After all these years Le mans on DC is still graphically...
- Thread starter SegaR&D
- Start date
jarrod said:
how dare you question the out-of-boredom (sega) bashers? regardless whether a thread has a direct or a remote connection to sega, popping up just to say the magical 'ohh p-lease, not sega again' gives you this stature.
ps:
on a sidenote, the quality of recent threads on the console board has gone south. in every thread that has the potential of being at least of some use to even a handful of people there always pop up several guys full of so much "constructivism" that the thread becomes of no use to anybody. last minute newsflash: if the thread topic reads something loathfull to you, like 'ps2 bumpmapping' or 'dc rocks' you can actually _not_ click on the thread's link. thanks for your attention.
darkblu said:
I second that. It's a shame to see topics closed because of a few individuals that just can't do better than to derail threads/topics they don't like and have it closed in the process. It's a shame because there are also other members posting in there actually going through an effort making quality posts worth reading - posts that once got me to sign up here in the first place! Don't the mods or whoever's responsible for closing threads see that by simply closing threads, more and more posters aren't bothering to post quality replies or not at all anymore? It's a continuation of the degration of this forum - and it's a bloody shame indeed. :?
Sorry I meant obscure defunct dreamcast stuff
Darkblu,
Take it easy. I'm not a sega basher as I happen to buy lot's of sega games. I'm a old fan that has more recently starte liking sega again. So I'm not bashing sega in my post, I was more bothered with the sega spammig we get from sega R&D because it's not unlike the xbox spammig we got from chap (althogh admittedly sega R&D is not wrecking other threads.). After the first 10 sega getting back into hardware threads and claims of being an ex sega of japan employee, I guess I haven't treated his topics with a level of respect. Well my bad, long live sega
Re: After all these years Le mans on DC is still graphically
What is it in that picture that makes you believe there is no real time filtering involved, and that it is using RIP maps rather than anisotropic (trophic relates to nutrition, I think) filtering?
I'm happy to take the developer's word for it myself.
Squeak said:
So that's the famed anisotrophic filtering! Looks a lot like they are just using RIP maps instead of MIP maps to me. No realtime filtering nvolved.
What is it in that picture that makes you believe there is no real time filtering involved, and that it is using RIP maps rather than anisotropic (trophic relates to nutrition, I think) filtering?
I'm happy to take the developer's word for it myself.
Re: After all these years Le mans on DC is still graphically
The road texture is obviously filtered more in one direction than the other. This could be for a speed blur effect but the filtering does the exact same thing real-time anisotropic would do to a slanted texture, only here it’s also “done†when it is not necessary.
The road texture is obviously filtered more in one direction than the other. This could be for a speed blur effect but the filtering does the exact same thing real-time anisotropic would do to a slanted texture, only here it’s also “done†when it is not necessary.
I think you're seeing the "motion blurred" road textures, and assuming this is what people are referring to when they talk about the game's anisotropic filtering. Creating static textures with a certain look (be it shading, motion blurring, focus or whatever) is entirely separate and unrelated to the filtering performed on those textures at render time.
Suffice to say, anisotropic filtering doesn't make things look strectched. When it works you can't really see it, only the absence of (well, reduction in) aliasing.
Having road textures which imply a sense of speed in the direction of the road is a very old trick (and one that you could use along side any number of filtering techniques).
Suffice to say, anisotropic filtering doesn't make things look strectched. When it works you can't really see it, only the absence of (well, reduction in) aliasing.
Having road textures which imply a sense of speed in the direction of the road is a very old trick (and one that you could use along side any number of filtering techniques).
That's exactly my point. They didn't do real anisotropic (I believe), and they didn't have to, when they used those textures (at least for the road which I believe is all they claimed to have filtered).function said:
If you apply a RIP map to a square billboard polygon it is going to look stretched.
And also a nice cheap way of doing two things at once.Having road textures which imply a sense of speed in the direction of the road is a very old trick (and one that you could use along side any number of filtering techniques).
Squeak said:
Ah, but using those textures doesn't mean you don't need to (as in 'benefit from') using anisotropic filtering ...
Not quite sure what you're saying I'm afraid. Anisotrpic filtering doesn't invlove the stretching of texures, only a different sampling pattern from normal bilinear.
And also a nice cheap way of doing to things at once.Having road textures which imply a sense of speed in the direction of the road is a very old trick (and one that you could use along side any number of filtering techniques).
The road textures being "pre-blurred" to imply a sense of speed (fake temporal blurring I supose) doesn't replace aniso though. Being able to see this doesn't have any bearing on whether aniso could be, or was, used...
