[ODFAEG] (Open Source Development Framework Adapted for Every Game)

[Lolilolight]

Ha yeah I've forgotten, I'll post a small source code example soon : (but for the game it'll take more time, because I need to add a gampeplay)

I think I'll post less often.

And even if it was, this one is not it - lack of RAII, lack of OpenGL understanding, just generally messy code, a developer who is "full of himself/overstating own abilities", lack of use of modern C++11, etc etc..

Sorry but I use all this technique and it works, but maybe I need to review the source code a bit.

Sorry to disappoint you but I understand opengl very well, RAAI, I use it because the std::vector are using it and even if they don't I destroy them in the destructor.

And I use modern c++11 techniques also and a lot of modern techniques. (Entity scene nodes, pattern observer, based component pattern, type erasure, CRTP, etc..., etc...)

But you'll see it once I'll have finished the tests and the demo.

But I'm not in a hurry, all I want now, is to have a good design. (The demo'll come just after)

But now I think that the design is ok, so, I'll post less, otherwise this thread'll never end, and, I'll post again when the demo'll be finished.

[G.]

I browse the SFML forums using the "Unread" feature, it shows every threads with new posts. I come here more than once a day and this thread always has "something" new. I don't want to hurt your feelings because I think you're a little bit retarded, but nobody takes this project seriously. It is almost 200 posts long and yet there is nothing to see. 

Maybe you should post less here and more on your devblog. Come here to show big updates, not useless things that change every day.
And claiming that you're "one of the most experienced C++ dev in the world" doesn't help, especially when you have achieved nothing and have been unemployed for years. No you're not that awesome and it's not a shameful thing to be average.

[Njifra]

I browse the SFML forums using the "Unread" feature, it shows every threads with new posts. I come here more than once a day and this thread always has "something" new. I don't want to hurt your feelings because I think you're a little bit retarded, but nobody takes this project seriously. It is almost 200 posts long and yet there is nothing to see. 

Maybe you should post less here and more on your devblog. Come here to show big updates, not useless things that change every day.
And claiming that you're "one of the most experienced C++ dev in the world" doesn't help, especially when you have achieved nothing and have been unemployed for years. No you're not that awesome and it's not a shameful thing to be average.

I agree, and ull never make that example...
You have finished 0.00009% of it...

[Grimshaw]

As much as the other things may or may not be discussable, you do not understand OpenGL at all. You're exactly the definition of a beginner using it as you've proven all along this thread. Even the little you're getting done with OpenGL is all taken from tutorials around the web, as everyone can see :p

Be honest, you are mocking us or being serious?

[StormWingDelta]

I say build for one game type at a time than find out what is common among all the types of games and build a framework around that rather than what makes them different.  If you build around what they all have in common than branch out from there you'll be able to pull this off.  It isn't impossible but I think the old saying, "Don't bite off more than you can chew.", comes to mind.  Also if you need help just ask, trying to do all this by yourself isn't going to end well as I found out the hard way.  At the very least we can help by finding more useful tutorials.  I'd recommend ignoring anything more than 3 years old unless it is really good or the codes for it haven't changed much.


If you need a few helpers there are bored people like me that might help if able.

[Lolilolight]

Hi, yes it's right I'm unemployed since several years, this is why I want to make this kind of framework, because, over one year, I'll loose the insertion allocations, and I don't want to do something other than computer sciences. (And to have made studies for nothing)
I'm intending  to  create my game developpement own studio so if I'm still unemployed in one year (I hope this'll work), if not, I'll continue this as a passion.

The tutorials what I've found on the internet are sufficient for the moment and I don't have time to read many books or articles with explain a lot of tehcniques.
But, if someone have done it and if he really think that he really have better knowledge than me, he can help, or he can shut up.

I've the intention of creating a game soon for the demo but only a 2.5D game because it's easier to start to do tests and after, I'm intending to create a full 3D game. (A simple FPS)

This thread becomes to be..., a mess, so I think I'll recreate another one once when I'll have finish to encode the demo and correcting all the remaining issues.

I don't know if people are passing on the devblog, I've more answers on a forum most of time.

