|
Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
#246490 - 02/14/11 05:34 AM
|
|
|
Since the last update, I've been spending a lot of time looking at lighting and texture modulation in Scud Race. I've figured out some important things but there are still more questions than answers. Overall, various shading effects look better than they did before, and the changes have also generally benefited Sega Rally 2 and Virtual On 2. Elsewhere, they may be causing problems.
I've also discovered the source of some of the missing geometry in Scud Race. The tiki head and staircases on the mountain course, the ship on the beginner course, and the circular tower on the night course are all visible now.
I have a better understanding of the source of some annoying graphical glitches in Sega Rally 2 relating to polygon RAM. I'm not yet sure how to fix them but similar problems are also showing up in some other games I added over the weekend, so now that I have more data to compare against, I'm hoping that a precise solution will become apparent.
I added Le Mans 24 and got it up and running. It is very playable but suffers from some graphical problems that I suspect are related to polygon color and fog modulation.
So far, Le Mans 24 holds the dubious distinction of being the ugliest Model 3 game. It appears to have been released after Scud Race and features many of the same vehicles, making its sub-par quality all the more noticeable. It would have made an average-looking Model 2 game.
Today, I hooked up Fighting Vipers 2. It runs pretty well on the whole but it appears to be referencing objects in VROM that are not actually models. The result is that very large polygons are sent to my renderer, dramatically slowing the game down. This is loosely related to some issues I've encountered in Virtual On 2 during boot-up and Sega Rally 2.
Star Wars Trilogy mostly looks nice but I haven't spent very much time playing with it. I'll return to it later down the road.
And last, but certainly not least: about those lighting effects I mentioned at the beginning...
Look closely at where the vehicle is hitting the wall. See the pale yellow circle? My top priority right now is to figure out the Model 3's spotlight effect. The devil's in the details
Edited by Bart Trzynadlowski (02/14/11 06:24 AM)
|
Bart
|
|
iq_132 |
FBA Dev.
|
|
|
Reged: 10/01/04
|
Posts: 150
|
Loc: Michigan, USA
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246492 - 02/14/11 05:58 AM
|
|
|
> Since the last update, I've been spending a lot of time looking at lighting and > texture modulation in Scud Race. I've figured out some important things but there are > still more questions than answers. Overall, various shading effects look better than > they did before, and the changes have also generally benefited Sega Rally 2 and > Virtual On 2. Elsewhere, they may be causing problems. > > I've also discovered the source of some of the missing geometry in Scud Race. The > tiki head and staircases on the mountain course, the ship on the beginner course, and > the circular tower on the night course are all visible now. > > > I have a better understanding of the source of some annoying graphical glitches in > Sega Rally 2 relating to polygon RAM. I'm not yet sure how to fix them but similar > problems are also showing up in some other games I added over the weekend, so now > that I have more data to compare against, I'm hoping that a precise solution will > become apparent. > > > I added Le Mans 24 and got it up and running. It is very playable but suffers from > some graphics problems that I suspect are related to polygon color and fog > modulation. > > > So far, Le Mans 24 holds the dubious distinction of being the ugliest Model 3 game. > It appears to have been released after Scud Race and features many of the same > vehicles, making its sub-par quality all the more noticeable. It would have made an > average-looking Model 2 game. > > Today, I hooked up Fighting Vipers 2. It runs pretty well on the whole but it appears > to be referencing objects in VROM that are not actually models. The result is that > very large polygons are sent to my renderer, dramatically slowing the game down. This > is loosely related to some issues I've encountered in Virtual On 2 during boot-up and > Sega Rally 2. > > > Star Wars Trilogy mostly looks nice but I haven't spent very much time playing with > it. I'll return to it later down the road. > > > And last, but certainly not least: about those lighting effects I mentioned at the > beginning... > > > Look closely at where the vehicle is hitting the wall. See the pale yellow circle? My > top priority right now is to figure out the Model 3's spotlight effect. The devil's > in the details
Wow! Nice!
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246495 - 02/14/11 06:03 AM
|
|
|
very interesting this development, and graphics and details are emerging and looks that are a good way
congratulations
|
|
|
Stiletto |
They're always after me Lucky ROMS!
|
|
|
Reged: 03/07/04
|
Posts: 6472
|
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246502 - 02/14/11 07:51 AM
|
|
|
> Star Wars Trilogy mostly looks nice but I haven't spent very much time playing with > it. I'll return to it later down the road.
Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG compressed and thus not coming soon to MAME?
