6 replies to this topic

[tuccio`]

Well, i was setting up an hardcoded cornell box scene with opengl because i want to learn glsl and hopefully write a fragment shader path tracer

so i copied these data (here's the code), but i had a result i didn't expect, and i don't understand



There's no reason for having an enlighted left wall and such a dark right wall.. also that floor shouldnt be so dark if the bottom of the left wall is still enlighted

But if i invert floor and right wall normal direction (right wall normal = (-1, 0, 0) and floor normal = (0, -1, 0)) i get something similiar to what i expected



i feel like i'm missing something :/

Edited by tuccio`, 29 June 2012 - 07:02 AM.

[Brother Bob]

The light parameters are 4-dimensional vectors, not 3-dimensional vectors as in your code. OpenGL is currently reading the fourth component from outside the corresponding arrays.

[tuccio`]

The light parameters are 4-dimensional vectors, not 3-dimensional vectors as in your code. OpenGL is currently reading the fourth component from outside the corresponding arrays.

oh, thank you

that helped, and now it works fine if i set position's w component to 1 (so that the light propagates in all directions), as you can see:



but if i put w to 0 to get a directional light, this is what i get:



the code now looks like this

i guess i'm still missing something :/

Edited by tuccio`, 29 June 2012 - 08:00 AM.

[Brother Bob]

What do you think is wrong with the directional light? If the light is coming from the upper left in the image, then the bottom and right faces are lit since they face the light, and the right and top faces are not lit since they are not facing the light.

I haven't looked much into the exact details of where the light is pointing, where the view point is facing, and which wall is where though, so I don't know for sure if the light really is supposed to come from the upper left. If the incorrect faces are lit, then make sure that the normals are in fact correct, and that you are viewing from the correct place.

[tuccio`]

What do you think is wrong with the directional light? If the light is coming from the upper left in the image, then the bottom and right faces are lit since they face the light, and the right and top faces are not lit since they are not facing the light.

I haven't looked much into the exact details of where the light is pointing, where the view point is facing, and which wall is where though, so I don't know for sure if the light really is supposed to come from the upper left. If the incorrect faces are lit, then make sure that the normals are in fact correct, and that you are viewing from the correct place.

the light is in the middle (sort of) of the room, actually closer to the ceiling than the floor, and the direction points towards the floor

also, changing the direction of the light and the spot cutoff doesnt change the result, and this is weird too

[Brother Bob]

You have a directional light, not a positional one. Spotlight settings only work for positional lights. Set the w-component of the position to 1 instead.

Also, don't expect very good results. Lighting is calculated per vertex so you need to define a much higher resolution cube to have some sensible results.

[tuccio`]

You have a directional light, not a positional one. Spotlight settings only work for positional lights. Set the w-component of the position to 1 instead.

Also, don't expect very good results. Lighting is calculated per vertex so you need to define a much higher resolution cube to have some sensible results.

yup i think i understand

and yes, i didn't expect great results, i'm using a low tesselated cube because, as i said, i want to learn glsl and use fragment shaders to compute direct and indirect lighting

thank you again