the non-programmers, which probably outnumber programmers more than 10 to one.

Realizing these issues is an important step towards fixing them. Although there are some impressive projects being worked on by a few people, a good amount of people are pretty helpless. The question is how to meet their needs without "gimping" the engine.

Consumers don't know what they want. They just don't have any idea of the constraints of hardware, and what technology allows. Generally, technology is just a magic box to them and they have no understanding of what is happening. When I think back to my experience as a consumer in this market, before I learned to program, this is how I thought. What they're really thinking when they evaluate a product is "is this over my head"? They'll overlook all manner of limitations and glitches if they find something they can understand how to make simple changes to, because they want to believe it will work. This was true of me when I started with DarkBasic. You hear other people say "you'll run into problems down the road if you use that" but you don't care because the alternative they suggest is utterly incomprehensible to you.

So when someone says a product is "professional" often times what they really mean is "I can change a few values and pretend I made something myself". They're not going to say this, of course. But what they really want is a game they can tinker around with and make changes to, not an engine. What we usually see, in every 3D engine, is the users just take the default demo and add their own models. This is just what people want.

Understanding these issues, a question arises of flexibility versus ease of use (or rather, lack of use at all). What do you do? Retard the system and make a point-and-click game maker? If that were the only option, I would rather not be designing an engine at all. There must be a way to make things simple and structured without limiting the functionality of the engine.

The answer has come from my testing of the Lua scripting language. This solves several problems at once:

Users can program the engine without an external programming language. This is a much bigger deal than you might think.

Code is modular and entity-oriented.

Let me explain the second point. Each model can have a .lua script. The script can have one or more pre-defined functions that will be called at certain points in the program:

Entity:Spawn()
Entity:Kill()
Entity:Update()
Entity:Collision()
Entity:SetKey()
Entity:GetKey()

Since scripts are associated with a model, a user can make a model, write a behavior script, and upload the whole thing on the forum. Another user can download it, insert it into their scene, and it just works. This is a huge advantage, because with programming languages, collaberation is impossible or extremely impractical. In fact, probably about 10% of programmers will even get to the point where they are writing behavior for a class of entities and loading them from a scene.

Because a script is loaded automatically with a model, this works with a regular engine application, and doesn't interfere with whatever programming language you choose to use. A whole program can be written in script, or you can write your app in C++ and just use the script for entity behavior. Lua makes the game structure and behavior easier, but it does it in a way that pros will take advantage of it, instead of just making a dumbed-down system.

I am happy with this resolution. It adds functionality and structure without penalizing the skilled users. I don't think things like weapons, GUIs, and AI belong in the main engine, but I don't mind adding them in the script level. It took a while to get to this point, but since we have taken our time making sure the core engine is well-designed, I think this will give us all a more solid foundation than we might have otherwise.

What I want to see now is collaberative programming; I'm not going to write your whole game for you, but if someone writes a GUI or entity in Lua, everyone else will be able to easily plug it into their own programs. We can actually work together now, even as we make our separate projects. This modular approach is something I think no one else has really nailed yet, and I am very excited about the possibilities.

Windows 7 for sale?

2009-07-15T14:08:00.000-07:00

Didn't I just buy this last year?

How do you think this will play out? I am not too happy at the prospect of paying for another OS already.

2009-07-11T20:07:00.000-07:00

Here's a preview of what I have been working on. I'm not ready to make any big announcements yet, but I have been working a lot on some new stuff. So my online presence will probably be limited until I'm done.

Google Chrome OS

2009-07-10T07:49:00.001-07:00

Looks like Google is entering the OS business. What do you think? I predicted a while ago that Windows would no longer be the dominant OS within ten years. I also said that neither Linux or MacOS would.

2009-07-09T16:15:00.000-07:00

The first issue of the BlitzMax coder magazine it out, and it's actually pretty good.

2009-07-08T22:16:00.000-07:00

I accidentally solved a small but annoying problem that has existed for some time. Changing resolutions during runtime is a problem for many commercial games. However, it seems to work if you simply call GLShareContexts() at the beginning of your program, before initializing any graphics windows. I know what this is doing internally, but I assumed it was done automatically, and did not need to be called explicitly. Initial tests allowed me to switch screen resolutions in a running application with no problems.

2009-07-06T22:05:00.000-07:00

One new feature I have had in mind a long time is something I am calling "tagged" textures.

In the material file:
texture0="tag::renderdepth"

In your program:
TagTexture(GetDepthBuffer(gbuffer),"renderdepth")

The engine then automatically sets the depth texture whenever the material is used. Previously we have done this by loading the material and setting the material texture to the buffer depth component. This is tedious and has to be updated any time the rendering buffers are recreated.

Not sure if I will stick to this naming convention when it is released, but the point is, this makes it so you can use lots of heat haze and refraction effects without having to hard code the material texture assignment.

2009-07-05T10:05:00.000-07:00

I'm working on a number of technologies that are all somewhat intertwined. I've also got two people working on specific technologies they have specialized knowledge in. I'm not at liberty to discuss it right now, but I can assure you it is really cool.

2009-07-01T18:29:00.000-07:00

I had an idea last night to use conditional rendering to lower the cost of updating directional and point shadows, and spent the whole day implementing that. In the end it turned out the fastest I could get the conditional method was about 8% slower than the regular technique I had before. Not what I expected, but now I know.

NEVER do this!

2009-06-30T13:17:00.000-07:00

In BlitzMax, don't declare a type in a function. It will compile and run, but it will cause random memory errors down the line. I spent about four hours finding the source of a mem leak, and this was it.

For that matter, you should stay away from functions in functions as well.

2009-06-29T16:57:00.000-07:00

Well, I've got it worked out so I can see that everything in the new editor and Lua will work. I'm getting some help to make the Lua integration more efficient, and the remaining editor functionality is mostly just a matter of copying and pasting code from Sandbox.

2009-06-27T07:04:00.000-07:00

According to Valve's hardware survey, 60% of users now have SM4 cards, and between 60-82% can run Leadwerks Engine. Back in December it was a big deal when SM4 users nearly reached 50%. At this rate SM4 users might be around 70% by Christmas, and SM3 could reach 80-90%.

Zenimax buys id Software!

2009-06-24T14:23:00.000-07:00

Wow.

2009-06-24T10:01:00.001-07:00

Well, the Lua implementation I am using is too slow for real-time use. I have to write my own. If anyone knows Lua well, speak up.

2009-06-22T17:24:00.000-07:00

Well, it took all day, but Leadwerks Engine now supports up to 256 bones per character mesh.

If you are just so pleased you can't contain yourself, why not celebrate with some nice textures?

Goodbye, Winamp

2009-06-22T09:20:00.000-07:00

I have discovered foobar2000. There's even a .umx plugin.

Goodbye Winamp bloatware.

2009-06-21T11:54:00.000-07:00

Paper on simulating ocean water:
Click

Yay, now I can delete the bookmark from my browser.

More thoughts on Lua

2009-06-21T09:45:00.000-07:00

Lua Scripting preview

2009-06-20T13:49:00.000-07:00

You must watch this video: