Resolutions and aliasing

[Moloch]

Clear Type is adjustable; it's clearly a matter of preference and I respect that but if you fiddle around with the settings a bit I'm sure you could find a happy medium that won't make your text to appear blury.

I'm using it on this CRT even now with a 1600 desktop. I usually use a high quality text printout to compare it to.

I messed with it with the tool and never got any "happy medium."

 

[Sxotty]

I will have to try clear type again on my CRT, it is a trinitron 21". I used it previously but did not get the tuner tool (it is apower toy right) Anyway it sucked without tuning it properly. On my laptop I got the tuner and it looks much nicer afterwards. In otherwords perhaps the lack of tuning killed it on my crt I will have to see...But I stll say i don't really need AA @ 1600 or greater, but maybe I just need glasses, or perhaps my annoyance level from jaggies is not as great. I can see them, but they are not distracting. I would always get higher res if the choice is up to me. (I thnk 21" is perfect at 1600x1200 though, no real need for higher, though I would not deny that it is fun to get an even higher res)

 

[Ailuros]

I enlarged the font sizes a bit to get text even readable for 1600*1200 on the desktop; no idea if that's a factor here too or not.

I didn't get a satisfying result with it at first either; tried a couple of combinations first until I got just a small of pinch of font smoothing.

 

[Bolloxoid]

I'd ask another question before anything else: is there a theoretical limit in dpi values where the human eye can still see a difference?

Of course there is, but that won't solve the problem of aliasing. We may note, for example, that human beings are able to resolve periodic gratings of up to 50-60 cycles per degree of visual angle. But if we render images containing higher frequencies on a display capable of that spatial frequency (theoretically the limit of human vision), we still get aliases in the visible frequency range that people can perceive.

 

[Ailuros]

But if we render images containing higher frequencies on a display capable of that spatial frequency (theoretically the limit of human vision), we still get aliases in the visible frequency range that people can perceive.

I'd think so too; the next best question though is if we'd get more overall side-effects than we'd get today with AA/AF(tri) or less on average.

Reverend's questions are hardly ever a coincidence; example:

FSAA—full scene antialiasing—implies that the full scene is completely correctly sampled. In other words, even if little tiny slivers of triangles 1/1000th of a pixel wide cover the screen, each pixel's color will be correct as if you exactly correctly divided each pixel into the area covered by each little triangle and integrated the colors. Of course, this isn't practical, and it isn't how any hardware or software works. We approximate that result using a technique known as "point sampling." We evaluate, or sample, each pixel in one or more points, and assume that will be a good estimate.

As an aside, relative to your question, this is not an "inefficient" fake method—this is an extremely efficient fake method! It's a lot less work to do a small amount point sampling than actually evaluating the area across a potentially infinite number of points. Given that we're willing to sample more than one point per pixel, it's also efficient in hardware, and faster, to assume that you can sample these points (2, 4, or more) at the same time, and process them together. This is part of the cleverness of multisample antialiasing. In the long term, I hope that display resolutions will get large enough that the amount of sampling will become less important. In the short term, though, I believe that FSAA as we know it is here to stay.

Now, on to anisotropic filtering. You know, trilinear filtering was good enough for the best of the best flight simulators for many many years, and every time you or I have been in an airplane, the pilot flying it was probably trained in a simulator that used trilinear filtering! Anisotropic filtering is a relatively subtle effect of texture filtering, allowing severely oblique-angled textures to look both sharp and smooth at the same time. It's not really a "fake" method—it's a better estimate of the "perfect" filtering than either bilinear or trilinear filtering. And, there are still better techniques beyond anisotropic filtering, but they require even more effort and hardware.

http://www.extremetech.com/article2/0,1697,1745057,00.asp

My highlights; if I'd have far more sophisticated texture filtering algorithms than today in the future and very high resolutions would aim to "replace" in some way multisampling, I'm not so sure it couldn't work out in the end.

 

[Moloch]

but high res just makes them smaller.
I just want gpus to become highly effcient with high levels of fsaa.. 8 sample msaa+ 2x TAA with 20% performance drop.
:smile:

 

[DemoCoder]

The human eye itself is subject to aliasing. See for example

I once read a report of a guy who had underwent a wavefront topological laser scan of his eye and laser probing of aberrations in his eye, then custom contact lenses were constructed for him, much like the Keck Telescope, which not only correct for the shape of his eye, but internal aberrations as well.

The result was 20/5 vision. The eye starting out had 20/15 or so. He was a star baseball player, and said his execellent vision helped resolve the ball better in flight. But with the contact lenses, he had 20/5. He reported that the world looked as if a blurry veil had been lifted. Everything was crisp and sharp. Eventually it was discovered he saw *aliasing* artifacts when he looked at certain angles.

There's a paper here that talks about the theory Making the best contact lenses in theory

 

[Reverend]

but high res just makes them smaller.

Yup.

Damn, I need to blame somebody for this. Guess who I usually end up blaming...

