Anyone think Model 3 games look better than Naomi?
Kevin
checking them out lately, I've come to the conclusion that Model 3 games
still look better than the supposedly more powerful Naomi-based games.
Daytona USA 2 and the old Super GT are especially good examples of this.
The textures look more realistic and there's this smoothed out 'sheen' to
everything that makes the games look so much better than the Naomi games.
Really, I'd say Daytona2 and SuperGT look as good as any racing game ever
made and stand up to the quality of the PS2 racer shots. In comparison,
Naomi games tend to have an antisceptic, "clean" look reminiscent of PC
games. Even Yu Suzuki's much lauded Ferrari game doesn't look as good as
the aforementioned racers.
Is this just a matter of art direction? ie, choice of colours, design of
the models, etc? And what's with the extra smoothness in the Model 3 games?
Is it some kind of anti-aliasing technique or what? Because it makes games
look sooo nice.
For examples of the smoothness difference, check out VF3 DC vs VF3 arcade.
Even ignoring the polygon difference, you can tell there's something
different about the visual quality of the VF3 arcade. It looks glossy while
the DC version is gritty. I'm certain this is not just due to resolution or
higher quality monitor. There's more to it but it's hard to put into words.
Brian Osserman
>Is this just a matter of art direction? ie, choice of colours, design of
>the models, etc? And what's with the extra smoothness in the Model 3 games?
>Is it some kind of anti-aliasing technique or what? Because it makes games
>look sooo nice.
>
>For examples of the smoothness difference, check out VF3 DC vs VF3 arcade.
>Even ignoring the polygon difference, you can tell there's something
>different about the visual quality of the VF3 arcade. It looks glossy while
>the DC version is gritty. I'm certain this is not just due to resolution or
>higher quality monitor. There's more to it but it's hard to put into words.
Yes, I have noticed the same thing you describe, and I'm honestly not sure
what does it. I believe it to be something in the rendering technique itself,
but the Model3 texturing simply looks much more natural to me for some reason.
It's not inconceivable that it's the lighting doing it, since there's usually
much more variation in lighting algorithms than texturing algorithms. It's
also possible that it's some sort of antialiasing, it's been a long time since
I looked closely at a Model3 game, so my memory is a little vague, and I don't
know a whole lot about the architecture.
I wish I had a clearer picture of what's making the difference, because all
the Model3 games simply jump out when I look at them as looking 'nicer' in terms
of sheer image quality. Does anyone have a link to a detailed Model3 spec?
Brian
www.sega-dreams.com
Jeff Williams
"Kevin" <kev...@home.com> wrote:
> I'm lucky enough to have several well-stocked arcades in my town and
after
> checking them out lately, I've come to the conclusion that Model 3
games
> still look better than the supposedly more powerful Naomi-based games.
>
> Daytona USA 2 and the old Super GT are especially good examples of
this.
> The textures look more realistic and there's this smoothed
out 'sheen' to
> everything that makes the games look so much better than the Naomi
games.
> Really, I'd say Daytona2 and SuperGT look as good as any racing game
ever
> made and stand up to the quality of the PS2 racer shots. In
comparison,
> Naomi games tend to have an antisceptic, "clean" look reminiscent of
PC
> games. Even Yu Suzuki's much lauded Ferrari game doesn't look as
good as
> the aforementioned racers.
Well, first, I'm not sure Naomi is really more powerful than Model 3.
The main advantage of Naomi is that it's muuuuuuuuuuuuch cheaper than
Model 3 while giving comparable power. So it's a lot easier for
arcades to stock Naomi games than Model 3, which means more money in
the end for both the arcades and for Sega (and Sega runs their own
arcades too, so it means more quarters being pumped into their arcades
with a much lower production cost). I don't think Naomi's *less*
powerful, I just think they're different, and the games may look a
little different. But Naomi costs a lot less.
