Shader tests

Big Fan · Post by **Big Fan** » Tue Aug 12, 2008 7:02 pm

it would be cool to declare textures and shaders globally (keeping using the index thing) instead into a material,

God give me a break I only just did the index thing I dont need it redone already

How about we are just happy with what we have for a while
Too much more of this stuff and people burn out with exporter updating

Plenty of other things could get attention without getting convoluted in shaders - reminds of the tropical sea water adventure

Post by **fused** » Tue Aug 12, 2008 7:58 pm

okay, heres some colored gradient, im on a mac right now, so no chance to test it(maybe someone wants to give it a try?):

edit: they dont seem to work

edit: now they do
edit:dont work as expected, forget it

Post by **fused** » Tue Aug 12, 2008 9:05 pm

got it now: renders will follow soon:

Code: Select all

<albedo>
	<shader>
		<shader>
			<![CDATA[	
					
			# 2D colored gradient shader V

			def colorCalc(real x) vec3 :
				vec3(
					# color1 = red (0.75,0,0)
					# color2 = green (0,0.75,0)
					# compute color mix

					#  color1 \/                      color2 \/
					add(mul(0.75,x),mul(sub(1.0,x),0.00)),
					add(mul(0.00,x),mul(sub(1.0,x),0.75)),
					add(mul(0.00,x),mul(sub(1.0,x),0.00))
				)
						
			def eval(vec3 pos) vec3 :
					colorCalc(dotj(getTexCoords(0)))
			]]>
		</shader>
	</shader>
</albedo>

heres the z up gradient thingy with color:

Code: Select all

<albedo>
	<shader>
		<shader>
			<![CDATA[	
					
			# 3D colored gradient shader Z (up) with center = 0.5
			
			def colorCalc(real x) vec3 :
				vec3(
					# color1 = blue (0,0,0.75)
					# color2 = green (0,0.75,0)
					# compute color mix

					#  color1 \/                      color2 \/ 
					add(mul(0.00,x),mul(sub(1.0,x),0.00)),
					add(mul(0.00,x),mul(sub(1.0,x),0.75)),
					add(mul(0.75,x),mul(sub(1.0,x),0.00))
				)
						
			def eval(vec3 pos) vec3 :
					colorCalc(div(mod(add(dotk(pos),div(0.6,2.0)),0.6),0.6))
			]]>
		</shader>
	</shader>
</albedo>

edit: just notced there was some color clamping going on in the 3d gradient. fixed now

Deus · Post by **Deus** » Tue Aug 12, 2008 10:38 pm

Still cool but useless images

Ill post something... (image shack messed image up bad)

here i used a shader that looks like this:

Code: Select all

	<material>
		<name>land</name>
	      <diffuse>
  			<texture>
		            <uv_set>uv</uv_set>
		            <path>world.png</path>
				<exponent>1</exponent> 
			</texture> 

			<albedo>
				<shader>
					<shader>
						<![CDATA[	
						
						# This is a comment
						
						def eval( vec3 pos ) vec3 :
					               sample2DTextureVec3(
					                  0,
					                     add( 
						                     	mul( 
						                     		vec2(
						                     	  		mul( cos(mul(noise(pos), 6.28) ) ,fbm(pos,10) ),
						                     	  		mul( sin(mul(noise(pos), 6.28)) ,fbm(pos,10) ) 
						                     	  	     )
						                     	  , 0.001 )
						                     	,
						                     	getTexCoords(0)
					                     	)
					                  ) 
						]]>
					</shader>
				</shader>
			</albedo>
		</diffuse>
	</material>

It is used to preturb the texture coordinates slightly so I can get away with a much lower resolution for the ground without getting square looking patches. Look closely at the green patches on the beach. Looks pretty nice. Trees still looks like shit but I did reduce the beach "size" after suggestion from you guys

Oh while I am at it. Ono what is the ouput range of noise functions? 0-1? -1 - 1 ? 0-X? talking about noise/fbm

Cheers!

/ D

zsouthboy · Post by **zsouthboy** » Wed Aug 13, 2008 2:56 am

I like how Deus casually posts something awesome, like it's no big deal.

