Supermodel Forum

by **HarryTuttle** » Fri Jun 16, 2017 8:40 am

My first approach was that fixed shading doesn't automatically exclude lightning, so that doesn't take in consideration ph.LightEnabled() value. For what is saw the contrary is true.

So I tested fixed shading polys in both illuminated and self-illuminated cases. The values that I found seem to always work.

My only problem is how similar are in respect to the arcade counterpart. We know that sometime (L.A. M.) we have incorrect viewport sun light values and only in some scenes. Well, that will influence fixed shading when using ambient values.

We also know that some object in the same viewport has scaled down normals, the "x" component of which could give an erroneous value for fixed shading.

EDIT:
I forgot to say an important thing: every value I'm talking about is the "shade" one, that is the value that is later multiplied by the polygon color.

by **Ian** » Sat Jun 17, 2017 2:33 am

HarryTuttle wrote:Some before / after comparison:

This is LA machin guns running under dolphin

It's quite a good colour match since it's not recorded off CRT or something funky .. Seems to look pretty close to current supermodel. Although the actual arcade could be different. Can't find any good quality footage.

by **HarryTuttle** » Sun Jun 18, 2017 3:44 am

by **Ian** » Sun Jun 18, 2017 4:46 am

This is the best arcade quality footage I found

by **Ian** » Sun Jun 18, 2017 1:22 pm

Harry,
have you ever looked at the fixed shading in this part of scud? I think it might gives us some clues to the missing offset
Have close look at the brightness of the wall, it changes as the car goes around the corner.

Bright wall

Suddenly gone darker

Side by side comparison

Video of it here
https://www.youtube.com/watch?v=qrJkQ9e9uwQ#t=11m22s

This offset can't be in the poly headers themselves as they are (I think ROM models and are fixed)
So it must be something either in the culling nodes (possibly) or the viewport itself (more likely).
This change in brightness here has puzzled me for a long time ..

by **HarryTuttle** » Mon Jun 19, 2017 9:25 am

Ian wrote:Harry,
have you ever looked at the fixed shading in this part of scud?

Yes. And that part exactly was one of my reference to find the right fixed shading values. In particular I watched the lightning of the wall, the road and the inclined plane between them.

The variation is caused by sunlight dynamically changing direction as you pass by. I remember the same happens on the first course after the dolphin tunnel.

Anyway that video is from Abe, he has also uploaded on Mega the much higher quality version. He posted the link for all his videos somewhere in this forum.

by **Ian** » Tue Jun 20, 2017 6:50 am

Gonna do some more digging tonight
But you are probably right that the viewport ambient plays a role. It makes no sense (to me) that it does this. But i've explored almost every other option.
I did a binary diff on the viewports before and after the lighting changed in scud, and besides the light angle and brightness values, only the fog value changed. The whole castle section in scud that clearly uses some different fixed shading values, just so happens to exist in its own viewport, which it's own lighting values.

It's possible something in the culling nodes might effect fixed shading, but probably unlikely

by **HarryTuttle** » Wed Jun 21, 2017 2:09 am

HarryTuttle wrote:The variation is caused by sunlight dynamically changing direction as you pass by.

My bad, after a second check I have to correct that statement, there's no variation in that particular case (the wall). I think that's an artifact caused by the capture hardware.

I remember that part gave me much headache and it was hard to find the right values for fixed shading. But at least it seems that for step 1.5 hw the formula for the "shade" component is:

(fixed shaded polys)
- illuminated = ambient + (float)((UINT8)(ix & 0x7F)) / 127.0f
- self-illuminated = 2.0f

(sun shaded polys)
- self-illuminated = 1.5f

by **Ian** » Wed Jun 21, 2017 10:06 am

Have a go with my current build. It's still not finished, I need to figure out exactly what the hw is doing with regards to vp ambient with 2.x hw.
But currently it's as close as I can get it

by **Ian** » Thu Jun 22, 2017 3:08 am

I have a feeling with step 2.x hw that the viewport ambient with unsigned values is simply the min value

so unsigned

shade = std::max(vpAmbient,shade)

signed version

shade = std::max(1-vpAmbient, shade)

Needs more testing ..

Supermodel Forum

[Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Re: [Patch] Fixed Shading

Who is online