>
> Is this just a matter of art direction? ie, choice of colours,
design of
> the models, etc? And what's with the extra smoothness in the Model 3
games?
> Is it some kind of anti-aliasing technique or what? Because it makes
games
> look sooo nice.
I think it's a combination of things, including your own perception. I
think Model 3 games have a *harshness* to them that I don't really
mind, but it's basically the opposite of what you perceive. It could
also be the monitors - Daytona 2 uses slightly older, possibly lower-
res monitors than Ferrari F355, which may provide some artificial anti-
aliasing (in the form of lower sharpness). But, different 3D hardware
does look different. Compare a TNT card running a Direct3D game with a
Voodoo 3 card running the same game in Direct3D, and you *will* see a
difference. It doesn't mean one's better or worse, just different.
>
> For examples of the smoothness difference, check out VF3 DC vs VF3
arcade.
> Even ignoring the polygon difference, you can tell there's something
> different about the visual quality of the VF3 arcade. It looks
glossy while
> the DC version is gritty. I'm certain this is not just due to
resolution or
> higher quality monitor. There's more to it but it's hard to put into
words.
They probably do look a little different, although I don't think VF3 is
really a valid comparison because it was obviously a rushed port. But
even if Sega had had the time to really do all they could to make VF3
look *exactly* like the arcade in the DC version, it would probably
look a bit different. How you perceive that difference, though, is
probably up to each individual.
I kind of feel this way about Soul Calibur myself, actually, even
though I know that was originally on a much lower class of hardware
than the DC when it was in arcades. One of my local game stores still
has a machine and I still play it sometimes, and despite the lower
resolution and polygon counts it does have that arcade harshness that I
think Model 3 games also do, and that's what I'm used to from arcade
games. In SC's case, I think it is a combination of the monitor that's
in the arcade machine, and the contrast being cranked all the way up
(which would make sense in an arcade). I think if I wasn't used to
that from the arcade, and you showed me the arcade and DC versions side
by side, I'd think the arcade version - ignoring the higher resolution -
looks like utter crap. But because I'm used to the "arcade look" of
how the display is set up in the arcade machine, there's something
about it that I prefer over the DC version. Obviously I'll take the DC
version overall with the revamped graphics, but that one aspect of the
way it looks I'm just used to from the arcade.
--
// Jeff Williams
// ge...@nervhq.org
Sent via Deja.com http://www.deja.com/
Before you buy.
Raymond McKeithen II
Brian Osserman <osse...@math.mit.edu> wrote in message
news:86prha$o...@schubert.mit.edu...
> Kevin <kev...@home.com> wrote:
> >Is this just a matter of art direction? ie, choice of colours, design of
> >the models, etc? And what's with the extra smoothness in the Model 3
games?
> >Is it some kind of anti-aliasing technique or what? Because it makes
games
> >look sooo nice.
> >
arcade.
> >Even ignoring the polygon difference, you can tell there's something
> >different about the visual quality of the VF3 arcade. It looks glossy
while
> >the DC version is gritty. I'm certain this is not just due to resolution
or
> >higher quality monitor. There's more to it but it's hard to put into
words.
>
sure
> what does it. I believe it to be something in the rendering technique
itself,
> but the Model3 texturing simply looks much more natural to me for some
reason.
> It's not inconceivable that it's the lighting doing it, since there's
usually
> much more variation in lighting algorithms than texturing algorithms. It's
> also possible that it's some sort of antialiasing, it's been a long time
since
> I looked closely at a Model3 game, so my memory is a little vague, and I
don't
> know a whole lot about the architecture.
> I wish I had a clearer picture of what's making the difference, because
all
> the Model3 games simply jump out when I look at them as looking 'nicer' in
terms
> of sheer image quality. Does anyone have a link to a detailed Model3 spec?
>
I know, however, that Model 3 games do use some sort of antialiasing, which,
so far I don't think Naomi and DC are using, even though it's supposed to be
available.