- Stiletto
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Stiletto]
#246506 - 02/14/11 09:46 AM
|
|
|
Thanks for the update Bart. It's appreciated.
|
|
|
Cable |
retro gamer
|
|
|
Reged: 08/30/08
|
Posts: 131
|
Loc: UK
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246508 - 02/14/11 11:53 AM
|
|
|
Every bit of progress you post is great news! I might not fully understand some the technical details, but the updates are really appretiated. I cant wait to get my hands on star wars trillogy again. God knows how much money ive plowed into that game over the years
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Stiletto]
#246509 - 02/14/11 03:02 PM
|
|
|
> > Star Wars Trilogy mostly looks nice but I haven't spent very much time playing with > > it. I'll return to it later down the road. > > Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG > compressed and thus not coming soon to MAME? > > > - Stiletto
It sounds like MPEG-1 layers 1,2 patents have expired so it should be possible assuming Model 3 didn't use mp3 compression. Which is still patented according to this article.
http://www.kuro5hin.org/story/2008/7/18/232618/312
|
|
|
Foxhack |
Furry guy
|
|
|
Reged: 01/30/04
|
Posts: 2409
|
Loc: Spicy Canada
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Malice]
#246513 - 02/14/11 03:58 PM
|
|
|
> > > Star Wars Trilogy mostly looks nice but I haven't spent very much time playing > with > > > it. I'll return to it later down the road. > > > > Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG > > compressed and thus not coming soon to MAME? > > > > > > - Stiletto > > It sounds like MPEG-1 layers 1,2 patents have expired so it should be possible > assuming Model 3 didn't use mp3 compression. Which is still patented according to > this article. > > http://www.kuro5hin.org/story/2008/7/18/232618/312
One, that guy is not a lawyer. Two, those companies have more money than God and will stop at nothing to prevent people from messing with their codec if they're not going to profit out of it.
I'd err on the side of caution and not include it, just in case.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Foxhack]
#246514 - 02/14/11 04:18 PM
|
|
|
> > > > Star Wars Trilogy mostly looks nice but I haven't spent very much time playing > > with > > > > it. I'll return to it later down the road. > > > > > > Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG > > > compressed and thus not coming soon to MAME? > > > > > > > > > - Stiletto > > > > It sounds like MPEG-1 layers 1,2 patents have expired so it should be possible > > assuming Model 3 didn't use mp3 compression. Which is still patented according to > > this article. > > > > http://www.kuro5hin.org/story/2008/7/18/232618/312 > > One, that guy is not a lawyer. > Two, those companies have more money than God and will stop at nothing to prevent > people from messing with their codec if they're not going to profit out of it. > > I'd err on the side of caution and not include it, just in case.
not to mention that also non-US patents will have to expire if we want to avoid making MAME a US-only emu...
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246519 - 02/14/11 06:06 PM
|
|
|
"I suspect are related to polygon color and fog modulation"...
Looks like it's missing vertex color to my opinion... if model2 had vertex color (?) This would smooth ie the girl's leg on the pics if vertex color is used... I know model 1 didn't have vertex colors, i'm quite sure model 3 had... but model 2...?
Vertex color was already used on latest Playstation 1 games iirc...
Anyway, very good progress, two thumbs up
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: MAMu_]
#246522 - 02/14/11 06:21 PM
|
|
|
Those aren't the graphical glitches that I was referring to but you bring up a good point. Supermodel is effectively performing flat shading right now. The renderer supports vertex normals, of course, but I don't know how to compute them. Model 3 only stores polygon normals. In most situations, such as with inanimate objects, it's not noticeable, but on human figures, it is glaringly obvious.
Here is a screenshot of Lost World from the actual machine, courtesy of Abelardo who has been supplying me with very useful videos of this game and Scud Race to compare against:
Now compare with flat-shaded Supermodel:
Ugly, right? It's been really bugging me. Using a slightly different way to compute lighting, similar to how specular lighting is usually done, I can do a little better:
But still, the fundamental problem is missing information. A polygon normal is not sufficient. What if I search for all faces with the same vertex and interpolate the normals? The result looks perfect:
The Le Mans girls will look perfect as well using this method and even the cars in-game start to look a lot nicer. Unfortunately, this calculation is O(n^2) in polygons, and is therefore unacceptably slow. I refuse to believe that the Model 3 actually did this. But look how close it looks to the actual thing!
Looking at polygon dumps, I don't see any per-vertex attributes that could control shading or sharing of polygon normals.
It's a frustrating problem, to say the least.
Ignore the FPS, by the way. I'm developing on an old laptop with integrated graphics. Supermodel seems to run full speed on any reasonable new hardware with a dedicated video card.