But for the advanced techniques I must admit that I need help.

PS : ok, as you think that I'm not serious I'll post less, and starting to encode the demo.

It's just that many people make me wasting a lot of time by mocking me, last years I had a proposition but never answers for a stage, they promise to create job, they force us the search for a jobs, but, there's one job for 50 people and we have to past tests, so, there's no chance.

And a few years ago, a people was telling me that he have a stand at the japan expo to present the project but it as a lie.

I think this domain is simply satured.

But i'm not here to talk about that, but please don't make links between thing which they haven't any rapport.

It's not because I'm unemployed since several years that I'm bad.

[Nexus]

You're not bad, but you want to do too much at once, which is simply not realistic. Have a look at existing 3D frameworks like Ogre, Irrlicht, CrystalSpace (or even all-in-one engines like Unity), but also 2D ones like Allegro, SDL, SFML. They have existed for years if not decades, and the people developing them are often very experienced, as they have worked years in game development industry before. This experience is something you don't have, and even with that experience it would be almost impossible to create a framework of such a scale and quality in less than a year. You must acknowledge that fact.

You talk about saturated market, yet you want to develop something that other products already do much better. If you want to be successful, you should rather find a niche, where your framework has potential. If you start small, it's not only more likely that you will ever finish, but you also gain experience that you can use later for bigger projects. And you have something to show that you achieved -- people won't question what you did the whole time, if there is a result. My suggestion is that you cease the development of ODFAEG and find a more specific problem domain where your work is recognized by the community.

The reason why you are not taken seriously is because you never consider our advice and because you have imaginations of your framework that are beyond reality -- not only your future visions; you also completely overestimate what you have already done. In this forum there is a great exchange of knowledge, but if you want to benefit from it, you should dismiss your attitude of knowing already everything (which you obviously don't) and be open for suggestions from others. Every single person responding here has already said that (in fact, I alone am saying it for the fifth time or so), so please consider the possibility of it being true. And before, the exact same story has happened in the French forum.

Lolilolight, it is really really time that you realize the whole situation and take care of what you post. This can't go on like that.

[Lolilolight]

Hi!

I've implemented the vbo in the vertex array and render target classes.

Now, I've still to correct some positionning problems in the FastRenderComponentManager class for multi pass rendering with the camera, and the core of the framework'll be almost finished.

Before beginning to encode the last version, I'll write a demo, I end the discution on this topic here, and I'll start a new one, if anyone can help me to correct the remaining bugs, he's welcome.

[Grimshaw]

I don't think encode means what you think it means :p

[Lolilolight]

Yeah probably.

I've a little example of source code which runs, I'll add a tutorial and a source code example soon.

And later the same but with the network.

But I've still a problem when I draw component in the fastRenderComponentManager class, when combining the view transform, I've a part of the screen which is black. :/

So I draw directly entities on the windows and component directly on the window, without applying my shaders.
I think I need to find a tutorial about billboard or something like for multi-pass shader rendering but if anyone can help me he's welcome.

Once I'll have find a way to do multi-pass rendering, it'll become more funny. (Because I'll be able to start to make the first game with these framework.)

[Grimshaw]

If a part of the screen is not being rendered to (is black), you probably need to call glViewport() to the whole window, to make sure its used. Also make sure the scissor test is disabled. That should help.

What do you need multi pass rendering for?

[Lolilolight]

I've done this but it doesn't work, I think I'll try with the source code of of 3DEE. (I hope it'll word)

I need the multi pass rendering to render 3D semi transparent objects correctly.