Model 3's resolution is lower than Naomi/DC. At 24.83kHz scanrate, the
theoretical limit would be 640x400 (vs. 640x480), however according to what
I've read Model 3 is still lower than that, some oddball number like 590x389
(not accurate).
I don't think outside of some PR bs, Sega has ever truly claimed Naomi was
"more powerful" than Model 3. Naomi was meant more as a low-cost replacement
since Model 3 games cost so much and thus get limited distribution. For
certain Sega was specific even in the PR to NEVER say it was "more powerful"
than Model 3 Step 2 which Daytona 2 uses. We know now that Sega still plans
to make some sort of "Model 4" though it may not be called that, and I
imagine the now-announced VF4 will use it.
In Super GT's case, I think art direction is part of what makes the game
look so nice. There was a letter in an old NextGen issue that talked about
this....it mentioned how Konami's Racing Jam (the "Cobra" game) was more
"realistic" looking in terms of looking like "real life" but that it lacked
any style and was thus far more boring-looking compared to Super GT. [One of
the points I was trying to make to "Mr. Realism" in that other thread, if
you want that kind of realism, look out a window...]
--
Raymond
remove "suchiepai" for email
Pio
full effect on. In fact, that machine is a beast because it has separate
chips to handle the effects : lighting, etc. For the powerVr, there is 1
chip doing the rendering so maybe the lighting is not as polished.
For a spec sheet of both check : http://www.system16.com
Kevin <kev...@home.com> wrote in message
news:gpYj4.19059$632.9...@news1.rdc2.on.home.com...
> I'm lucky enough to have several well-stocked arcades in my town and after
> checking them out lately, I've come to the conclusion that Model 3 games
> still look better than the supposedly more powerful Naomi-based games.
>
> Daytona USA 2 and the old Super GT are especially good examples of this.
> The textures look more realistic and there's this smoothed out 'sheen' to
> everything that makes the games look so much better than the Naomi games.
> Really, I'd say Daytona2 and SuperGT look as good as any racing game ever
> made and stand up to the quality of the PS2 racer shots. In comparison,
> Naomi games tend to have an antisceptic, "clean" look reminiscent of PC
> games. Even Yu Suzuki's much lauded Ferrari game doesn't look as good as
> the aforementioned racers.
>
Andrew Ryan Chang
> Yes, I have noticed the same thing you describe, and I'm honestly not sure
>what does it. I believe it to be something in the rendering technique itself,
>but the Model3 texturing simply looks much more natural to me for some reason.
Just larger textures?
>It's not inconceivable that it's the lighting doing it, since there's usually
>much more variation in lighting algorithms than texturing algorithms. It's
>also possible that it's some sort of antialiasing, it's been a long time since
>I looked closely at a Model3 game, so my memory is a little vague, and I don't
>know a whole lot about the architecture.
AIUI, the Model 3 games tend to do trilinear filtering and have
some sort of edge antialiasing, which would certainly help picture
quality. Check out the page mentioned below (around the middle).
<http://www.canadawired.com/~gvink/Sega/Tech/tech_gpu.html>
While there are more variations in lighting algorithms than
texturing algorithms, I wasn't aware of any realtime 3D hardware that does
it differently from the tried-and-true "SGI" way... (gouraud shaded phong
illuminated.) A cursory search on google makes me think nothing's changed
yet... Really, cheesy though it might be, the image quality can be upped
a lot with additional hacks like fullscreen antialiasing, more and smaller
polys, and better quality textures and good texture filtering.
--
"_Dornbeast:_ Smart adventurers run the other way when they hear 'Hurumph',
the battle cry of the deadly dornbeast." -Encyclopedia Frobozzica.
Brian Osserman
> Just larger textures?
It'd have to be more than this, or it wouldn't also look good at long range.
> AIUI, the Model 3 games tend to do trilinear filtering and have
>some sort of edge antialiasing, which would certainly help picture
>quality. Check out the page mentioned below (around the middle).