Edited by Bart Trzynadlowski (02/14/11 06:25 PM)
|
Bart
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246527 - 02/14/11 07:25 PM
|
|
|
> Those aren't the graphical glitches that I was referring to but you bring up a good > point. Supermodel is effectively performing flat shading right now. The renderer > supports vertex normals, of course, but I don't know how to compute them. Model 3 > only stores polygon normals. In most situations, such as with inanimate objects, it's > not noticeable, but on human figures, it is glaringly obvious. > > Here is a screenshot of Lost World from the actual machine, courtesy of Abelardo who > has been supplying me with very useful videos of this game and Scud Race to compare > against: > > > Now compare with flat-shaded Supermodel: > > > Ugly, right? It's been really bugging me. Using a slightly different way to compute > lighting, similar to how specular lighting is usually done, I can do a little better: > > > But still, the fundamental problem is missing information. A polygon normal is not > sufficient. What if I search for all faces with the same vertex and interpolate the > normals? The result looks perfect: > > > The Le Mans girls will look perfect as well using this method and even the cars > in-game start to look a lot nicer. Unfortunately, this calculation is O(n^2) in > polygons, and is therefore unacceptably slow. I refuse to believe that the Model 3 > actually did this. But look how close it looks to the actual thing! > > Looking at polygon dumps, I don't see any per-vertex attributes that could control > shading or sharing of polygon normals. > > It's a frustrating problem, to say the least. > > Ignore the FPS, by the way. I'm developing on an old laptop with integrated graphics. > Supermodel seems to run full speed on any reasonable new hardware with a dedicated > video card.
Sure, last pic seems far better
Can't you find anywhere in roms a sort of list with xyz coordinates, or just a list of unique numbers, followed by something that could be RVB colors information anywhere...? As each vertex on a 3d model can be listed as a list of coordinates, or just a number (vertex 1, vertex 2, etc..), with a color information, these informations can be stored elsewhere than in the model itself.. I know we used it when working at psygnosis, and usually we used a special plug in in 3ds max to vertex color the models, and these informations were exported with our plug elsewhere than in the exported 3ds model. One file had modelisation informations (xyz, maps, color of faces), the other had only vertex colors in. I was graphist, not programer, but I hope my poor explanations in my poor english can help.
Regards And again, congrats on the progress
|
|
|
drewcifer |
One bad Mutha-(shut yo' mouth!)
|
|
|
Reged: 07/01/04
|
Posts: 428
|
Loc: Sweden
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: MAMu_]
#246536 - 02/14/11 08:30 PM
|
|
|
> I was graphist, not programer, but I hope my poor explanations in my poor english can > help.
Good thoughts, Mamu, however, I think Bart's right - he's looking for per-vertex normals in this case, probably not per-vertex colors:
I think a message to Mr. Naibo might help clear things up
http://www.mameworld.info/ubbthreads/sho...part=1&vc=1
Andrew
* Waves at Bart * Congrats on getting yourself some more free time, and great progress! A fresh look is always a nice thing.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: drewcifer]
#246550 - 02/14/11 10:29 PM
|
|
|
i always loved those blocky,low poly boobs and asses..way to go Bart
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: drewcifer]
#246565 - 02/14/11 11:29 PM
|
|
|
I hadn't seen that post, very interesting indeed! I've corresponded with Naibo a few times regarding some other things but it seems he's a pretty busy guy and I haven't heard from him in a while.
If you're reading this Naibo, some information on the vertex normal problem would be very helpful
I'm now 99.999% certain that the mesh data itself does not contain explicit vertex normals. When Naibo says "tricky", I take it to mean that they are computed automatically in some manner based on the polygon normals (otherwise it would seem to be a tremendous waste of memory to store polygon normals with such precision merely for backface culling).
There doesn't seem to be any explicit linkage information either, other than what already exists for polygon strips (allowing only the previous polygon's vertices to be reused, which is not enough for computing normals). What I'm looking into now is whether there might be an assumed fixed ordering of polygons, with a limit to how many faces a vertex can share. I highly doubt it, though.
Edited by Bart Trzynadlowski (02/14/11 11:31 PM)
|
Bart
|
|
CptGuapo |
Beat'em-ups Lover
|
|
|
Reged: 03/18/08
|
Posts: 342
|
Loc: Off to Never Never Land
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246586 - 02/15/11 01:28 AM
|
|
|
Quote:
Today, I hooked up Fighting Vipers 2. It runs pretty well on the whole but it appears to be referencing objects in VROM that are not actually models. The result is that very large polygons are sent to my renderer, dramatically slowing the game down. This is loosely related to some issues I've encountered in Virtual On 2 during boot-up and Sega Rally 2.
