Project Ninja Star - help wanted - ported to windows on 26th

[bitplane]

I think those IQ tests are mostly done by stupid people who are worried about their IQ, which lets any semi-intelligent person get a decent score. A real IQ test has questions that are impossible for normal people, that one was easy.

Sorry for the off-topic post.

[hayate]

... A real IQ test has questions that are impossible for normal people, that one was easy.
...

OT
I disagree with that , sometime the difference between "normal" and "above" or "below" "normal" it's just about the time you take to answer a question possible even for "normal" people. I tried this test but I think it isn't very useful because it doesn't have a time limit and, as already stated, it's like a sort of basic admittance test
/OT

@ devsh: When do you plan to release a demo of your game? I'd like to give it a try (and you can collect datas from the performance on other peoples machine)

[Prott]

So the fact of the matter is that my depth of field shader cannot be processed any faster than 74FPS.

Interesting .. what shader are you using? You could take advantage of that shader by using low detail/high detail models according to the intensity of bloom.

(Why buy quad cores, people?)

Because of realtime physics - for example...

And.. you could use IrrEdit maps instead of BSP. It`s better, faster, etc.

[devsh]

and that's right!!!

[devsh]

Anyone KNOWS of a better way of blurrring in a shader than this:

Code: Select all

for (i = -4; i<4; ++i) {
   for (j = -4; i<4; ++i) {
      sum += texture2D(RT, texcoord + vec2(i,j) * weight*difference_between_focal_point_in_depth_tex) * sampler [i+4];
   }
}

gl_TexCoord[0] = sum * 0.16;

As You can imagine one pixel can get sampled up to 16 times!!
the problem is that i could make a way more efficient shader IF I could store variables in between shader passes (so after the pixel 0,0 has been done i can pass data to pixel 0,0.003) because all the global and local variables get wiped, or initialised AGAIN.

do I use varying.

Or does One have a better blurring algorithm than getting the values of all the 15 surrounding pixels in a square around the original one?

[lostclimategames]

not sure how it works for glsl but in hlsl you can make an array of offsets, and just run thru the for function using the index to get the values. also you can do it that way to make the blur less banded.

[devsh]

I think my lines of code have reached the magic 2500

[devsh]

So my team (me and Darell) decided to try to pose professional sensible people and we have set ourselves a TODO list:

http://code.google.com/p/project-ninja- ... dated=TODO

100 out of 237 are things to do with modeling, so NEED MORE ARTISTS (NOT MORE, MANY because WE HAVENT GOT ANY)

[devsh]

today is a very big day in which i am going to create the graphics options, which will include HDR and bloom.

[wITTus]

Screenshots!!

[devsh]

Almost, I will post in a couple of min. then i will implement the solutions in split screen

[Prott]

HDR

So you are actually using expanded colorspace?

[devsh]

I do not have a clue of what you are talking about,but if you mean floating point thingie, NO. I have went for a less accurate method which still kind of works.




P.S. I think i better switch to high res

[Prott]

It works, but it`s not HDR (high dynamic range lightning). It`s a glow (only one part of HDR process - others are saving lighting values with range above 255 "steps", rendering only some portion of that and adding noise to dark areas). But don`t worry .. I`m using it too.

[devsh]

yeah i do save values above 255 but i do it with less accuracy.

but tell me about noise in dark areas