><http://www.canadawired.com/~gvink/Sega/Tech/tech_gpu.html>
Good trilinear filtering together with antialiasing could have a really
huge impact on quality, yeah. The problem is, I haven't run across a specs
page that actually discusses what's used in the average game (both of
these effects are also listed for NAOMI, but essentially never used).
But I suppose these are the most logical explanations.
> While there are more variations in lighting algorithms than
>texturing algorithms, I wasn't aware of any realtime 3D hardware that does
>it differently from the tried-and-true "SGI" way... (gouraud shaded phong
>illuminated.) A cursory search on google makes me think nothing's changed
>yet...
Phong illuminated? I think very few games do phong, even at vertices. When
they do, it tends to jump out, as you get a much shinier effect. I think for
games the most common is still generic diffuse lighting with gouraud shading.
But your point still stands, I don't suppose there's much chance that someone
has gone out and come up with a substantially better illumination model and
implemented it in real time and nobody's heard of it.
>Really, cheesy though it might be, the image quality can be upped
>a lot with additional hacks like fullscreen antialiasing, more and smaller
>polys, and better quality textures and good texture filtering.
Cheesy? Heh, these are the backbone of a good image. Full screen antialiasing
alone has a big impact and even dimishes the need for good (ie, trilinear) texture
filtering somewhat. I would like to say that we can assume polys aren't a factor
here, though, as the differences observed tend to feel a little less tangible
than that, but I suppose anything is possible.
Brian
www.sega-dreams.com
John A. Robertson
: Kevin <kev...@home.com> wrote:
:>Is this just a matter of art direction? ie, choice of colours, design of
:>the models, etc? And what's with the extra smoothness in the Model 3 games?
:>Is it some kind of anti-aliasing technique or what? Because it makes games
:>look sooo nice.
:>
:>For examples of the smoothness difference, check out VF3 DC vs VF3 arcade.
:>Even ignoring the polygon difference, you can tell there's something
:>different about the visual quality of the VF3 arcade. It looks glossy while
:>the DC version is gritty. I'm certain this is not just due to resolution or
:>higher quality monitor. There's more to it but it's hard to put into words.
: Yes, I have noticed the same thing you describe, and I'm honestly not sure
: what does it. I believe it to be something in the rendering technique itself,
: but the Model3 texturing simply looks much more natural to me for some reason.
I think the biggest factor is that the Model 3 keeps 100+ megabytes of
textures in ROM, whereas the Dreamcast is stuck with 8 Megabytes minus whatver
it takes for the framebuffer(s). I'm guessing that they can't effectively
stream textures off of the GD-ROM, whereas the Model 3 can obviously move them
out of ROM into VRAM more quickly.
Shortly after the Dreamcast launch, the head of one of the AM divisions was
asked what he thought of the relative power of the Model 3 and Dreamcast.
He replied that (among other things) the Dreamcast could handle more
polygons, but the Model 3 had superior texturing capabilities because of
the ROM storage format.
I beleive that the Model 3 uses trilinear filtering (I know it is supported.)
A PowerVR SG PC chip takes one cycle to render a bilinear filtered pixel but
two to render a trilinearly filtered pixel and 4 for anisotropic, so I think
that we're going to see bilinear exclusively.
Something that the Model 3 supposedly had was what they referred to as
'microtexturing.' I'd like to know what that referred to. Does anyone
know? The best I could guess is that it referrs to some sort of sub-pixel
precision in their texturing routine, but that is kind of vague considering
what today could be considered sub-pixel precision; pretty much anything but
point sampling.
One thing that I notice on the Dreamcast is the severely obvious mipmap levels.
Check out the grass in Sonic Adventure or the floors and walls in the long
rooms in Shadowman. The textures go from relatively clear to very blurry right
along a very pronounced line. I don't recall seeing this on Model 3 games.
