The Framerate Analysis Thread part 2

[Jazz9]

on the RE5 X360 demo in 1080p there are lot of tearing and not in 720p but it's the same native resolution and AA, just upscale, it's strange... X360 upscale not free?
select the first level, don't move, just rotate the camera and look the tearing

I can confirm that.
I first played the demo in 1080p and the tearing is absolutely brutal.
I then tried it in 720p and I don't remember the game tearing even once (maybe it did but I sure didn't notice it).
Makes me wonder about leaving the 360 on 720p despite having a 1080p display, but I think RE5 may be something more of an anomaly.
I'll be waiting for Grandmaster's test eagerly

 

[GarretASUS]

I can confirm that.
I first played the demo in 1080p and the tearing is absolutely brutal.
I then tried it in 720p and I don't remember the game tearing even once (maybe it did but I sure didn't notice it).
Makes me wonder about leaving the 360 on 720p despite having a 1080p display, but I think RE5 may be something more of an anomaly.
I'll be waiting for Grandmaster's test eagerly

Well one would hope it is a bug of some sorts in the demo, but we will see in a month.

 

[GuestLV]

Originally Posted by Quaz51
on the RE5 X360 demo in 1080p there are lot of tearing and not in 720p but it's the same native resolution and AA, just upscale, it's strange... X360 upscale not free?
select the first level, don't move, just rotate the camera and look the tearing

It is a Capcom engine bug seen first in Dead Rising demo (when i set xbox display settings on ''optimal res'' on my 32 philips 1366х768) screen tear constantly present, next time i choose 720p in xbox display menu and tear completely gone, the same bug present in Re5 demo

 

[grandmaster]

Right, I've captured the Resident Evil 5 Shanty Town level up to the chainsaw guy, and included a 'spin the camera around the main character' moment at the beginning of the video.

The results are:

28,886 frames captured, 29.97fps, 0.37% torn frames.

For the public arena level, I captured it at 1080p60, then edited out all the black screens that crop up before and after cut-scenes as these world skew the frame analysis. I also did the 'spin around the character' exercise at the beginning of the stage too.

17,032 frames captured, 29.2fps, 5.73% torn frames.

This is clearly at odds with what Quaz is seeing. My feeling is that if there is a bug in the Capcom engine, it only affects certain 360s. My unit is an NTSC Elite (Zephyr model). The only other explanation is that the 1080p signal is not running at the right timing for the display. Quite why only Capcom games are affected is a mysterious though.

The only other thing I should point out is that I'm running at 1920x1080, ie the console sees TrueHD as a DVI monitor rather than an HDMI display. What I have noticed is that if you select 1920x1080, it asks you to confirm that the picture is OK before locking the resolution. If you say NO, it invokes a second 1920x1080 mode using slightly different timings.

However I also use an HDCP decrypter that identifies itself as an HDTV, and this brings up the more usual HDTV settings on the console set-up. In previous tests, this has produced the same output as 1920x1080.

In terms of these Resi captures though, I can scan through and confirm the results by eye, and I can see the torn frames - where they are - too.

 

[Boxercide]

hay grandmaster, i know you said you'd run a full workup on Killzone when it's out, but since you do the opening of levels a lot, could you maybe try the opening sequence until the player gets control in the demo? It seems slightly slower than 30 fps in at certain times, but nothing too terrible.

 

[GarretASUS]

Right, I've captured the Resident Evil 5 Shanty Town level up to the chainsaw guy, and included a 'spin the camera around the main character' moment at the beginning of the video.

The results are:

28,886 frames captured, 29.97fps, 0.37% torn frames.

For the public arena level, I captured it at 1080p60, then edited out all the black screens that crop up before and after cut-scenes as these world skew the frame analysis. I also did the 'spin around the character' exercise at the beginning of the stage too.

17,032 frames captured, 29.2fps, 5.73% torn frames.

