Sky Color Modders?

[Bucic]

The answer is in the text.

You can take a screenshot (like you did), pick a probe of RGB values of the current color and compute the coefficients to multiply the output with.

1 in shader = 255, 0 in shader = 0.

You must have yourself a goal not to answer my question directly :D

OK then, I'm on it and I'll post the results (if I get to that point) later.

 

Edit:

So I guess generally

 

RGB_desired = RGB_original * coefficient

 

 

What I did:

 

sky_basic.dat:

mov oD0,r0

changed to:

def c23, 1.67, 1.25, 0.9, 1

mul r0, r0, c23

mov oD0,r0

 

sky_new.dat:

mul oD0,r0, c22

mov oD0, r0

changed to:

mul oD0,r0, c22

def c23, 1.67, 1.25, 0.9, 1

mul r0, r0, c23

mov oD0, r0

 

Effect: black sky = error, same with

 

sky_new.dat:

mul oD0,r0, c22

mov oD0, r0

changed to:

def c23, 1.67, 1.25, 0.9, 1

mul r0, r0, c23

mov oD0, r0

 

(spaces removed)

 

BTW, What is w in the coefficient definition?

 

 

In the meantime I chose the right colors for the first tests

Edited December 9, 2010 by Bucic

[SilentEagle]

To be honest Bucic, I had absolutely no clue what I was doing. I tried for days to figure out what this pixel shader color coding was about and just went with plain luck and trial & error. I'm not completely satisfied with the sky color Blaze and I have come up with, but until we figure out how to get exactly what we want, that's our best effort. Thanks to DarkWanderer for the information! Not sure it helps me convert rgb to code though. We'll see.

[aaron886]

Confusing as all get out. I'm no programmer, but I dabble with "higher-level" languages and no doubt I look at your explanation with a stupid look on my face, DarkWanderer.

[Bucic]

Confusing as all get out. I'm no programmer, but I dabble with "higher-level" languages and no doubt I look at your explanation with a stupid look on my face, DarkWanderer.

That's because, and I'm not being grumpy here, DarkWanderer's explanation is not sufficient in a technical sense. Take the w value of the coefficient for example.

 

Anyway, I think I got the whole coefficient multiply thing right, I've set up a spreadsheet for quick calculations of desired coefficients. This theoretically should allow me to precisely and deliberately control sky color. The only problem is that after the modifications I mentioned before I got black sky which usually indicates a syntax error in the shader, e.g. when you use a previously undeclared constant (coefficient).

Edited December 9, 2010 by Bucic

[DarkWanderer]

That's because, and I'm not being grumpy here, DarkWanderer's explanation is not sufficient in a technical sense. Take the w value of the coefficient for example.

Sure, it's always someone else who is at fault...

 

Here's a code with simple change:

vs.1.1
dcl_position v0

;
; Constants:
;
;  c0-c3  - View*Projection
;  c4-c6  - inverted View
;  c7     - C0.x, C0.y, SpanX, SpanY
;  c8     - (-2*h),1/R,Cz,(-4*(h^2-R^2)
;  c9     - 0.0,0.5,1.0,h
;  c10-c12- RGB conversion
;  c13    - LumA,LumB,LumC,LumD
;  c14    - LumE,Ax,Bx
;  c15    - Cx,Dx,Ex,Ay
;  c16    - By,Cy,Dy,Ey
;  c17    - 1/zenith_rad,1/x_rad_zenith,1/y_rad_zenith,0.3
;  c18    - lightVector
;  c19    - acos1,acos2,acos3,acos4
;  c20    - acos5,pi/2,log2(e),alt_adjust

def c22, 0.85, 0.85, 1.00, 1.00
;
;Set position in cam space:
mad r0.x,v0.x,c7.z,c7.x
mad r0.y,v0.y,c7.w,c7.y
mov r0.z,c8.z

;transform position to world space
dp3 r1.x,r0,c4
dp3 r1.y,r0,c5
dp3 r1.z,r0,c6
mov r1.w,c9.z

;normalize position:
dp3 r0.w,r1,r1
rsq r2.w,r0.w
mul r0,r1,r2.w
;;mov r0.w,c9.z	;1.0

;calc distance to sphere
mul r1.w,c8.x,r0.y	;-2h*p.y
mad r1.x,r1.w,r1.w,c8.w	;^2-4*(h^2-R^2)
mov r1.y,r1.x
rsq r1.x,r1.x
mul r1.y,r1.y,r1.x
add r1.w,r1.w,r1.y
mul r1.w,r1.w,c9.y