I haven't been to the arcade for a while. Perhaps Model 3 has something like
per-pixel mipmapping. It isn't rare today (the GeForce supports it,) but it
would have been impressive for 1996.
On a side note, It's amazing that they got such unreal performance from the
Model 3, especially the Step 1. It had 50 Megapixel/sec fill rate and the
step two had 100 Megapixel/sec rate. That's pitiful, but there's no denying
the performance.
[Off topic comments follow...]
I can't wait to see the Dolphin in action. The textures on the PS2 seem
to suffer from the same blurry appearance as on the Dreamcast but ArtX inc.
licensed S3TC (also known as DXTc) from (of course) S3 . Anyone who
has seen the difference knows how amazing it is. Of course if they had
waited a few more weeks they could have licensed 3dfx's slightly superior
FXT1 royalty free.
I think that even if the Dolphin doesn't push more polys than the PS2, it
will still look superior thanks to S3TC combined with apparently a lot
more embedded DRAM, like 8-16 MB total according to some fellow at NEC.
(NEC is providing the eDRAM format (1-T eDRAM) for the Dolphin.)
I currently suspect that the Dolphin vs. PS2 will be similar to N64 vs. PSX.
Similar # of polys, but better image quality on the Nintendo machine.
: It's not inconceivable that it's the lighting doing it, since there's usually
: much more variation in lighting algorithms than texturing algorithms. It's
: also possible that it's some sort of antialiasing, it's been a long time since
: I looked closely at a Model3 game, so my memory is a little vague, and I don't
: know a whole lot about the architecture.
: I wish I had a clearer picture of what's making the difference, because all
: the Model3 games simply jump out when I look at them as looking 'nicer' in terms
: of sheer image quality. Does anyone have a link to a detailed Model3 spec?
: Brian
: www.sega-dreams.com
Warthawg
In article <86sfo8$8...@schubert.mit.edu>,
osse...@math.mit.edu (Brian Osserman) wrote:
> Andrew Ryan Chang <arc...@sfu.ca> wrote:
> > Just larger textures?
>
> It'd have to be more than this, or it wouldn't also look good at long range.
>
> > AIUI, the Model 3 games tend to do trilinear filtering and have
> >some sort of edge antialiasing, which would certainly help picture
> >quality. Check out the page mentioned below (around the middle).
> ><http://www.canadawired.com/~gvink/Sega/Tech/tech_gpu.html>
>
> Good trilinear filtering together with antialiasing could have a really
> huge impact on quality, yeah. The problem is, I haven't run across a specs
> page that actually discusses what's used in the average game (both of
> these effects are also listed for NAOMI, but essentially never used).
> But I suppose these are the most logical explanations.
>
> > While there are more variations in lighting algorithms than
> >texturing algorithms, I wasn't aware of any realtime 3D hardware that does
> >it differently from the tried-and-true "SGI" way... (gouraud shaded phong
> >illuminated.) A cursory search on google makes me think nothing's changed
> >yet...
>
> Phong illuminated? I think very few games do phong, even at vertices. When
> they do, it tends to jump out, as you get
Brian Osserman
>Something that the Model 3 supposedly had was what they referred to as
>'microtexturing.' I'd like to know what that referred to. Does anyone
>know? The best I could guess is that it referrs to some sort of sub-pixel
>precision in their texturing routine, but that is kind of vague considering
>what today could be considered sub-pixel precision; pretty much anything but
>point sampling.
I actually asked someone at Sega about this a while back, and got a rather
vague response. It is indeed a sort of sub-pixel precision thing; the best I
could make out was it may allow for something like texel offsets for
calculating filtering, to avoid the squarish look of bilinear. But he also
amde it sound like it was pretty irrelevant, since the Model3 textures are
so high res to start with (of course, he was pushing the DC's capabilities
at the time, too). In fact, if I remember correctly, I was asking him about
precisely in the context of trying to figure out why I thought Model3 games
look better.
