The Framerate Analysis Thread part 2
- Thread starter grandmaster
- Start date
user542745831
Veteran
As far as i know, only the US NTSC version of Uncharted: Drake's Fortune features a 1080i/1080p mode with 960 x 1080p rendering.
Edit:
http://www.eurogamer.net/articles/no-1080i-in-uncharted-pal-after-all
Edit:
http://www.eurogamer.net/articles/no-1080i-in-uncharted-pal-after-all
Last edited by a moderator:
user542745831
Veteran
http://www.digitalfoundry.org/blog/?p=682
Nice
.So, are you going to use your EU PAL version or another US NTSC version?
grandmaster
Veteran
Well I'm not going to buy the game again I'm using my EU debug review code which appears to be to all intents and purposes final. It's a curious game in that it's one of these titles that often has screen tear in the first few lines, mostly hidden in overscan. So should it be included as a torn frame? Strictly speaking it should be, but it's almost impossible to notice. I think I'll crop out the first eight lines from the analysis and re-do it.
I'm using my EU debug review code which appears to be to all intents and purposes final. It's a curious game in that it's one of these titles that often has screen tear in the first few lines, mostly hidden in overscan. So should it be included as a torn frame? Strictly speaking it should be, but it's almost impossible to notice. I think I'll crop out the first eight lines from the analysis and re-do it.the ignoramus
Regular
A performance analysis of PGR4 and Formula 1 Championship Edition would be awesome.
Both of these games are consistently overlooked in any discussion about great looking games and it annoys to me to no end.
I mentioned this before, but PGR4s combination of particle effects, motion blur, dynamic lighting and trackside geometry make the game look absolutely stunning in motion.
F1CE doesnt look as good as PGR4, but the trackside lighting in the time trials and the rain effects look sensational. No game has done a better job of nailing the scary sensation of driving at high speeds in wet weather. ( force and movement of the rain drops is calculated based on forces/velocities per pixel) Given the fact that it was a launch game for '06, (over 28 months old) I think it's aged quite well.
Both of these games are consistently overlooked in any discussion about great looking games and it annoys to me to no end.
I mentioned this before, but PGR4s combination of particle effects, motion blur, dynamic lighting and trackside geometry make the game look absolutely stunning in motion.
F1CE doesnt look as good as PGR4, but the trackside lighting in the time trials and the rain effects look sensational. No game has done a better job of nailing the scary sensation of driving at high speeds in wet weather. ( force and movement of the rain drops is calculated based on forces/velocities per pixel) Given the fact that it was a launch game for '06, (over 28 months old) I think it's aged quite well.
Last edited by a moderator:
Well I'm not going to buy the game again
I find on my NTSC US version that screen tear gets really bad center image when things get hectic. I've been told by others that it's nit picking, but if it's there, it's there.
Rendering at 1080 or 720?
Both. After I learned of this site, and yours, I stopped forcing 1080p. But even at 720, Uncharted can get some bad (to me; maybe I'm too sensitive?) screen tear. Doesn't ruin the game, but like you say "...is like getting poked in the eye with..."
My main hope for Uncharted 2 is that they can reduce/minimize it or limit it to the top.
I play Wipeout HD a lot split screen and even though that causes near constant tear, I hardly ever notice it because it's not dead center.
grandmaster
Veteran
Yes, the position of the tear varies dramatically and no v-lock is the game's biggest IQ problem. It's not nit-picking, in places it looks terrible - for example on the cinematic reveal of the U-Boat. That's why I'm looking to remove it completely from my vids and produce 'pure' 30fps footage - to see how much better it looks.
Certainly not from a frame rate point of view.
It would be neat to compare them to in-game quality though, if only to see what they cranked up specifically.grandmaster
Veteran
Not all of them are - for example the U-Boat reveal I pointed out previously.
No that kind of analysis,i'm talking about some like KZ2 engine analysis...
http://www.digitalfoundry.org/blog/?m=200902
brain_stew
Regular
Well I'm not going to buy the game again
I really think it should, many people make an effort to make sure they have no overscan and those are likely the exact same people that would be bothered the most by tearing. Include both figures if you want, but to my eyes any tearing at all looks hideous and any games that features heavy tearing should be called out for it. Its a pet peeve of mine, I just see no instance where tearing is an acceptable compromise, and I force triple buffer vsync in every PC game I play precisely because of that.
grandmaster
Veteran
Well on every video done so far (each is 9-10 minutes long), screen tear has been consistent at 45%. Almost every other frame then with just the odd exception. I don't think it's a particularly useful measurement. The whole point of having a torn frame indicator is to show where IQ is being compromised and a tear in the first couple of lines of resolution doesn't really impact it at all IMHO.
There's also the small matter that tearing on a static scene is pretty much undetectable by the human eye but still easily detectable in analysis.
There's also the small matter that tearing on a static scene is pretty much undetectable by the human eye but still easily detectable in analysis.
Last edited by a moderator:
I think the two figures are necessary to be accurate. We need a true torn-frame count as this is an absolute measure for technical comparison, but also a somewhat subjective count to account for visible IQ impact is wanted. 45% torn frames is 45% torn frames, but that metric itself is misleading and needs to be balanced, as you say, to accomodate perception, because 45% of torn frames in the top 2 lines isn't anything like as bad as 5% torn frame slap-bang middle of the screen.
Perhaps including a count of vertical position would lead to a suitable reference figure? A mean average tear with ranges. Thus 50% frames torn at line 2 would equate to an average per-frame tear of 1 pixel. A 5% tear on line 500 would be a per-frame tear average of 40 pixels. those comparing metrics without understanding (most people who like to look at figures for comparison!) will read the 40 and see that's clearly worse than 1, and appreciate the game with the ScreenTear Score of 1 is the better looking game, proven mathematically
Perhaps including a count of vertical position would lead to a suitable reference figure? A mean average tear with ranges. Thus 50% frames torn at line 2 would equate to an average per-frame tear of 1 pixel. A 5% tear on line 500 would be a per-frame tear average of 40 pixels. those comparing metrics without understanding (most people who like to look at figures for comparison!) will read the 40 and see that's clearly worse than 1, and appreciate the game with the ScreenTear Score of 1 is the better looking game, proven mathematically
Similar threads
