Messages

76

SFML projects / Re: NNBrain SFML Demo

« on: November 03, 2013, 03:46:25 pm »

@StormWingDelta: Sounds like a cool idea! I would be very interested to see how well the AI performs in such an application. You could train it against itself at first, and then when it is good enough, it can learn from the players it is playing against.

Also: A manual for how to use the system is coming very soon!

77

SFML projects / NNBrain SFML Demo

« on: November 02, 2013, 10:02:00 pm »

Hello all,

I would like to present to you my new self-learning AI library, NNBrain!
It includes a demo that uses SFML to visualize the cart-pole balancing problem.

Right now, there are 3 agents (AI's), 1 of which is well know while the other two are more original in implementation:

• Actor-Critic
• Neural-Q
• CACLA

All agents are applicable to both continuous and discrete task domains.

Feel free to use it in a game or other application to hopefully get some awesome self-improving AI!

Download Link: https://sourceforge.net/projects/nnbrain/?source=navbar

78

SFML projects / Re: GLLight2D

« on: September 13, 2013, 03:46:52 am »

Unfortunately there is no documentation... yet.

GLLight2D can be run in real-time on more powerful machines, but depending on your target platforms, you can pre-generate a light map (texture with all lighting rendered to it) on level load using it instead. GLLight2D operates entirely differently from LTBL, it is a path tracer, which has the benefit of giving you full global illumination, but is much more processing intensive. So, it is a looks vs. performance trade-off. If you want speed, go with LTBL, if you want global illumination, go with GLLight2D. I personally think GLLight2D is much cooler

79

SFML projects / Re: NEAT Visualizer

« on: July 25, 2013, 08:42:12 pm »

I posted an article describing how you can create your own AI similar to that of the NEAT Visualizer.
I know that I am not the best writer, but hopefully it is well written enough to follow 

Here it is for those interested:

http://www.gamedev.net/page/resources/_/technical/artificial-intelligence/reinforcement-learning-for-games-r3271

80

General discussions / Re: SFML Game Jam

« on: July 21, 2013, 08:31:27 pm »

Neither is that a specific feature of SFML, nor is that argument very expressive. You can extend it to "SFML allows you to use the Audio module for sound, and Irrlicht for graphics". Like this, you can justify any other library.

Well, SFML wraps OpenGL. So you would be forcing the use of a wrapper, whose simplicity should provide enough justification for most people to use it instead of OpenGL. If it is so powerful (which I think it is), then it shouldn't need to be forced. I am just one of those guys so used to OpenGL that they won't use anything else 

Also, I don't see an "Irrlicht" header in SFML, but I see an "OpenGL" one.

But fine, if OpenGL is still off the table, I will probably compete regardless.

81

General discussions / Re: SFML Game Jam

« on: July 21, 2013, 07:59:13 pm »

I must continue the fight for OpenGL 

One of the cool features of SFML is that you can so easily integrate it with plain OpenGL! I think this feature should be advertised like any other. SFML includes OpenGL integration, so using OpenGL is still using a feature of SFML. If SFML provides something, you should be able to use it. Besides, wouldn't it be cool to have people make 3D games?

82

General discussions / Re: SFML Game Jam

« on: July 21, 2013, 07:35:56 pm »

@ Aster:
We can say that it at least needs to use SFML windowing and sound or something like that. Some minimum integration level. OpenGL integration is part of SFML's feature set, so I think you should be able to use it. If OpenGL were allowed, then why not allow 3D as well?

*raises hand up* I used sf::Vector2f in my GLFW game! Does that count?

We could have a SFML integration rating or something, or just ask those who rate it to keep in mind the extent to which SFML was used. Let the community decide if the developer was cheesing the use of SFML.

Games based on OpenGL defeat this purpose and may convey the impression that combining SFML with OpenGL is the standard way to do it (as it was for SDL if you wanted reasonable performance).

Well, if everyone ends up using OpenGL, then it IS the standard way of doing it  I think that games submitted would give a reasonable representation of who likes using SFML's drawing API and who prefers to use OpenGL with SFML.

I don't want to abandon my OpenGL 

83

General discussions / Re: SFML Game Jam

« on: July 21, 2013, 06:36:02 pm »

When allowing OpenGL and what-not, it will be more difficult for people not familiar with it to compete.

I don't think it is possible to eliminate the experience advantage, especially not by restricting how you can use SFML. OpenGL integration is a feature of SFML, and I think it should be permitted (I always use OpenGL directly, just because I am used to it  ). Besides, OpenGL users would likely produce similar graphics in the end if they stick to 2D. The SFML abstraction of OpenGL is supposed to make it easier to use, not limit it (again, when in the realm of 2D).

I would suggest SFML be required, but without any further restrictions.

84

General discussions / Re: SFML Game Jam 1 Theme Submission Thread

« on: July 20, 2013, 01:28:48 am »

Roguelike
Artificial Life
Game-ize an algorithm
Procedural Generation

85

SFML projects / Re: NEAT Visualizer

« on: July 06, 2013, 04:01:03 am »

Yet another update (now at 2.1), and again to the reinforcement learning! I heavily modified the previous algorithm so that it "drifts" away from undesired outputs and is "slapped" into taking the correct drift direction (there are 2 directions). There are 2 new adjustable parameters as a result, alpha for the drift speed and slapFactor for the directional slap. I have tested this new learning rule with the NEAT evolved networks as well as plain feed forward neural networks and it is able to easily learn logic gates.

86

SFML projects / Re: NEAT Visualizer

« on: July 03, 2013, 04:33:23 am »

@ DrSuperSocks: The video had about 20 mechs in different rooms. I haven't really stress tested it yet, but for the video it ran at 60 fps (vsynced) while fraps was recording at 60 fps as well.

87

SFML projects / Re: NEAT Visualizer

« on: July 02, 2013, 11:03:42 pm »

I found a couple of issues with the current reinforcement learning algorithm, so I replaced it with a completely new algorithm which is far superior to the previous. So, I now present version 2.0 of the NEAT visualizer. Reinforcement learning is now included in the demo as a second stage of learning after the evolutionary process in order to fully optimize it.

88

SFML projects / Re: NEAT Visualizer

« on: June 30, 2013, 09:23:30 pm »

New version 1.9 uploaded, with better reinforcement learning as well as Hebbian learning!

Edit: Here is a video of this NEAT package being used to animate a mech! The animation if physics based and will react to the environment as one would expect. The neural network controls motors in the limbs in order to allow the robot to walk.

89

SFML projects / Re: GLLight2D

« on: June 26, 2013, 10:01:08 pm »

Hmm, that's not the problem. It actually says it is 0x1 for me too, but when I change the material, the render reflects those changes, so gDEBugger is not correct.

Try validating the shader by changing

bool TracerScene::Create(const std::string &shaderName)
{
        return m_traceShader.CreateAsset(shaderName);
}
 

To

bool TracerScene::Create(const std::string &shaderName)
{
        if(!m_traceShader.CreateAsset(shaderName))
                return false;

        return Shader::Validate(m_traceShader.GetProgID());
}
 

90

SFML projects / Re: GLLight2D

« on: June 26, 2013, 09:43:27 pm »

The third one shouldn't be empty. I think that one is the material buffer, and no material probably results in black! I will look into it now.