Hmmm...these are awesome news, indeed. I'm looking forward to see this baby emulated...
Good times...My gratitude to you, Bart, and your phenomenal effort.
Edited by CptGuapo (02/15/11 01:29 AM)
|
"Mythology is what grownups believe, folklore is what they tell children and religion is both."
|
|
AnimalBear |
100% MAME Fan Forever!
|
|
|
Reged: 04/29/10
|
Posts: 206
|
Loc: Portugal, Porto
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CptGuapo]
#246591 - 02/15/11 02:47 AM
|
|
|
Great News man Good Job. I Believe you can get the trick to solve that errors. I Believe in you and i believe anyone can do this.
GO! Model 3! Go!
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246610 - 02/15/11 05:36 AM
|
|
|
Nice job emulating that Olmec Head in the Medium course in Scud Race
|
|
|
Moose |
Don't make me assume my ultimate form!
|
|
|
Reged: 05/03/04
|
Posts: 1483
|
Loc: Outback, Australia
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Foxhack]
#246619 - 02/15/11 10:03 AM
|
|
|
> > > > Star Wars Trilogy mostly looks nice but I haven't spent very much time playing > > with > > > > it. I'll return to it later down the road. > > > > > > Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG > > > compressed and thus not coming soon to MAME? > > > > > > > > > - Stiletto > > > > It sounds like MPEG-1 layers 1,2 patents have expired so it should be possible > > assuming Model 3 didn't use mp3 compression. Which is still patented according to > > this article. > > > > http://www.kuro5hin.org/story/2008/7/18/232618/312 > > One, that guy is not a lawyer.
LOL, I was trawling through some old PC magazines a few nights ago, and by pure random chance stumbled across an old "Tron Movie is Banned" report on the kuro5hin web site (because the movie shows the inside of a US nuclear reactor and Homeland Security are gettin' twitchy). It was purpotedly to have been written Lirpa Sloof, so even disregarding some of the other give aways, it was dead easy to spot the joke (but apparently quite a few swallowed the story hook, line, and sinker).
|
Moose
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246621 - 02/15/11 01:43 PM
|
|
|
Yeah, girls should look little more sexier
|
|
|
AnimalBear |
100% MAME Fan Forever!
|
|
|
Reged: 04/29/10
|
Posts: 206
|
Loc: Portugal, Porto
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246623 - 02/15/11 04:31 PM
|
|
|
> Yeah, girls should look little more sexier
Hope we see a uncensored version Oooops.....
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246634 - 02/15/11 05:58 PM
|
|
|
Nice work! Mind sharing where the polygon normals are hidden? Or is smooth shading achieved by an averaging of face normals, under control of bits in the polygon header?
|
Bart
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246642 - 02/15/11 06:50 PM
|
|
|
yep, you right, hardware calculated average of face normals, just use CalculateNormals
is there any difference in polygon lists in vipers/lostworld? looks like it doesn't transfer at all... display lists are empty here...
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246643 - 02/15/11 06:53 PM
|
|
|
Thanks. What do you mean by CalculateNormals()? Are you performing an O(n^2) calculation of face normals?
|
Bart
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246644 - 02/15/11 06:58 PM
|
|
|
you got one normal per 3-4 vertexes, for each different poly... so you can calc average then
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246645 - 02/15/11 07:00 PM
|
|
|
No, Lost World and Vipers are the same (well, Vipers _is_ referencing some things that aren't models -- have you seen the same thing in Sega Rally?)
Lost World boots without any patches over here and Fighting Vipers only needs a patch to some weird decrementer loop (seems a lot of games have this issue during boot up). Do the games start at all for you? Given that both games use different DMA hardware, the problems could be unrelated. I'll need more info to help you troubleshoot.
Edited by Bart Trzynadlowski (02/15/11 07:03 PM)
|
Bart
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246646 - 02/15/11 07:02 PM
|
|
|
Right, but how do you know which polygons a vertex touches? Unless there is connectivity information specified per-vertex (or in some other, fixed way), you'd have to scan the entire list of vertices for every vertex.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246647 - 02/15/11 07:09 PM
|
|
|
they boots but no geometry transfered, all display lists are empty... only 2D is output... seems strange, but maybe we missed something like banking or there is bug somewhre...