OH how I wish all PS2 racers used anisotropic filtering. Some of them have so much texture aliasing that it is funny when you think about it. Some of them look awesome though even though they could benefit from some nice anisotropic filtering which I think either the PS2 can't do in hardware or is very expensive to do.
It would be like doing MIP mapping twice, first pre-calc and then in real-time (maybe there would be a very small improvement, but not enough to justify it's use)function said:
Anisotropic filtering involves downsampling the texture more in one direction than the other. That means if the texture is not applied to a polygon where it is accordingly compressed in the direction in which it was oversampled, it is going to look stretched.
Not up close of course (where it takes on the role you describe) but on the higher mip levels is begins to work as anisotropic.And also a nice cheap way of doing to things at once.Having road textures which imply a sense of speed in the direction of the road is a very old trick (and one that you could use along side any number of filtering techniques).
The road textures being "pre-blurred" to imply a sense of speed (fake temporal blurring I supose) doesn't replace aniso though.
Fox5 said:
You can use it to do both. Anisotropic filtering samples more of the texture elements that are supposed to be 'under' each pixel; this changes not only with distance from the camera (something trilinear filtering takes into account) but the slope of the textured surface with respect to the camera. This way, anisotropic filtering can reduce shimmer and moire, and allow you to push the changes between mip map levels further out into the distance (reducing blur).
Squeak: What you can see in that road texture is not what anyone would try and pass off as anisotrpic filtering. It'll still need mip mapping to hide texture aliasing, and it'll still suffer from moire and shimmer at acute angles (and therefore benefit from aniso filtering).
Hopefully, someone far better qualified can take a moment to explain this in a straight forward manner!
Squeak,
your saying is not completely unreasonable yet:
a) longitude-blurred road textures can give you only this much headroom. from a certain distance on you still get undersamling (ie. shimmering) with bilinear (using trilinear at these degreees of anisotropy would be murderous) unless your texture is complete DC (meaning here 'direct current', ie. 0 freq) along the longitude direction. 24HLM on the DC does not exhibit any visible road texture shimmering, with road visibility reaching very far ahead.
b) the tracks in 24HLM are not airstip runways - they are well winding - ripmapping would fail miserably at such scenarion.
c) the roadside billboards in the game are still sampled better than what biliner alone can provide, and they don't have any prebaked longitudal blurring.
d) rip-mapping is a bit wastefull technique for an 8MB console platform. 24HLM on the DC is a textures-lush game - i'm pretty sure melbourne house have made better use of those 8MBs than ripmapping.
e) ..and last but not least, it's kind of hard to do ripmapping w/o the hardware being designed for that (or being very programmable) - the PVR2DC is a hardwired functionality unit.
bottomline being, i'm pretty positive the 24HLM on the DC utilises anisotropic filtering. it may be a low degree (say, 4th) and part of the effect may indeed come from the longitudal blur, but the latter alone is definitely not sufficient for the virtually flowless road sampling exhibited by this title on this platform.
ps: i have about 200h in this game, so you can take my observations on it with less salt.
your saying is not completely unreasonable yet:
a) longitude-blurred road textures can give you only this much headroom. from a certain distance on you still get undersamling (ie. shimmering) with bilinear (using trilinear at these degreees of anisotropy would be murderous) unless your texture is complete DC (meaning here 'direct current', ie. 0 freq) along the longitude direction. 24HLM on the DC does not exhibit any visible road texture shimmering, with road visibility reaching very far ahead.
b) the tracks in 24HLM are not airstip runways - they are well winding - ripmapping would fail miserably at such scenarion.
c) the roadside billboards in the game are still sampled better than what biliner alone can provide, and they don't have any prebaked longitudal blurring.
d) rip-mapping is a bit wastefull technique for an 8MB console platform. 24HLM on the DC is a textures-lush game - i'm pretty sure melbourne house have made better use of those 8MBs than ripmapping.
e) ..and last but not least, it's kind of hard to do ripmapping w/o the hardware being designed for that (or being very programmable) - the PVR2DC is a hardwired functionality unit.
bottomline being, i'm pretty positive the 24HLM on the DC utilises anisotropic filtering. it may be a low degree (say, 4th) and part of the effect may indeed come from the longitudal blur, but the latter alone is definitely not sufficient for the virtually flowless road sampling exhibited by this title on this platform.
ps: i have about 200h in this game, so you can take my observations on it with less salt.
Similar threads