Especially for semi-transparent objects which are in collision with others (like water textures), I can't display them in a certain order because some semi-tranparent fragment of a texture can be behind or before the fragments of the second semi-transparent texture, and as the default opengl depth test don't consider semi-transparent fragments, I need a shader with test also the alpha component of the pixel before drawing it, like this :

 const std::string vertexShader =
            "void main () {"
                "gl_Position = gl_ModelViewProjectionMatrix * gl_Vertex;"
                "gl_TexCoord[0] = gl_TextureMatrix[0] * gl_MultiTexCoord0;"
                "gl_FrontColor = gl_Color;"
            "}";
            const std::string  depthGenFragShader =
            "uniform sampler2D depthBuffer;"
            "uniform sampler2D texture;"
            "uniform vec2 resolution;"
            "void main () {"
                  "vec2 position = ( gl_FragCoord.xy / resolution.xy );"
                  "vec4 color = texture2D(depthBuffer, position);"
                  "vec4 pixel = texture2D(texture, gl_TexCoord[0].xy);"
                  "if (gl_FragCoord.z > color.z && pixel.a >= color.a) {"
                     "gl_FragColor = vec4(0, 0,  gl_FragCoord.z, pixel.a);"
                  "} else {"
                     "gl_FragColor = color;"
                  "}"
            "}";
            const std::string frameBufferGenFragShader =
            "uniform sampler2D depthBuffer;"
            "uniform sampler2D frameBuffer;"
            "uniform sampler2D texture;"
            "uniform vec2 resolution;"
            "void main () {     "
                "vec2 position = ( gl_FragCoord.xy / resolution.xy );"
                "vec4 depth = texture2D(depthBuffer, position);"
                "vec4 color = texture2D(frameBuffer, position);"
                "vec4 pixel = texture2D(texture, gl_TexCoord[0].xy);"
                "if (gl_FragCoord.z >= depth.z) {"
                    "gl_FragColor = pixel;"
                "} else if (color.a < pixel.a) {"
                    "gl_FragColor = pixel;"
                "} else {"
                    "gl_FragColor = color;"
                "}"
            "}";
 

So more transparent pixels doesn't hide more opaque pixel when their z position is greater. (And I can draw textures in any order that I want.)

[Lolilolight]

I've found the bug.

So, only some network tests are remaining, some source code is to clean up and to put in .impl and .cpp files and it should be ok for the version II and the demo.

Everyone who thinks he can improve the source code design and to add some idioms is welcome.

[Lolilolight]

Hi!

The second version is nearby finished, I've still to clean the code a bit, add some idioms (RAII, etc...) and then, finishing the documentation and the demo!

I'll put the listener into a class called InputSystem, and the cache into a class called ResourceSystem and use the monostate pattern like I do in the network class. (I've to avoid the use of the singleton pattern in a multi-threaded context, and I want to make each functionnalities independant of the framework, by example, the developper can use another framework for his application and also using the input system of odfaeg)
I've also corrected the source code of the network module. (And the developper'll be able to choose if he wants to make a secure connexion or not, in a further version I'll add an https class to add the https protocol)

The best code design

The framework'll use this following patterns or idioms so :

-RAAI for resources acquisition and liberation.
-CRTP (curiously recursive template pattern) for calling 2D or 3D entity manager's function. (it depends if the entity manager which is stored in the world
-Type erasure. (For signals and slot management and also for storing resources of different type into the cache)
-Based Component Design pattern : (The 2D or 3D entities'll be able to be dawn on components, very usefull if we need to do multi-pass rendering with differents views or with different layouts, the component manager class'll be able to use advanced rendering functions (new blend modes, advanced alpha test, etc...) but only if your pc support shaders)
In the same optic, the view class'll combine a lookat matrix with a transformation matrix, so, no need to use billboards)
I don't want to make this framework very big like Irrlicht, Ogre, etc..., but I want to combine all the techniques of existant framework or libraries to make the source code simplier to use and most powerfull. (And also to make my life easier)
-Entity scene nodes : the transformations between the entities of the scene graph'll be able to be combined if each graphic entities have children or parent node. (But you can redefines transformation functions if you need to update other's informations like collision areas or collision volumes of the entities)
The framework pre-transform the vertices of a root entity and each childs. (The framework also pre-transform vertices if your pc support vbo to have a very fast rendering time)

[StormWingDelta]

Keep in mind when making this that it should be as unrestricted as it can be like SFML is if you are intending people to be able to use for any game.   Also you could have a list of game types and than one of those could be a fully unrestricted type where they can use all the tools rather than limiting based on the game.  This will help quite a bit.