;get sphere point in world space

mul r2,r0,r1.w
mov r2.w,c9.z

;;transform and output
dp4 oPos.x, r2, c0
dp4 oPos.y, r2, c1
dp4 oPos.z, r2, c2
dp4 oPos.w, r2, c3
;----------------------------------------
;normalize world pos
dp3 r0.w,r2,r2
rsq r0.w,r0.w
mul r4,r2,r0.w     ;r4=normalized world pos
max r4.y,r4.y,c9.x
;
;cos gamma
dp3 r0.w,r4,c18
;***************************************
;acos gamma
;get |gamma|
mul r1.w,r0.w,r0.w
mov r2.w,r1.w
rsq r2.w,r2.w
mul r1.w,r1.w,r2.w ;r1.w=|gamma|

;calc z for |gamma|<0.5
mul r0.x,r0.w,r0.w		;r0.x=z for |gamma|<0.5
;calc z for |gamma|>0.5
add r0.y,c9.z,-r1.w
mul r0.y,r0.y,c9.y		;r0.y=z for |gamma|>0.5
;calc x for |gamma|>0.5
mov r1.x,r0.y
rsq r1.x,r1.x
mul r0.z,r0.y,r1.x		;r0.z=x for |gamma|>0.5

;set r1.x to 1 if |gamma|>=0.5
sge r1.x,r1.w,c9.y
;if (r1.x==1) r2.z=r0.y; else r2.z=r0.x //r2.z=r0.y*r1.x+(1-r1.x)*r0.x <=> r1.x*(r0.y-r0.x)+r0.x
add r1.y,r0.y,-r0.x
mad r2.z,r1.x,r1.y,r0.x

;if (r1.x==1) r2.x=r0.z;else r2.x=r1.w //r2.x=r0.z*r1.x+(1-r1.x)*r1.w <=> r1.x*(r0.z-r1.w)+r1.w
add r1.y,r0.z,-r1.w
mad r2.x,r1.x,r1.y,r1.w

;main_run(gamma)=(((( 4.2163199048E-2 * z+ 2.4181311049E-2)
;	          * z+ 4.5470025998E-2)
;                 * z+ 7.4953002686E-2)
;                 * z+ 1.6666752422E-1)
;                 * z
;                 * x+ x

mad r3.z,r2.z,c19.x,c19.y
mad r3.z,r3.z,r2.z,c19.z
mad r3.z,r3.z,r2.z,c19.w
mad r3.z,r3.z,r2.z,c20.x
mul r3.z,r3.z,r2.z
mad r3.z,r3.z,r2.x,r2.x

;;if (|gamma|>0.5) asin(|gamma|)=pi/2-2*r3.z; else asin(|gamma|)=r3.z
add r3.w,r3.z,r3.z
add r3.w,c20.y,-r3.w ;r3.w=pi/2-2*r3.z

;if (r1.x==1) r3.x=r3.w; else r3.x=r3.z //r3.x=r3.w*r1.x+(1-r1.x)*r3.z <=> r1.x*(r3.w-r3.z)+r3.z
add r1.y,r3.w,-r3.z
mad r3.x,r1.x,r1.y,r3.z

;if (gamma<0) r1.y=-1;else r1.y=+1
slt r1.y,r0.w,c9.x  ;0 if >=0, 1 if <0
slt r1.z,-r0.w,c9.x ;1 if >0,0 if <=0
add r1.y,r1.z,-r1.y ;1-0, 0-1

mul r3.x,r3.x,r1.y
add r3.x,c20.y,-r3.x ;r3.x=acos(gamma) in 30 commands
;**************************************************
;
;radiance = (1.0f+A*exp(B/Cos_theta)*(1.0f+C*exp(D*gamma)+E*cos_gamma*cos_gamma);
;e^y=2^(y*log2(e))
;--------------------
;r1.x=b/cos_theta
mul r1.y,r4.y,c20.w ;verpos.y*alt_adjust
rcp r1.x,r1.y
mul r1.x,r1.x,c13.y ;b/(verpos.y*alt_adjust)
;e^r1.x=expp(r1.x*c20.z)