 

[Ailuros]

but high res just makes them smaller.
I just want gpus to become highly effcient with high levels of fsaa.. 8 sample msaa+ 2x TAA with 20% performance drop.
:smile:

No doubt about that; yet if you have all polygon intererior data antialiased and you'd look at a 4096*2304 resolution on just a 21" inch screen I'm questioning whether you'll see any of the polygon edge aliasing at all anymore, due to the jaggies being so extremely small.

I'm judging actually from 2048*1536 on this 21"; poly edge aliasing is still to quite small degree noticable at almost twice as high resolution I doubt you could notice anything anymore. Presupposition though that poly interior data will be properly antialiased. Just a thought anyway.

As a rather passionate AA advocate if it would be possible I would beg str8 away for Pixar AA quality/sample density for both FSAA and motion blur heh; I'm just wondering where this stuff with the high resolutions actually leads to and if there's anything behind it that makes sense.

 

[KimB]

No doubt about that; yet if you have all polygon intererior data antialiased and you'd look at a 4096*2304 resolution on just a 21" inch screen I'm questioning whether you'll see any of the polygon edge aliasing at all anymore, due to the jaggies being so extremely small.

Increasing the resolution will do nothing more than reduce the frequency of scenarios where aliasing is blatantly obvious.

Even at 4096x2304 resolution, for example, a wide flight of stairs viewed from far away (as seen in City of Heroes) will be horribly aliased if no anti-aliasing is used: the moire effects will be extremely visible no matter the resolution. All that higher resolution does is forces you to move back further before seeing the moire.

 

[Moloch]

Yup.

Damn, I need to blame somebody for this. Guess who I usually end up blaming...

Geoge Lucas?

 

[Ailuros]

Increasing the resolution will do nothing more than reduce the frequency of scenarios where aliasing is blatantly obvious.

Even at 4096x2304 resolution, for example, a wide flight of stairs viewed from far away (as seen in City of Heroes) will be horribly aliased if no anti-aliasing is used: the moire effects will be extremely visible no matter the resolution. All that higher resolution does is forces you to move back further before seeing the moire.

Since you mention moire patterns I don't think you understood me; oversimplyfied I could think of a scenario like that if the very high resolution would be used only to compensate for polygon edge/intersections aliasing.

I've fiddled around with quite a few nasty scenarios in 2048 and hence my presupposition that polygon interior data has to be antialiased for such a scenario to work. Probably badly phrased from my behalf.

radeonic2,

Weren't there recently ultra high resolution shots from UT2007 with a 3200 width (don't recall the height)?

 

[KimB]

Since you mention moire patterns I don't think you understood me; oversimplyfied I could think of a scenario like that if the very high resolution would be used only to compensate for polygon edge/intersections aliasing.

Er, I was highlighting a scenario where polygon edge aliasing results in a moire effect.

 

[Moloch]

Since you mention moire patterns I don't think you understood me; oversimplyfied I could think of a scenario like that if the very high resolution would be used only to compensate for polygon edge/intersections aliasing.

I've fiddled around with quite a few nasty scenarios in 2048 and hence my presupposition that polygon interior data has to be antialiased for such a scenario to work. Probably badly phrased from my behalf.

radeonic2,

Weren't there recently ultra high resolution shots from UT2007 with a 3200 width (don't recall the height)?

I dont know.. I dont really keep ip on mainsteam games aside from what I see at techreports frontpage and here.. and apparrently I missed the boat here

 

[Ailuros]

Er, I was highlighting a scenario where polygon edge aliasing results in a moire effect.

I don't think I've ever noticed such a scenario, probably because MSAA in the past years was rarely ever disabled on my rigs. Any example where it occurs (as in a game for instance)?

 

[KimB]

I don't think I've ever noticed such a scenario, probably because MSAA in the past years was rarely ever disabled on my rigs. Any example where it occurs (as in a game for instance)?

City of Heroes, Galaxy City, just get far away from the stairs by the police station.

 

[Ailuros]

Don't have either/or, but now that you mentioned staircases I'll keep an eye for it; I've been willing to run a couple of tests in 2048 anyway.

 

[KimB]

Alrighty, here's a screenshot. The moire was degraded slightly due to compression (sorry about that), and it is due only to geometry. Setting was 1280x960 with 2x AA. Moire did occur at a closer distance as well, though it wasn't as pronounced.

 

[Ailuros]

Uhmmm don't shoot me, but did you have "quality" AF enabled?

 

[KimB]

Uhmmm don't shoot me, but did you have "quality" AF enabled?

Heh, this was done in Linux, and I had the game set to "soft" texture quality, so I really don't think it was texture aliasing.

It is true, however, that if I turn off anisotropic filtering, it does become harder to see the moire pattern. It is still there, but the contrast is just smaller. I think this is due to a difference in the MIP maps used, rather than due to texture aliasing.

Edit: yes, the stairs do appear much darker with anisotropic enabled. I suspect that's the difference, not the texture aliasing.