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246649 - 02/15/11 07:14 PM
|
|
|
I doubt it has anything to do with banking -- those games don't do anything out of the ordinary. The place I would look first if I were you is IRQ handling. Some games can be fussy about that; it's especially important for games using the SCSI controller for DMA.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Moose]
#246655 - 02/15/11 07:29 PM
|
|
|
> > > > > Star Wars Trilogy mostly looks nice but I haven't spent very much time > playing > > > with > > > > > it. I'll return to it later down the road. > > > > > > > > Eagerly anticipating this one. Plans to handle the soundtrack, IIRC it is MPEG > > > > compressed and thus not coming soon to MAME? > > > > > > > > > > > > - Stiletto > > > > > > It sounds like MPEG-1 layers 1,2 patents have expired so it should be possible > > > assuming Model 3 didn't use mp3 compression. Which is still patented according to > > > this article. > > > > > > http://www.kuro5hin.org/story/2008/7/18/232618/312 > > > > One, that guy is not a lawyer. > > LOL, I was trawling through some old PC magazines a few nights ago, and by pure > random chance stumbled across an old "Tron Movie is Banned" report on the kuro5hin > web site (because the movie shows the inside of a US nuclear reactor and Homeland > Security are gettin' twitchy). It was purpotedly to have been written Lirpa Sloof, so > even disregarding some of the other give aways, it was dead easy to spot the joke > (but apparently quite a few swallowed the story hook, line, and sinker).
Well that site may not have been the best to link to. But it poped up on Google first and i was being lazy.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246660 - 02/15/11 08:03 PM
|
|
|
D3DXComputeNormals doing all this work actually it's easy enough when meshes objects are used...
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CaH4e3]
#246666 - 02/15/11 08:38 PM
|
|
|
Ah, I don't have this luxury in OpenGL It's easy enough to do, but computationally expensive. For static objects, it's not a problem, but for dynamic meshes, doing it each frame is prohibitively slow. Dynamic meshes often have the highest polygon counts in a given scene. Of course, there are ways to be clever
I'm still not convinced that this is how Model 3 actually computes its normals. It doesn't have mesh caching hardware and seems to lack adjacency data. I wonder how it computed the normals so quickly each frame?
|
|
|
drewcifer |
One bad Mutha-(shut yo' mouth!)
|
|
|
Reged: 07/01/04
|
Posts: 428
|
Loc: Sweden
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246671 - 02/15/11 09:09 PM
|
|
|
> Ah, I don't have this luxury in OpenGL It's easy enough to do, but computationally > expensive. For static objects, it's not a problem, but for dynamic meshes, doing it > each frame is prohibitively slow. Dynamic meshes often have the highest polygon > counts in a given scene. Of course, there are ways to be clever > > I'm still not convinced that this is how Model 3 actually computes its normals. It > doesn't have mesh caching hardware and seems to lack adjacency data. I wonder how it > computed the normals so quickly each frame?
Agreed, and unless it has a per-vert (huh huh huh, pervert) "smoothing" flag somewhere, it seems unlikely that the hardware is incapable of rendering hard edges. 3d hardware that can't render a proper cube seems pretty out there, even for Sega .
Andrew
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: drewcifer]
#246675 - 02/15/11 09:29 PM
|
|
|
Right. I haven't looked for it yet, but there is certainly a setting that controls shading -- flat shading is supported by the hardware. I do have some cube and sphere data to look at. The cubes would be flat shaded and the spheres smooth shaded, so I may be able to use them to figure out the shading parameters.
Here's what I do know:
- There is a crude vertex sharing capability, but it only allows polygons to reuse vertices from the previous poly. The obvious thing to do is to average the normals for shared vertices, but this hardly has any effect.
- Separately-stored adjacency matrices and vertex normals can be ruled out (or does anyone have information to the contrary?)
- It is not sufficient to search only polygons within a limited window (say +/- 32 polys) to detect adjacent faces. The whole model has to be scanned for smooth shading to look right in most cases.
- As far as I can tell, although there may be a bit or two that controls shading in the polygon header, there really isn't anything that indexes vertex normals.
- Likewise, there is no room for vertex normals in the vertex data itself, which consists of 4 32-bit words: x, y, z, and u and v (16 bits each). I thought it could be hiding in the lower 3 bits of the U and V data, because I once observed that ignoring the 3-bit fraction has no noticeable effect on texturing, but I'm starting to become convinced that it is totally unrelated to normals or polygon linkage.
Lost World has a rippling water effect in the intro which requires smooth shading and requires the entire mesh to be scanned. Might be a good test case because it's guaranteed to have smooth shading enabled.
EDIT: At one point, I thought it was strange that polygon normals are in 2.22 format. It seems like it could be done with far less precision. Logically, I looked here for possible clues, but it doesn't seem like there's anything in there...