Borgleader · Post by **Borgleader** » Wed Aug 13, 2008 4:07 am

zsouthboy wrote:I like how Deus casually posts something awesome, like it's no big deal.

+1

He is one letter away from Zeus...so I guess awesome is normal for him

Lord of the Rings Junkie · Post by **Lord of the Rings Junkie** » Wed Aug 13, 2008 4:45 am

OnoSendai wrote:Another test:

(Piccy of brushed metal sphere)

Looks good as a bump test, but the shading is far from anisotropic.

A true anisotropic shader would be awesome.

Deus · Post by **Deus** » Wed Aug 13, 2008 5:18 am

Oh thanks. It depends on who you are. I cannot model an ordinary vase properly.. math and programming I can do in my sleep. I envy most of you guys in this forum for your artistic skills

If you wanna steal ...err use the shader yourself preturb textures more then increase the 0.001 parameter (its the pertubation distance)

pixie · Post by **pixie** » Wed Aug 13, 2008 7:50 am

Borgleader wrote:
zsouthboy wrote:I like how Deus casually posts something awesome, like it's no big deal.
+1

He is one letter away from Zeus...so I guess awesome is normal for him

You might not know but Deus is* God...

*is as in means

StompinTom · Post by **StompinTom** » Wed Aug 13, 2008 12:02 pm

zsouthboy wrote:I like how Deus casually posts something awesome, like it's no big deal.

nah it really isnt a big deal. so 90s.

CTZn · Post by **CTZn** » Wed Aug 13, 2008 3:26 pm

Lord of the Rings Junkie wrote:
OnoSendai wrote:Another test:

(Piccy of brushed metal sphere)
Looks good as a bump test, but the shading is far from anisotropic. A true anisotropic shader would be awesome.

If you mean an anisotropic shader like in biased renderes, I don't agree. Or, as far as I'm concerned, you are not looking after a realistic effect

Post by **fused** » Wed Aug 13, 2008 6:28 pm

Deus wrote:Still cool but useless images Ill post something...

whoa, thats cool!

Post by **fused** » Wed Aug 13, 2008 6:42 pm

some new output, trivial and not as cool as deus' but funky with displacement

rendered as always by zomB

edit:

heres the mat:

Code: Select all

    <material>
        <name>shader</name>
        <diffuse>
            <albedo>
                <shader>
                    <shader>
                        <![CDATA[   
 
                        # 3D colored gradient shader Z (up) with center 
= 0.5
 
                        def colorCalc(real x) vec3 :
                            vec3(
                                # color1 = blue (0,0,0.75)
                                # color2 = green (0,0.75,0)
                                # compute color mix
 
                                #  color1 \/                      color2 \/
                                add(mul(0.00,x),mul(sub(1.0,x),0.00)),
                                add(mul(0.00,x),mul(sub(1.0,x),0.75)),
                                add(mul(0.75,x),mul(sub(1.0,x),0.00))
                            )
 
                        def eval(vec3 pos) vec3 :
                            colorCalc(div(mod(add(dotk(pos),div(0.1,2.0)),0.1),0.1))
                        ]]>
                    </shader>
                </shader>
            </albedo>
            <displacement>
                <shader>
                    <shader>
                        <![CDATA[   
 
                        # ~1/3pi sinus waves displacment U
 
                        def eval() real :
                            mul(sin(div(mod(doti(getTexCoords(0)),0.1),0.1)),20.0)
                        ]]>
                    </shader>
                </shader>
            </displacement>
        </diffuse>
    </material>

Deus · Post by **Deus** » Thu Aug 14, 2008 12:38 am

That was cool

1.1.9?

CTZn · Post by **CTZn** » Thu Aug 14, 2008 1:00 am

Wondering: let's consider two special, but practical cases:

I must use a given texture for whatever parameter, but:

Case A: the texture I have is of a too too low x-y resoulution,

Case B: the pixel depth in the texture is too low (ie 8bit for displacement)

Can you see, for both cases, a way to improve the textures (dithering, smoothing, whatever) ? Provided that we would have better precision in UV space than just texture space (better results with shader than by photogimping).

I hope I wasn't too unclear, just trying to be prgmatical

Already, thanks !

Shader tests

Who is online