This is clearly at odds with what Quaz is seeing. My feeling is that if there is a bug in the Capcom engine, it only affects certain 360s. My unit is an NTSC Elite (Zephyr model). The only other explanation is that the 1080p signal is not running at the right timing for the display. Quite why only Capcom games are affected is a mysterious though.

The only other thing I should point out is that I'm running at 1920x1080, ie the console sees TrueHD as a DVI monitor rather than an HDMI display. What I have noticed is that if you select 1920x1080, it asks you to confirm that the picture is OK before locking the resolution. If you say NO, it invokes a second 1920x1080 mode using slightly different timings.

However I also use an HDCP decrypter that identifies itself as an HDTV, and this brings up the more usual HDTV settings on the console set-up. In previous tests, this has produced the same output as 1920x1080.

In terms of these Resi captures though, I can scan through and confirm the results by eye, and I can see the torn frames - where they are - too.

If it helps, my 360 is PAL and tried through 1080p component and HDMI 50 & 60Hz and still the same tearing.

 

[makattack]

Not sure if this will contribute anything useful, but I used my NTSC 360 connected via HDMI to a Gateway FHD2400 monitor to generate this video captured off-the-screen with a Flip video recorder:

http://tridml.bc.edu:9080/b3d/r5_1080p.avi

http auth required to prevent "accidental" d/l's of the big file:
u:b3d
p:screens

Details:
r5_1080p.avi: RIFF (little-endian) data, AVI, 640 x 480, 30.00 fps, video: XviD, audio: ADPCM (mono, 44100 Hz)
35187110 Feb 8 21:42 r5_1080p.avi

It's big... 35MB approx. but I can't really detect any screen tearing while panning the camera. Tried max elevation, max declination, and mid-level.

360 set to 1080p, monitor upscaling disabled.

Edited to add: Obviously, this is very primitive compared to grandmasters capabilities, but is quick and dirty for those curious.

 

[AlNom]

The only other thing I should point out is that I'm running at 1920x1080, ie the console sees TrueHD as a DVI monitor rather than an HDMI display.

So no overscan?

What I have noticed is that if you select 1920x1080, it asks you to confirm that the picture is OK before locking the resolution. If you say NO, it invokes a second 1920x1080 mode using slightly different timings.

Curious... Do you have further details on this?

 

[user542745831]

So no overscan?

If you connect your Xbox 360 to a Computer/DVI display over the Xbox 360 HDMI output, then the available resolutions in the dashboard differ from the usual HDTV resolutions (720p, 1080i, 1080p) which are available when connecting a Xbox 360 to a 'real' HDTV display device over the HDMI output of the Xbox 360.

In fact you get the resolutions which also become available when using a VGA cable. Like 1024 x 768, 1280 x 1024, 1440 x 900, 1680 x 1050 and so on.

And instead '720p' you can then choose '1280 x 720' and instead '1080p' you can choose '1920 x 1080'.

What he meant was likely that his Digital Foundry TrueHD capture system is recognized by his Xbox 360 as a Computer/DVI display instead of a 'real' HDTV display device. So he can't choose '1080p' but only '1920 x 1080' in the dashboard.

So this has nothing to do with overscan.

I could imagine he mentioned it, because he wasn't sure if the '1920 x 1080' output of the Xbox 360 slightly differs from the '1080p' output in timings and so on.

Furthermore his Digital Foundry TrueHD capture system simply has nothing to do with overscan, ... it's not a display .

If you say NO, it invokes a second 1920x1080 mode using slightly different timings.

I have another own question:

What exactly does the PS3 / Xbox 360 output if it outputs 720p/1080i/1080p over HDMI?

Does it output 720p59.94 / 1080i59.94 / 1080p59.94?

Or does it output 720p60 / 1080i60 / 1080p60?

If it uses both (59.94 and 60), does it depend on what the console is displaying (games / movies)?

And what about using a VGA cable for the Xbox 360 (counting in PC resolutions)? Is it 59.94 or 60?