mov r2.x,c9.z

mul r7.w,r1.x,c20.z
expp r7,r7.w
mad r1.x,r7.z,c13.x,r2.x
;add r1.x,r1.x,c9.z ;r1.x=(lumA*exp(B/cos_theta)+1)
;;;D*0.3*gamma
mul r1.y,r3.x,c17.w
mul r1.y,r1.y,c13.w
mul r7.w,r1.y,c20.z
expp r7,r7.w
mul r1.y,r7.z,c13.z        ;r1.y=lumC*exp(D*0.3*gamma)
;;E*cos_gamma*cos_gamma
mul r1.z,r0.w,r0.w
mad r1.z,r1.z,c14.x,r2.x
;add r1.z,r1.z,c9.z ;r1.z=lumE*cos_gamma*cos_gamma+1
add r1.y,r1.y,r1.z
mul r1.x,r1.x,r1.y
mul r1.y,r1.x,c17.x      ;r1.y=luminance
;-----------------------------------------
;r5.x=b/cos_theta
rcp r5.x,r4.y
mul r5.x,r5.x,c14.z ;Bx/verpos.y
;e^r5.x=expp(r5.x*c20.z)

mul r7.w,r5.x,c20.z
expp r7,r7.w
mad r5.x,r7.z,c14.y,r2.x ;r5.x=(Ax*exp(Bx/cos_theta)+1)
;;;Dx*gamma
mul r5.y,r3.x,c15.y
mul r7.w,r5.y,c20.z
expp r7,r7.w
mul r5.y,r7.z,c15.x        ;r5.y=Cx*exp(Dx*gamma)

;;Ex*cos_gamma*cos_gamma
mul r5.z,r0.w,r0.w
mad r5.z,r5.z,c15.z,r2.x ;r5.z=Ex*cos_gamma*cos_gamma+1
add r5.y,r5.y,r5.z
mul r5.x,r5.x,r5.y
mul r5.x,r5.x,c17.y      ;r5.x=CIE_x

;----------------------------------------
;-----------------------------------------
;r6.x=b/cos_theta
rcp r6.x,r4.y
mul r6.x,r6.x,c16.x ;By/verpos.y
;e^r5.x=expp(r5.x*c20.z)

mul r7.w,r6.x,c20.z
expp r7,r7.w
mad r6.x,r7.z,c15.w,r2.x ;r6.x=(Ay*exp(Bx/cos_theta)+1)
;;;Dy*gamma
mul r6.y,r3.x,c16.z
mul r7.w,r6.y,c20.z
expp r7,r7.w
mul r6.y,r7.z,c16.y        ;r6.y=Cy*exp(Dy*gamma)

;;Ey*cos_gamma*cos_gamma
mul r6.z,r0.w,r0.w
mad r6.z,r6.z,c16.w,r2.x ;r6.z=Ey*cos_gamma*cos_gamma+1
add r6.y,r6.y,r6.z
mul r6.x,r6.x,r6.y
mul r6.x,r6.x,c17.z      ;r6.x=CIE_y
;-----------------------------------------
rcp r1.w,r6.x
mul r1.x,r1.w,r5.x
mul r1.x,r1.x,r1.y

add r1.z,c9.z,-r5.x
add r1.z,r1.z,-r6.x
mul r1.z,r1.z,r1.w
mul r1.z,r1.z,r1.y

dp3 r0.x,r1,c10
dp3 r0.y,r1,c11
dp3 r0.z,r1,c12
min r0,r0,c9.z
mov r0.w,c9.x

;----------------------------------------
mul oD0, r0, c22

Here's the result:

 

P.S. W component is alpha-channel and is not used in LO sky shaders.

Edited December 9, 2010 by DarkWanderer

[Bucic]

Sure, it's always someone else who is at fault...

.

.

.

P.S. W component is alpha-channel and is not used in LO sky shaders.

You misunderstood me. It's not to be considered in 'who's fault was it' category. You're making us a huge favor by helping us getting rid of cartoonish sky and maybe you'll even become a community hero :) You just have your way of taking some things for granted others (even with C/C++ programming basics) couldn't possibly know. It generates more posts and I'm just afraid you'll quit on the subject leaving us to fail, when we are so close to the solution.

 

That's why I'm so pressing to get to the solution and I'm sorry if you felt urged by me.

 

Anyway, I'm off to test what you posted and I hope you'll bear with me just a little longer.

[Bucic]

OK, I've checked it out. I see that the only thing you've modified is

def c22, 0.57, 0.57, 0.6, 0.57 -> def c22, 0.85, 0.85, 1.00, 1.00

 

I'll try to get a nice orange sky to confirm that it works as intended and that we're not in number guessing game again.

 

EDIT:

Success!!!

