bump map not loading

Kram1032 · Post by **Kram1032** » Fri Mar 16, 2007 5:11 am

I somewhere read, you shouldn't change b or c normally...
just a is important.
So it can't represent RPG...
a should be...

f(x)=a*g(exponent)*2
hmmm...

Isn't there any good simple word for that? It seems that noone (I'm not a exception) can completely figure out, what these things mean... variaables are cionfusing... Sometimes, it would be better to see a formula with the meaning of the variables, instead of either shorts or meaningless letters...

CTZn · Post by **CTZn** » Fri Mar 16, 2007 5:27 am

The final, used value for each component is calculated like so:

"for each component" means than this equation is how R, G and B final values are computed for every pixel of the map. So, for each of these pixels

Code: Select all

R = a*Xexponent² + b*Xexponent + c
G = a*Xexponent² + b*Xexponent + c
B = a*Xexponent² + b*Xexponent + c

At least this is how I try to get the point, maths only killed my scolarity

are abc representing RGB
or something different than like a alpha, b bump, c color value???

f(x) is the final used value

So they don't represent just R, G and B. The whole equation does (each of the rgb components equals f(x)). As for variables, I'm lost too. I'm wondering if that's related with tonemapping, but I guess these calculations happen before...

a, b, c, and exponent are as described below.

that is ... where ?

F.ip2 · Post by **F.ip2** » Fri Mar 16, 2007 2:11 pm

i give up ...

CTZn · Post by **CTZn** » Fri Mar 16, 2007 2:29 pm

I'm sorry... I even don't know if what I've saied makes sense

Stur · Post by **Stur** » Fri Mar 16, 2007 9:09 pm

F.ip2, this is how I understand it :

- A improves gamma a lot
- B improves gamma reasonably
- C improves gamma a few

To be simple, just use B. If your texture is too dark, raise B a little. If your texture is to saturared, reduce B a little.

Hope it helps. (and hope I'm right

)

mrCarnivore · Post by **mrCarnivore** » Fri Mar 16, 2007 9:49 pm

Here's how I understand it:

a, b, c are values that effect the colour of the texture.

the formula f(x) = a * g(x)² + b * g(x) + c is calculated for each r,g,b to calculate the r,g,b values that are used in the render.

I will explain the process for the r-component. The two other components are treated the same:
For clarity I define r_texture to be the r-value of the original texture (g(x) in the original formula).
r_final is the final value of that texture pixel that is used in the render (f(x) in the original formula).
Written with these variables the f(x)-formula looks like this:

r_final = a * r_texture * r_texture + b * r_texture + c

READ FROM HERE IF YOU JUST WANT THE RESULTS:

c is just a constant offset.
It makes all pixels of the texture brighter by the same amount. So the brightness of an image is increased. Notice that clipping might occur if you use a too large value for c.

b makes all pixels brighter/darker linear to their colour value.
So for values of b<1 the contrast of an image is decreased. For b>1 the contrast is increased. Notice that you will get clipping of you choose b-values of greater than 1!

a makes all pixels brighter to the square of their value. It mainly does the same as the b value just with an even stronger effect. The a-parameter makes the darker values much more dark than the b-parameter, whereas the brighter pixels are affected almost the same by the a- and b-paramter. However, you will not get any clipping, regardless of the values of a you choose.

How about an example:
So if you have a pixel of value (r_texture,g_texture,b_texture) = (0.1, 0.4, 0.9)

The impact of the a,b,c values will be following:
a=0.1, b=0, c=0;

(r_final, g_final, b_final) = (0.001, 0.016, 0.081)

a=0, b=0.1, c=0;

(r_final, g_final, b_final) = (0.01, 0.04, 0.09)

a=0, b=0, c=0.1;

(r_final, g_final, b_final) = (0.0, 0.3, 0.8 )

Hope this helps. Actually thinking it through helped even my understanding a,b,c more.

manitwo · Post by **manitwo** » Fri Mar 16, 2007 10:22 pm

good explanation! The standard indigo user won't benefit from this tho.
(except if the exporters would support a,b,c with a live-preview

)

mrCarnivore · Post by **mrCarnivore** » Sat Mar 17, 2007 12:03 am

Actually it generating a preview for the texture according to the a,b,c parameters should be quite quick to calculate. So if somebody has the time and skills, it should be straightforward to implement in the exporters...

Post by **OnoSendai** » Sat Mar 17, 2007 12:05 am

Well.. most people will only need to use b (gain).
The c parameter is there so that, say, if you want to use a negative of a texture for a blend material, you can set c=1 and b=-1.
The a is there because... well it might come in useful for something, and it's very cheap to compute

mrCarnivore · Post by **mrCarnivore** » Sat Mar 17, 2007 12:42 am

I can see why you namen the variables a,b,c. It makes sense looking at the formula.

From the whole concept about indigo and your responses I can gather that you think very mathematically. However, it might be useful if you replaced (or explained) the b with contrast, c with brightness. I guess those are equivalent terms to the effects that those coefficients have.

Having said that, it might be enough if the exporters just renamed the variables in their GUIs and you kept a,b,c in the xml...

Wedge · Post by **Wedge** » Sat Mar 17, 2007 2:09 pm

Thanks everyone for the information.

I read it all and understand better the meaning for these.

F.ip2 · Post by **F.ip2** » Sat Mar 17, 2007 4:16 pm

I agree! I wish I would understand the math much better.
I hate just knowing where to click to get something but not knowing
why I have to do it.

I think replacing A,B,C with B like Bump depth value for example will help
non programmers a lot to understand what they are doing!!!

Anyway Ono great job for the render engine. This is my first time using it and I have to say for my first image I am very happy!

I just wish velvet and anisotropic reflections would be possible. I ran into many anti aliasing issues with to fine brushed image maps. They produced a lot of moires.

claas