New SSF :D

vbt

Staff member
There is a new SSF release
smile.gif
here

It should be faster but on my PC it displays nothing and no options can be changed. Also now there is a pad support
biggrin.gif
 
Isn't SSF the one that runs in pure software? In which case, it should end up being better than GiriGiri, since it'll avoid all problems with trying to pass quads through a triangle setup engine
smile.gif
 
It should, but not until the author implemenst some ASM or Dynarec for some major speed ups. I would like to see full hardware d3d/ogl rendering though. Girigiri is only partly d3d.

Satourne is full ogl, but it's compatibility is lacking.
 
I think part of SSF is ASM but the latest release has no graphic output as the author says, also why would d3d or opengl help saturn emulators when the original saturn had no 3d acceleration?
 
Originally posted by JonnyCasino@Jul 15, 2003 @ 12:12 AM

Why would d3d or opengl help saturn emulators when the original saturn had no 3d acceleration?

Have you *ever* seen a PSX game running on EpsxE with a HW accelerated video plugin at 1600x1200, with 4X antialiasing and bilinear filtering on?

To put it simple: instead of using software engine code to plot pixels when drawing distorted quads, they can just send them over an OGL/D3D interface and allow the game to be played at whatever resolution you wish. It also allows the usage of bilinear filtering.

Some shots of GiriGiri in action

The previous SSF release, while still very slow, was a bit faster than the older ones, and the compatibility has gone through the roof. Also the graphic output was 99% identical to a real Saturn's. The only difference is the gouraud shading, that isn't blocky as the Saturn's.

They pretty much got very advanced at core and graphics emulation, and are now moving into sound. Then they can speed it up, and we'll have a hell of solid emulator.
 
Originally posted by M3d10n@Jul 14, 2003 @ 07:38 PM

Have you *ever* seen a PSX game running on EpsxE with a HW accelerated video plugin at 1600x1200, with 4X antialiasing and bilinear filtering on?

To put it simple: instead of using software engine code to plot pixels when drawing distorted quads, they can just send them over an OGL/D3D interface and allow the game to be played at whatever resolution you wish. It also allows the usage of bilinear filtering.

Uh... no.

Saturn uses Quads.

Direct3D and OpenGL use Triangles.

They both support "Quads" but they get converted to triangles in setup.

Therefore, any Saturn emulator which uses hardware acceleration... will not look right in some Saturn games. Any Saturn games that use curved quads will end up having a weird 'folded' look.

The thing is, if the four vertices of a quad aren't coplanar, the quad will have a nice, even curve to it. If you convert that to triangles, you'll get two flat triangles. Whoopity-doo.

Any pure-software Saturn emulator will end up more compatible and accurate than a hardware-accelerated one.

And another thing. Bilinear filtering CAN be done in software, thank you very much, it's just SLOW.
 
You can do quads using triangles and texture coordinates without any problems. Also, since all Saturn polygons are 2D they are coplanar by definition. Furthermore, not all OpenGL/D3D accelerators use triangles (eg. 1st-gen PowerVR) even though most do.
 
You can do quads using triangles and texture coordinates without any problems. Also, since all Saturn polygons are 2D they are coplanar by definition. Furthermore, not all OpenGL/D3D accelerators use triangles (eg. 1st-gen PowerVR) even though most do.

Nnnnnope. Shining Force 3 uses curved quads all over the terrain. Most obvious example is on the sides of mountains, especially the really short ones... put a character's movement cursor over it and look at the alignment of the four little corner arrows. Rotate the camera. They aren't coplanar.

I wasn't aware that the PowerVR PCX cards could use Quads. Please link me to any site which lists that they do.

Also old nVidia cores - NV1 and NV2, i.e. before Riva128 - used Quads exclusively, but they don't support OGL or D3D at all.
 
Since VDP1 takes only 2D screen ccordinates, how do you propose to draw these non-coplanar polygons of yours? You're probably looking at two triangles sharing two vertices.

PowerVR isn't (wasn't) quad-based, but plane-based. It's been years since I read up on this, but IIRC it was based on taking a plane that stretched infinitely in all directions and then cutting away everything that wasn't in the polygon you wanted. It's not too easy to provide references since most of the stuff available then has disappeared and the official docs were never made public. However, see this to start with. There have been other technologies as well - nVidia's NV1 was based around quadratic patches (not quads) - but mostly due to Direct3D being triangle-based they have all disappeared from the desktop market.
 
Ah, yes, the INCREDIBLY weird implementation... infinite planes are fun.