>One thing that I notice on the Dreamcast is the severely obvious mipmap levels.
>Check out the grass in Sonic Adventure or the floors and walls in the long
>rooms in Shadowman. The textures go from relatively clear to very blurry right
>along a very pronounced line. I don't recall seeing this on Model 3 games.
>I haven't been to the arcade for a while. Perhaps Model 3 has something like
>per-pixel mipmapping. It isn't rare today (the GeForce supports it,) but it
>would have been impressive for 1996.
This could be explained by trilinear, assuming it actually uses it. What do
you mean by per pixel mipmapping, though? I'm not sure I'm familiar with that.
Brian
www.sega-dreams.com
John A. Robertson
: John A. Robertson <jaro...@gnu.uvm.edu> wrote:
<snip>
:>One thing that I notice on the Dreamcast is the severely obvious mipmap levels.
:>Check out the grass in Sonic Adventure or the floors and walls in the long
:>rooms in Shadowman. The textures go from relatively clear to very blurry right
:>along a very pronounced line. I don't recall seeing this on Model 3 games.
:>I haven't been to the arcade for a while. Perhaps Model 3 has something like
:>per-pixel mipmapping. It isn't rare today (the GeForce supports it,) but it
:>would have been impressive for 1996.
: This could be explained by trilinear, assuming it actually uses it. What do
: you mean by per pixel mipmapping, though? I'm not sure I'm familiar with that.
: Brian
: www.sega-dreams.com
You probably know most/all of this, but.....
3d cards can choose different ways of generating mip-map levels. The more
finely partitioned, the better of course. I think the the old days,
x-number of mipmap levels were generated and when a polygon was rendered
with mipmapping, the mipmap level was chosen to be the closest fit of the
lot. I assume that they used the center of the polygon to determine the closest
mipmap level, but for polygons that span a large chunk of the z-axis, the levels
could get pretty screwed up.
I am not realy sure of the method of calculating the best fit. The distance
from the camera or something more generic like z-axis distance. Obviously
two objects at the same z value in world space will not be the same distance
if they are at different distances (x,y) from the center of the screen
(or from the z-axis itself.) I assume that the z-buffer is used these days.
z-buffer and z-axis (world space) of course being different.
The correct way would be to do distance from the viewer (camera), but since
the discrepancy between the mipmap levels is always parallel to the edges
of the screen, I think some cards must just be calculating a z-value, world
coordinates.
Someone chime in if I'm wrong here. Something in my explanation doesn't
add up.
At any rate, some cards (I think that the TNT2 is one) generate per-polygon
mipmap levels and (of course) generate a new mipmap for each polygon drawn,
so there are no static mipmaps. [I think an example of a card that
generated static mipmaps was the nVidia Riva 128.]
The GeForce goes one better than the TNT2 and generates a mipmap for each
pixel. If you want to see an example of this, you can check it out in
Quake III. There is an option (a console command) to display a (err..)
'false color' image of a scene that shows the mipmap levels. I don't know
what the command is off the top of my head. A number of sites have posted
screen shots of this, but I couldn't readily find a link.
If you enable this option, it displays an image of the screen that has various
shades of red and blue and green denoting different mipmap levels. If you
enable this on a TNT2, you will see each polygon a differet color.
If this is enabled on a GeForce, there will be a color gradient across
each polygon demonstrating a new mipmap level for each pixel.
The reality is that the difference between per-polygon and per-pixel
mipmaps isn't a big deal (at least on the TNT2 and GeForce, which have
excellent image quality) unless you have huge polygons, but the older
mipmap methods really look terrible. I used to use a Riva 128 and the
image quality was wretched.
\subsection{aside} Nod to any LaTeX users here.