Can you tell what PS3 is outputting when '1080p 24 Hz (HDMI)' option for Blu-ray Disc is activated, too? 23.976 or 24? If it uses both (23.976 and 24), does it depend on what the Blu-ray Disc / AVCHD video is encoded at?

 

[AlNom]

So this has nothing to do with overscan.

? All I wanted to know was if the 1920x1080 mode had overscan as a PC mode versus regular "1080p". It was just a fact-question I was curious about.

 

[grandmaster]

The modes look identical to me.

Almost all (if not all) 60Hz outputs are actually running at 59.94fps - a hangover from the NTSC drop-frame format.

With regards the two 1920x1080 modes, I know this is the case because on the older Digital Foundry hardware, you'd get an ugly strip of corruption on the right-hand side of the screen in one mode, but it was fine on the other. The newer TrueHD locks on to either automatically. I never really got to the bottom of this.

 

[user542745831]

Almost all (if not all) 60Hz outputs are actually running at 59.94fps.

Thanks.

 

[the ignoramus]

There's the PC intro to Devil May Cry 4 in the TrueHD 1080p Workstation section of www.digitalfoundry.org. That's the cleanest 1080p60 video produced for PS3 to date. I am also building a new PC based on the GTX295 and will be making some Crysis and Crysis Warhead 1080p60 videos. Encoding these videos is a bit of a fine art, but I've got better at it since I did the DMC4 one.

Yes, Killzone will be deconstructed big-style once I have the code.

With regards RE5, I'll take a look at that in the next day or so. My capture hardware can downscale 1080p to 720p on the fly, so a comparison video based on the bit Quaz points out should be very quick to do and bung up online.

The real test of whether Xenos scaling is free or not would - I suspect - come from a game like Call of Duty 4/WaW: I'd imagine a v-locked 60fps game that drops frames would drop more frames and be far more noticeable.

I see your F.E.A.R. 2 analysis is up at Eurogamer.
But where is the Killzone 2 frame rate analysis? Will it be posted on digitalfoundry within the week?

 

[grandmaster]

Yes, I hope so. There'll be a teaser of the demo section (but from the full game) which will be unabridged gameplay and I think that'll go up on Digital Foundry tomorrow, and then there'll be excerpts of all the campaigns, along with an in-depth analysis which should be up before the launch.

 

[makattack]

Hope I'm not pre-empting grandmasters plan to announce this himself, but I just noticed this while looking for SF4 videos:

http://www.digitalfoundry.org/blog/?p=481

... the first of his and Alex Gohs KZ2 analysis.

 

[grandmaster]

Part two is up today along with Alex's 'primer' on deferred rendering. The final four campaigns (spread across three videos as the last two are very short) will be tomorrow.

 

[Arwin]

Part two is up today along with Alex's 'primer' on deferred rendering. The final four campaigns (spread across three videos as the last two are very short) will be tomorrow.

Great stuff, thanks!

By the way, the last youtube video on the page is still private ...

 

[FirewalkR]

Part two is up today along with Alex's 'primer' on deferred rendering. The final four campaigns (spread across three videos as the last two are very short) will be tomorrow.

Haha I read both posts and watched the first of yesterday's videos, but now I'm avoiding your site like the plague. It's only for a few days, until I get and finish the game. Hope you don't mind.

Great work, as always.

 

[grandmaster]

Great stuff, thanks!

By the way, the last youtube video on the page is still private ...

I was linking to the wrong vid. All nine have been annotated and uploaded, but I only 'unprivate' them when the post is on the blog. Doh!

Thanks for the heads-up. Weirdly embedded YouTube vids only work sporadically for me, so I can't test.

I have to admit that embedded quality with YouTube HD is amazing. There's a nice little hack for making the HD streams embeddable (simply found with Google).

 

[bagofsuck]

Can anyone tell me why i notice more tearing when i play the re5 demo through HDMI and when i use component is almost impossible to notice the tearing..?