And are you sure Saturn isn't capable of curved quads? I know I've seen "CONVEX quads" mentioned in at least one Saturn tech doc, and I'm sure those mountains are curved in SF3. Take a look yourself some time...
 
Yes, I'm sure. VDP1 can do convex, concave and even crossed quads, but is completely 2D.
 
i love when you guys start talking like this. and im not saying it to be asmart ass. sometimes i just sit here and try to figure out everything you guys are saying....usually doesnt work but always a good read :cheers
 
Originally posted by antime@Jul 15, 2003 @ 08:38 AM

Yes, I'm sure. VDP1 can do convex, concave and even crossed quads, but is completely 2D.

2D, 3D, that's not what he's talking about.

When the saturn draws a convex or a crossed quad, the convex edges are drawn curved, instead of simply being distorted. Thus one can simulate partially curved surfaces in a 3D game using non-coplanar vertices, since those will result in a convex 2D quad when viewed from certain angles (the SF3 small mountains).

But, yeah, when you use 3D acceleration, everything et's converted into two triangles, and there is the evil PSX texture warping effect (but they could easily fix that by using perspective correction). Just look at the screenshots I posted. So the curved quads would not be possible, but very *few* games uses them, and the visual impact will be minimal, when compared to the fact that the game is running with a much higher resolution.

I need to put some Saturn Quake screenshots online, taken with GiriGiri. It looks almost the same as GL Quake.
 
Originally posted by M3d10n@Jul 15, 2003 @ 10:58 AM

and there is the evil PSX texture warping effect (but they could easily fix that by using perspective correction).

Perspective correction needs z-values to work. The psx doesn't have a z-buffer or use z-values which is why Perspectice correction isn't possible in current psx emulators. The only way would be to HLE the psx, but that would mean compatibility would be very low.

Sub Pixel Accuracy also needs a z-buffer. This would alleviate all the wobbly polygon edges in Higher resolution psx emulation.

Does the saturn have a z-buffer or use z-values? It would be possible for perspective correction if it did.
 
Originally posted by fivefeet8@Jul 15, 2003 @ 07:43 PM

Does the saturn have a z-buffer or us z-values? It would be possible for perspective correction if it did.

Since it is just as 2D as the Playstation, no. (As I believe I've mentioned here a few times.)
 
Originally posted by Tagrineth+Jul 14, 2003 @ 08:31 PM--><div class='quotetop'>QUOTE(Tagrineth @ Jul 14, 2003 @ 08:31 PM)</div><div class='quotemain'> <!--QuoteBegin-M3d10n@Jul 14, 2003 @ 07:38 PM

Have you *ever* seen a PSX game running on EpsxE with a HW accelerated video plugin at 1600x1200, with 4X antialiasing and bilinear filtering on?

To put it simple: instead of using software engine code to plot pixels when drawing distorted quads, they can just send them over an OGL/D3D interface and allow the game to be played at whatever resolution you wish. It also allows the usage of bilinear filtering.

Uh... no.

Saturn uses Quads.

Direct3D and OpenGL use Triangles.

They both support "Quads" but they get converted to triangles in setup.

Therefore, any Saturn emulator which uses hardware acceleration... will not look right in some Saturn games. Any Saturn games that use curved quads will end up having a weird 'folded' look.

The thing is, if the four vertices of a quad aren't coplanar, the quad will have a nice, even curve to it. If you convert that to triangles, you'll get two flat triangles. Whoopity-doo.

Any pure-software Saturn emulator will end up more compatible and accurate than a hardware-accelerated one.

And another thing. Bilinear filtering CAN be done in software, thank you very much, it's just SLOW. [/b][/quote]

any pure software saturn emulator will also prolly run at 5fps
 
Originally posted by Jurai@Jul 15, 2003 @ 10:46 PM

any pure software saturn emulator will also prolly run at 5fps

No you're wrong it only depends on the power of the CPU used to run the emu.
 
..and how well optimized it is. I doubt the VDP emulation is really the taxing one.

Also... just because it's rendered via software, this doesn't mean it needs to be 100% purely emulated in order for the emulator to run without glitches.

One can write a software renderer, with the features it needs to mimic the Saturn video features, and simply use that renderer much like GiriGiri uses a D3D renderer, but since the renderer is custom made, it can offer the features a HW renderer would lack, like curved quads, infinite planes and other things. Plus it wouldn't be *that* slow.

SSF sorta works that way, since the gouraud shading looks different. That hints that it isn't fully emulating the VDP for drawing, but routing the drawing calls to a custom software renderer.
 
Back
Top