I suppose on a tv, the image quality probablems are less apparent,
but I can still see the mipmap levels in some games. BTW, anyone know
the best deal on a Vega (err Wega)? Is there anything else out there
that is as flat, but not so bloody expensive? UVM pays their grad
students a pitance.]
I have to wonder what the PS2 does for this. It has the eDRAM bandwidth
to do such stuff, but it would seem that the 'Graphics Synthesizer'
is incredibly simplistic. Hmmm.....
[Off topic...]
I've noticed that you're a math wonk from MIT. You don't know anything
about confluent hypergeomtric functions, do you? I need a
linear combination of
x exp(-x^2 /2) 1F1(a,b,x^2)
and
x exp(-x^2 /2) U(a,b,x^2)
[a, b fairly arbitrary, although b really = 3/2]
that satasfies the boundary conditions that the combination goes to
zero at x=-infinity and becomes singular at x=+infinity.
I need something like it for a Green's function that I need to
construct. The solutions are closely related to the simple harmonic
oscillator Schrodinger's eqn.
[I do need the combination, but yes I am joking in posting it here.]
Back to work....
Brian Osserman
>you guys sound really smart. Are any of you developers ?
Not I, I'm just a geek. As far as I know, the few developers who are
here don't tend to participate in this sort of discussion.
Brian
(putting off a real job for as long as humanly possible)
Andrew Ryan Chang
>Warthawg <wart...@my-deja.com> wrote:
>>you guys sound really smart. Are any of you developers ?
>
> Not I, I'm just a geek. As far as I know, the few developers who are
>here don't tend to participate in this sort of discussion.
I'm a geek too! But I'm doing a co-op ("internship" ?) at a game
development company... (not that my work involves consideration of such
graphic details -- for this stuff, I'm just an interested bystander. AKA
geek.)
--
"_Lord Dimwit Flathead the Excessive_: The empire simply did not have
enough money to build [the new continent][...] Lord Dimwit, never
satisfied, proposed adopting everyone in the Kingdom and telling them he'd
cut off their allowances." -Encyclopedia Frobozzica.
Brian Osserman
>At any rate, some cards (I think that the TNT2 is one) generate per-polygon
>mipmap levels and (of course) generate a new mipmap for each polygon drawn,
>so there are no static mipmaps. [I think an example of a card that
>generated static mipmaps was the nVidia Riva 128.]
>
>The GeForce goes one better than the TNT2 and generates a mipmap for each
>pixel. If you want to see an example of this, you can check it out in
>Quake III. There is an option (a console command) to display a (err..)
>'false color' image of a scene that shows the mipmap levels. I don't know
>what the command is off the top of my head. A number of sites have posted
>screen shots of this, but I couldn't readily find a link.
OK, I kind of suspected this was what you meant, but I didn't actually
know that cards were doing this these days (haven't been keeping very close
track of the PC world, I'm afraid). Either way, it sounds computational
enough to do even per-polygon dynamic mipmap levels that I'd bet the DC
doesn't do them (looking at games like Sonic seems to confirm this, too --
it looks like it's using precalculated levels which are chosen on a per
pixel basis). In any case, thanks for bringing me up to date. :-)
>\subsection{aside} Nod to any LaTeX users here.
Ironically, I have two ssh windows open right now; one for trn, one
editing a latex file.
>I've noticed that you're a math wonk from MIT. You don't know anything
>about confluent hypergeomtric functions, do you?
I'm afraid not. I'm on the pure end of things.
>I need a
>linear combination of
>
>x exp(-x^2 /2) 1F1(a,b,x^2)
>
>and
>
>x exp(-x^2 /2) U(a,b,x^2)
Not being familiar with the lFl and U functions, I'm afraid I can't
be of much help.
>[I do need the combination, but yes I am joking in posting it here.]
No doubt. But ask about algebraic number theory and I'll be happy
answer, off topic or not. :-) Incidentally, I suppose you already
know this, but posting to sci.math or sci.math.research is often
helpful for getting help with this sort of thing.
