nAo talks about Heavenly Sword (You need to read this)
- Thread starter mckmas8808
- Start date
Fdooch said:
Pedant hat on - depends what you mean by "mathematically correct". Averaging in CIE space will give a different result to RGB space, so you might consider that it is "wrong". However it certainly smooths the edges in a visually pleasing way, so I would consider it "correct enough". You could in fact convert the buffer back to RGB *before* resolving the AA, and thus get more or less the same result - but so far I haven't found that to be necessary. In fact, doing the average in CIE will in theory give errors where the luminance is split over 2 8-bit channels and this isn't respected by the filtering... but in practice I haven't noticed any serious problems with this and so we're just ignoring this for now...
So I'm not sure "mathematical correctness" should be a consideration in this context. The only purpose is to make edges smoother, and so long as it does that, it's valid. But I suppose the short answer to your questions qould be "it's probably not correct unless you convert to RGB, but yes it mostly looks ok anyway".
MrWibble said:
The splitted component problem can be ameliorated reserving one additional carry bit in the most or least significant part, so that you can reconstruct a more
accurate luminance value later at expense of this bit (but 15 bits for the luminance are really a lot, so it's not really a big deal losing that bit)
Obviously a lerp in CIE space is not matematically correct at all since the function which maps a color from RGB to CIE is not linear, but there are quasi-linear approximations out there. Dunno if they're worth the extra shading cost (even though you can mostly hide that cost..)
mckmas8808 said:
I have seen it in action, and it is a very nice LCD display. It does not have the glossy screen coating like xbrite or similar technologies which improve contrast, but it does have a very high native contrast ratio as far as LCDs go. The backlight was fairly uniform just by quickly eyeballing the screen with no picture. brightness was good for an LCD, though certainly not up to plasma levels. We did not do any intensive twitch gaming, but I noticed no significant ghosting on the display while we were playing with it.
The display of course looks best at it's native resolution. When you use it with DVDs or the TV tuner it looks like a decent upper-midrange LCD. Not fantastic like a Bravia, but not bad either. When you give it a native signal it looks outstanding. We played world of warcraft for a while on it and it was absolutely impressive. I'd argue that the necessity for 1080p resolution is highly depending on what you are doing with the display. Games like World of Warcraft benefit greatly from it, while FPS or racing games probably would not benefit as much. For the price, it is still one of the best displays for dvd/tv watching you can get, and it significantly beats far more expensively displays for PC gaming and 1080p video. We watched the quicktime 1080p x-men 3 trailer on it and the detail was outstanding.
Nite_Hawk
london-boy said:
If you're spending that kind of money, you have to make sure there isn't something in the immediate horizon which would be better.
Looks like a wide selection of 1080p displays will be available this year for the first time.
Couple of things lacking in these new displays would be HDMI 1.3 and support for 24/48/72 refresh rates.
Still hoping Toshiba and Canon somehow pull off SED displays in the next couple of years.
Take the conversion function from CIE to RGB, call it f(x). Then define a blending function, say b(src, dst). Then look for the maximum of |f(b(x1,x2)) - b(f(x1),f(x2))|. Although, to make a bit more sense, this should be weighted by a perceptual threshold detector.Fdooch said:
The actual details are left out, as they are immediately obvious to the astute reader...
scooby_dooby
Legend
kyleb said:
No I was explaining DeanoC's own rationale when explaining why they would most likely target 720p, one of the reasons was because the 1080p install base was virtually non-existent.
You just inferred the rest.
DeanoC said:
Last edited by a moderator:
mckmas8808
Legend
wco81 said:
I thought the lower the refresh rate the better. Am I wrong in thinking this?
winterbird@nerdshack.com
Newcomer
yes, you're wrongmckmas8808 said:
Higher refresh = more frames capable of being displayed per second.
I think you're confusing refresh rate with refresh time. Refresh rate = fps, higher values = better. Refresh time, the delay in switching an LCD pixel, adds to ghosting and lower values = better.mckmas8808 said:
Response time needs to be as low as possible to avoid ghosting.
Refresh rate is the rate the screen gets updated in a certain period of time (per second) and usually the higher the better.
Refresh rate is the rate the screen gets updated in a certain period of time (per second) and usually the higher the better.
Why would racing games not benefit as much?Nite_Hawk said:
You do realize that FPS and racing games aren't really comparable right?
Racing games don't have things that move across the screen fast, and no crazy jumps (usally..) so I find that I have to use higher resolution and fsaa vs FPS games since it's easier to see aliasing when you're not moving around (the screen) very fast.