Edited by Bart Trzynadlowski (02/15/11 09:40 PM)
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9716
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246711 - 02/16/11 06:08 PM
|
|
|
> Ah, I don't have this luxury in OpenGL It's easy enough to do, but computationally > expensive.
Just to clarify: D3DX* is essentially the DX version of GLUT, although it's got a lot more functions and no teapots (or it didn't last time I read the docs ;-). So the call they're using may well be O(n^2) also.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: R. Belmont]
#246731 - 02/16/11 10:37 PM
|
|
|
Will Supermodel be released in future? This framerate is incredible
The best looking version of MAME running Lost World was the 0.102u3 (i belive) and it looked very good
An advance in this game should be interesting
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Stranno]
#246733 - 02/16/11 11:07 PM
|
|
|
I'm just glad this is being worked on by two different people with 2 different emulators. I just hope when they are released you don't need a top of the line PC to run it at full speed [if full speed is even possible on the emulators at all].
Edited by pjay (02/16/11 11:08 PM)
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: pjay]
#246739 - 02/16/11 11:36 PM
|
|
|
Yes, of course Supermodel will be released. Still quite a bit of work to do before I can even release an alpha version but it's coming along. Keep in mind I've only been working at it since just after Christmas. So it's at quite an early stage.
Right now, there's no dynarec. It's using the old Supermodel's C interpreter for PowerPC emulation. There are a few optimizations here and there in the core code I can make and the shaders will have to be streamlined.
I tried a few games on my office PC, a 2.4 GHz Core2 w/ GeForce 7300 (whose drivers have not been updated for 3 years now) and got the following results:
Performance varies from scene to scene depending on geometric complexity. My development PC is an old laptop with integrated ATI video, so performance is not spectacular there, but a friend of mine tried it on his GeForce-based laptop and got over 60FPS consistently in Scud Race.
Maybe it's not the fastest Model 3 emulator in town yet, but it's not too shabby
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: drewcifer]
#246747 - 02/17/11 01:15 AM
|
|
|
I've sent my research to Bart. Working on these data was a long story... :-)
> Good thoughts, Mamu, however, I think Bart's right - he's looking for per-vertex > normals in this case, probably not per-vertex colors: > > I think a message to Mr. Naibo might help clear things up > > http://www.mameworld.info/ubbthreads/sho...part=1&vc=1 > > Andrew > > * Waves at Bart * > Congrats on getting yourself some more free time, and great progress! A fresh look is > always a nice thing.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246749 - 02/17/11 01:43 AM
|
|
|
Wow, you get almost 60fps on a Geforce 7300 for Model 3 games so far, nice! Now I realize that there where 4 or more revisions to the Model 3 Hardware ranging from Power PC 66Mhz to 166Mhz so I don't expect that smoothness in all games, but still that is fantastic, keep up the good work.
> Yes, of course Supermodel will be released. Still quite a bit of work to do before I > can even release an alpha version but it's coming along. Keep in mind I've only been > working at it since just after Christmas. So it's at quite an early stage. > > Right now, there's no dynarec. It's using the old Supermodel's C interpreter for > PowerPC emulation. There are a few optimizations here and there in the core code I > can make and the shaders will have to be streamlined. > > I tried a few games on my office PC, a 2.4 GHz Core2 w/ GeForce 7300 (whose drivers > have not been updated for 3 years now) and got the following results: > > > Performance varies from scene to scene depending on geometric complexity. My > development PC is an old laptop with integrated ATI video, so performance is not > spectacular there, but a friend of mine tried it on his GeForce-based laptop and got > over 60FPS consistently in Scud Race. > > Maybe it's not the fastest Model 3 emulator in town yet, but it's not too shabby
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9716
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246782 - 02/17/11 07:38 AM
|
|
|
My osd/sdl peeps will be pleased to know Bart's new Supermodel is SDL-based and works just as great as is pictured here on Linux (and probably Intel OS X if I can find GLEW for it).
And for the record, it's damned hard to not crash in Scud Race at 200+ FPS. Thing needs a frame limiter ;-)
|
|
|
Stiletto |
They're always after me Lucky ROMS!
|
|
|
Reged: 03/07/04
|
Posts: 6472
|
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Naibo]
#246788 - 02/17/11 08:48 AM
|
|
|
> I've sent my research to Bart. > Working on these data was a long story... > :-)
Thank you!
Don't forget to send your updated code to MAMEDEV when it's ready.
- Stiletto
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: R. Belmont]
#246797 - 02/17/11 03:12 PM
|
|
|
> My osd/sdl peeps will be pleased to know Bart's new Supermodel is SDL-based and works > just as great as is pictured here on Linux (and probably Intel OS X if I can find > GLEW for it).