>Back to work....
Likewise...
Brian
www.sega-dreams.com
Game Otaku
>levels.
>Check out the grass in Sonic Adventure or the floors and walls in the long
>rooms in Shadowman. The textures go from relatively clear to very blurry
>right
>along a very pronounced line.
Yes, this is due to mip-mapping and a lack of tri-linear mapping. Not ALL
Dreamcast games suffer from this (as Sega Rally 2 for example, doesn't), but a
lot do. Tri-linear gives you subtle levels of smoothless instead of just two;
fuzzy and sharp.
-Game Otaku ^_^
Editor-in-Chief
Die Hard Gamer (www.diehardgamer.com)
We're back and still hiring writers so apply at our site or simply email me for
an application. Just remove 'municate'.
Michael R. Baraniecki
<20000201004607...@ng-fi1.aol.com>:
>>One thing that I notice on the Dreamcast is the severely obvious mipmap
>>levels.
>>Check out the grass in Sonic Adventure or the floors and walls in the
>>long rooms in Shadowman. The textures go from relatively clear to very
>>blurry right
>>along a very pronounced line.
>
>Yes, this is due to mip-mapping and a lack of tri-linear mapping. Not
>ALL Dreamcast games suffer from this (as Sega Rally 2 for example,
>doesn't), but a lot do. Tri-linear gives you subtle levels of
>smoothless instead of just two; fuzzy and sharp.
>-Game Otaku ^_^
>Editor-in-Chief
I was just wondering, do you know which other games for the DC don't
suffer from this, besides Sega Rally 2? Does Crazy Taxi? (haven't played
it in the arcades enough to notice) Shen Mue? (I would assume it does not
use tri-linear, just to keep the huge number of polys) I find it curious
that Sega Rally 2 uses it... seeing as how it's a Win CE game, but then
again, it was also a Model 3 game originally, so...
Anyway, thanks for any info, I like trying to learn more about these
techniques and reading posts such as these. :)
--
-Mike
Michael R. Baraniecki
uni...@netcom.com
http://www.geocities.com/TimesSquare/6470
Videogame MIDIs and Space Ghost sounds galore
Isamu
you started this thread! Not only am I learning alot of
information here, but you have done something I have always
wanted to do ever since I first saw a model 3 game (in my case,
Super GT). And that is, put into words how I feel about the way
Model 3 games look compared to NAOMI. When you described the
smoothness look to Super GT and Daytona 2, I couldn't help but
think "DAMM, he hit it right on the nail!" I whole heartedly
agree 100%! And after reading the last few posts in this thread,
now I know why. You're right! Naomi games look too much like PC
games in my opinion. They have an annoyingly clean, antisceptic
look that makes me think they're running off a voodoo 3 or
something. I first disliked this look when I first saw Cruisin
USA. I thought "well, maybe that's the last racing game that'll
look like that so it's not too bad". Then San Fran Rush came out
and I was like, "Ugh! Not again with this clean, papery look"!
Now Naomi games are suffering from this same look. And
Unfortunately, Ferrari 355 is the latest casualty of this. To
this day, I still say Super GT and Daytona 2 will forever be the
most GORGEOUS looking racing games of all time!!! RRV might come
close though! Proceed with thread!
* Sent from RemarQ http://www.remarq.com The Internet's Discussion Network *
The fastest and easiest way to search and participate in Usenet - Free!
Isamu
games including VF4 and *SUPPOSEDLY* a new sequel 2 OutRun and
both R said 2 B running off the next hardware in the "Model"
series. Not 2 mention the numerous reports that Sega has
mentioned how Naomi/Hikaru's reaching it's limit (aka the
upcoming Star Wars: Racer which I hear has Naomi bleeding at the
seams). So (fortunately) I think they're gunna proceed with the
Model series. Don't quote me on it tho.
0101011101110001
the end, Naomi might be a better choice.