 

 

And the shot below is an actual in-game shot

 

Check this one out for comparison

 

 

Unfortunately I can't resolve this problem with skycolor near the horrizon. I get white/yellow color for some reason :( As you can see the yelowish part is not affected by skyfog_b and skyfog_v which is reddish here.

 

Edited December 9, 2010 by Bucic

[aaron886]

Woahhh now we're talking! Keep it up DW and Bucic... don't give up on it! The sky color looks fantastic, if you can fix the horizon and "haze!"

[Boberro]

Honestly I think I like it as is without fixing white-yellow ;)

[aaron886]

I find the blood-red horizon a little disturbing. :D

 

EDIT: Darkwanderer, I noticed at the end, you removed "mov oD0, r0." (No idea what that actually would do of course...)

[Viper101]

The quest for the perfect sky color...

 

Great work, keep it up!:thumbup:

[Bucic]

The sky color will never be perfect because the graphics engine has gimmicks instead of solid foundations. We will tweak one thing and something will brake in other places. But as you could see before I will gladly take even plain white sky with nice blue haze just to never look at that default cartoonish blue sky ever again.

[DarkWanderer]

Bucic, could you post your current sky and fog shaders? I'll check it out.

[Bucic]

OK, I'll post them today, evening.

sky_basic.txt

sky_new.txt

Edited December 10, 2010 by Bucic

[Bucic]

Here's the latest version if someone is not repulsed by the yellowish horizon.

http://www.mediafire.com/file/lzqn4xwax3nagwk/SKY_3_dec14_bcc.7z

 

I have to wait for DW's response before attempting any further tests.

[LawnDart]

It would be neat if someone was able to create a shuffle/random skycolor mod (not sure exactly how you'd do that though). The sky never looks the same in the real world every day, and many of these sky color mods are good representations of various skies - but there's obviously never going to be a perfect match since it's an ever changing environment out there.

[SilentEagle]

Nice work! Couple questions...

 

1) Is that a slightly blue haze? If so, awesome!

2) Are you still working on how to change the horizon?

3) Why did you include so many files? Did you make modifications to all of these?

[Bucic]

It would be neat if someone was able to create a shuffle/random skycolor mod (not sure exactly how you'd do that though). The sky never looks the same in the real world every day, and many of these sky color mods are good representations of various skies - but there's obviously never going to be a perfect match since it's an ever changing environment out there.

It doesn't look the same with this mod either as it's not with the default one. I just target for a decent looking NOON summer, spring sky. Going after modding it correctly for more than two base conditions without a proper support (never given so far to sky modders) from ED I must say would be a waste of time. If anyone would like to pursue that madness :) be my guest. I'll tell you everything I know but count me out.

 

Nice work! Couple questions...

 

1) Is that a slightly blue haze? If so, awesome!

2) Are you still working on how to change the horizon?

3) Why did you include so many files? Did you make modifications to all of these?

1) Yes because slightly blue haze is an element made right in the default skyfog_b and skyfog_v, by accident :music_whistling: It's just a default haze and it looks more blue the more white you make the rest of the sky. In fact when DW responds I was planning to increase the blue value of haze/fog so that it looks blueish even when looked at from above, when the ground is in the background. As God intended. I chose the desert setting deliberately.

 

2) No. I have no leads whatsoever that would allow me to start any deliberate tryouts. I'm waiting for DW's response.

3) You can safely consider I mod only sky_basic.dat and sky_new.dat. I may upload the whole sky folder because that's how I handle it during tests. I juggle the whole folders around and alt tab the game from full screen which causes it to reload shaders.

Edited December 14, 2010 by Bucic

[SilentEagle]

Any progress or breakthroughs, Bucic? Can't wait to see haze like this!

 

[aaron886]

I would suggest emailing or PM'ing DW... seems he hasn't seen this thread lately. (Or perhaps I'm being too impatient.)

[T-Bone]

I created a "skyfix" for Falcon 4.0 many moons ago. I'm going to see if I can take a stab at this ;)

 

[aaron886]

Did that involve shader language, T-Bone? I remember your fix... it got some hours on my install. ;)

[T-Bone]

No... it was some RGB values coded for different times of day. Shaders weren't in games yet back in 1998 LOL

 

Which skyfix did you use? I had the platinum and v1-3, and Gold... yep lets just say I had too much spare time when I was younger ;)

 

I am guessing things are a bit more complicated these days.

[aaron886]

Much. I think you could call this lower-level shader programming.

 

I don't remember... but I remember it was one of yours!

[lucioperca]

I used it too

 

T Bone platinum I presume