Oh, excellent! Wish Demul could do that, too.
|
|
|
Nomax |
Ryo Hazuki
|
|
|
Reged: 11/19/03
|
Posts: 497
|
Loc: Belgium
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246804 - 02/17/11 05:00 PM
|
|
|
> So far, Le Mans 24 holds the dubious distinction of being the ugliest Model 3 game. > It appears to have been released after Scud Race and features many of the same > vehicles, making its sub-par quality all the more noticeable. It would have made an > average-looking Model 2 game.
I never had the "chance" to play it in arcades. Are the ugly graphics because of emulation or is the real game that ugly?
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Nomax]
#246831 - 02/17/11 09:28 PM
|
|
|
It's partly the emulator's fault but mostly it's just a rather sub-par game, especially for Sega! I was exaggerating a bit when I compared it to Model 2 but it can't hold a candle to Scud Race or Sega Rally 2. The texture mapping on the cars, for instance, is very crude, whereas Scud Race features environment mapping to make the car surfaces look reflective. The courses are boring and the background objects not particularly interesting or detailed.
The game has some redeeming features, however, namely the car handling, which although touchy, seems to try to more accurately model vehicle dynamics. It tries harder to be a racing simulator than Scud Race, which is more about speed, drifting, and fun.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246849 - 02/18/11 12:59 AM
|
|
|
Are the textures still missing on the door in lost world are due to a bad dump...?
I can't wait to test this
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246915 - 02/18/11 03:07 PM
|
|
|
> No, Lost World and Vipers are the same (well, Vipers _is_ referencing some things > that aren't models -- have you seen the same thing in Sega Rally?) > > Lost World boots without any patches over here and Fighting Vipers only needs a patch > to some weird decrementer loop (seems a lot of games have this issue during boot up). > Do the games start at all for you? Given that both games use different DMA hardware, > the problems could be unrelated. I'll need more info to help you troubleshoot.
Fighting Vipers 2 will probably need a patch because the weird decrementer is supposed to be tied to the dongle check.
The game will work for a few seconds and then be locked out because lack of correct dongle.
Other dongle games do this IIRC.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: R. Belmont]
#246917 - 02/18/11 03:25 PM
|
|
|
> And for the record, it's damned hard to not crash in Scud Race at 200+ FPS. Thing > needs a frame limiter ;-)
Scud Race Turbo!
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: AndyGeezer]
#246940 - 02/18/11 08:35 PM
|
|
|
I thought the dongle was emulated. How was the security data obtained for these games in MAME? Was it just based on examining what the game compares the data it reads from the dongle to?
I'm now looking into Daytona 2. My understanding is that the presently available ROMs are not properly dumped (32Mbit CROMs rather than 64Mbit), and yet Daytona 2 ran in earlier versions of MAME, I believe.
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9716
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246943 - 02/18/11 09:41 PM
|
|
|
> I thought the dongle was emulated. How was the security data obtained for these games > in MAME? Was it just based on examining what the game compares the data it reads from > the dongle to?
Some of the games make better use of the dongle than others. For ones which simply read the contents and strcmp() it with "WAKE UP!! ARF!" (actual Sega dongle data from System 32 Arabian Fight) it was easy for Ville to reverse-engineer.
> I'm now looking into Daytona 2. My understanding is that the presently available ROMs > are not properly dumped (32Mbit CROMs rather than 64Mbit), and yet Daytona 2 ran in > earlier versions of MAME, I believe.
The "presently available ROMs" have been the corrected redumps for the last 2 or 3 releases of MAME. Daytona 2 now passes the ROM test in the service menu. (Note: Daytona 2 uses the MMU capability of the 603; most Model 3 games do not).
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: R. Belmont]
#246944 - 02/18/11 09:45 PM
|
|
|
Thanks for the info. I was dreading have to deal with the MMU but now there's no turning back :P Here's hoping they aren't using paging and are just sticking to the segment registers. That would make life easier
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#246966 - 02/19/11 04:12 AM
|
|
|
I've noticed the missing texture in Lost World is it a bad dump?
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Jitterdoomer]
#246967 - 02/19/11 04:44 AM
|
|
|
No, it's probably a 4- or 8-bit texture that isn't emulated properly (or some texture modulation setting related to those formats that I haven't thoroughly investigated).
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#247115 - 02/21/11 04:08 PM
|
|
|
Daytona USA 2 Power Edition used to work in MAME. Well until a certain corner on the second track when it crashed!
The main difference seems to be the first oval track in original Daytona USA 2 is set in a giant glass dome with waterfalls, as opposed to a more realistic version in Power Edition. I saw the original Daytona USA 2 the other day and was thinking it's not been emulated yet!
Edited by VirtuaIceMan (02/21/11 04:08 PM)
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#247124 - 02/21/11 07:16 PM
|
|
|
Great work, man......
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Bart T.]
#247187 - 02/23/11 02:40 AM
|
|
|
Man, I'm excited about this emulator. Once this is released, I'll be done dealing with some crappy, slow emulation, and crashing messages.
Will this emulator work at full speed with an Intel "Celeron" dual-core processor? If so, I'll be VERY surprised! Oh, and will this have perfect sound and have support for the Sega Digital Sound Board on some of these games?
Edited by MegaManX500 (02/23/11 02:41 AM)
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: JustSaiyanPaul]
#247223 - 02/23/11 07:58 AM
|
|
|
> Man, I'm excited about this emulator. Once this is released, I'll be done dealing > with some crappy, slow emulation, and crashing messages. > > Will this emulator work at full speed with an Intel "Celeron" dual-core processor? If > so, I'll be VERY surprised! Oh, and will this have perfect sound and have support for > the Sega Digital Sound Board on some of these games?
the celeron is a pretty old cpu, he with a core 2 duo of 2,4 ghz is having 45/60 fps in lost world, so with that celeron you will probably get from 25/40 fps and if your gpu is weaker than that geforce 7300 the fps will drop even more
Edited by karadaniano (02/23/11 11:59 AM)
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: karadaniano]
#247230 - 02/23/11 02:00 PM
|
|
|
I see, because on the Model 2 Emulator, all the games run with lag without frameskip. I might need to get a fast PC or a laptop, though, which kinda sucks.
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: JustSaiyanPaul]
#247246 - 02/23/11 08:44 PM
|
|
|
> I see, because on the Model 2 Emulator, all the games run with lag without frameskip. > I might need to get a fast PC or a laptop, though, which kinda sucks.
Just wait for dexterMAMETM to be released. 60FPS for everything.
Kevin Eshbach
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: keshbach1]
#247277 - 02/23/11 11:52 PM
|
|
|
"dexterMAME"? Dude, I'm psyched now! Will Model 3 be perfectly emulated on that?
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: JustSaiyanPaul]
#247280 - 02/24/11 12:28 AM
|
|
|
It goes without saying that Model 3 will be better than perfect.
|
|
|
CiroConsentino |
Frontend freak!
|
|
|
Reged: 09/21/03
|
Posts: 6211
|
Loc: Alien from Terra Prime... and Brazil
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: bdam]
#247290 - 02/24/11 01:35 AM
|
|
|
great news about emulating this system.
oh wait. is supermodel 3 being coded for MAME or for a stand-alone emulator like SEGA Model 2 Emu ?
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CiroConsentino]
#247291 - 02/24/11 02:30 AM
|
|
|
It's a stand-alone emulator. I'm not the programmer, but I know this.
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9716
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: CiroConsentino]
#247300 - 02/24/11 04:17 AM
|
|
|
> great news about emulating this system. > > oh wait. is supermodel 3 being coded for MAME or for a stand-alone emulator like SEGA > Model 2 Emu ?
It's standalone. And to catch some questions further up: there's currently no sound. That's one reason it's fast...
|
|
|
|
Wrong ...
[Re: keshbach1]
#247325 - 02/24/11 10:00 AM
|
|
|
it will be called dexterSupermodel
|
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: R. Belmont]
#247332 - 02/24/11 01:12 PM
|
|
|
> It's standalone. And to catch some questions further up: there's currently no sound. > That's one reason it's fast...
Well assuming the author wouldnt mind, wont there be anything that is learned from Supermodel that can transferred to MAME's emulation of Model 3..?
Obvioulsy im not talking about speed... ...Id be happy if Model 3 games ran @10fps if they were actually drawing the polygons and textures correctly...
|
|
|
R. Belmont |
Cuckoo for IGAvania
|
|
|
Reged: 09/21/03
|
Posts: 9716
|
Loc: ECV-197 The Orville
|
|
Send PM
|
|
|
Re: Supermodel WIP Update: LeMans 24, Fighting Vipers 2, and Star Wars Trilogy
[Re: Ziggy100]
#247363 - 02/24/11 07:58 PM
|
|
|
> Well assuming the author wouldnt mind, wont there be anything that is learned from > Supermodel that can transferred to MAME's emulation of Model 3..?
Of course. Supermodel will remain open source so there are no secrets.
And I should mention that of course Supermodel will eventually have sound. I mean, that shady guy who did M1 has the source.
|
|
|