Frame Rate Analysis Thread (Simple Rules Post #2)

[Rangers]

Hey - did anyone else notice a huge amount of screen tearing in Star Wars: Force Unleashed? I'm doing some comparisons now - but I don't seem to get them in the PS3 version - but they are PROMINENT in the 360 version (I have them both being upscaled to 1080p forced on the same display)?

Jack

I thought there was a lot of tearing too. Very annoying. I'm about as far from a video expert as you can get though, so maybe something else was giving me that impression. But I really felt the tearing in the 360 demo was pretty bad IIRC.

Maybe it's the difference between the XBL DEMO and the final game?

 

[grandmaster]

Mercenaries 2 is 30fps on both Xbox 360 and PlayStation 3. The former version has occasional screen-tear - with the latter version it's ever-present. When things are calm, the tear is at the very top of the screen, hidden in the overscan area and barely noticeable. The tear creeps downwards the more the engine has to render.

 

[Cyan]

37,563 frames captured - an entire runthrough of the demo and the frame rate is 29.97fps. Screen tear measurement is negligable over the course of the demo - around 5% torn frames, but this is mostly down to the size of the sample. Mostly it's v-locked, but there are spikes - the cutscene at the beginning being one of them.

I know you use Digital Foundry TrueHD, the most advanced equipment in the world to capture frames, but the question that intrigues me the most concerns the duped frames.

How do you know a frame is duped? I mean, how do the "fossilisation" of the first duped frames compare with the other duped frames when they shouldn't look the same? afaik, AA patterns don't change with frames, but freely.

I mean, in any game, when you stop walking or whatever, the AA seems to be always "moving" even if everything else is standing still.

 

[grandmaster]

See post #53 for a description of methodology. It also includes a dump of the frame rate detector's output in a situation exactly as you describe - a scene from Race Driver GRID were the screen is seemingly completely stationary. Even in this situation we can see the 30fps update - every other frame is 0% different from the preceding one.

 

[Cyan]

See post #53 for a description of methodology. It also includes a dump of the frame rate detector's output in a situation exactly as you describe - a scene from Race Driver GRID were the screen is seemingly completely stationary. Even in this situation we can see the 30fps update - every other frame is 0% different from the preceding one.

Thanks for the reply. That makes it a lot more clear to me. At first I thought you did it manually, but it wouldn't be the definition of working, precisely, but having hardships.

Cheers, keep up the good work

 

[user542745831]

Anyone checked the framerate for BioShock PS3 demo in 720p mode?

And what about the framerate in 1080i/1080p mode?

 

[chambers]

Hi guys, I just registered and would like to point out something I noticed in the bioshock demo. In the options there is horizontal FOV lock option which is enabled by default, and when this is turned off the game seems to run much smoother and look slightly better given the expanded view.

 

[thambos]

The Brothers in Arms: Hell's Highway demo for PS3 seems to suffer from a low frame-rate. It seems very apparent during cut scenes. Anybody else notice it?

 

[el_rika]

The Brothers in Arms: Hell's Highway demo for PS3 seems to suffer from a low frame-rate. It seems very apparent during cut scenes. Anybody else notice it?

The Xbox360 version is pretty much identical as far as framerate and pop ups issues.
Considering how average it looks, the game could very well be the worse optimised game yet. Or at least top 3.

 

[pjbliverpool]

There's a LEGO Indiana Jones demo on PSN. Egads, that frame lock is the worst addition to a game in the history of gaming! For any that questioned grandmasters view on it, try it yourself. It's an afront to gamers. The developers should have targeted 60 fps from the other and stuck to it if they didn't want screen-tear, rather than offering juddervision (is that really just a lock to 30 fps? Other 30 fps titles don't jig about as IJ does at 30 fps.)

The PC version seems to be very well optimised.

I got the demo today and i'm getting a solid 60fps with vsync on at max graphics, 1920x1200, 16xCSAA, in game Edge AA (not sure if this works with the previous) and 16xAF.

Thats all on a relatively modest 8800GTS 640MB.

 

[woundingchaney]

The PC version seems to be very well optimised.

I got the demo today and i'm getting a solid 60fps with vsync on at max graphics, 1920x1200, 16xCSAA, in game Edge AA (not sure if this works with the previous) and 16xAF.

Thats all on a relatively modest 8800GTS 640MB.

Realistically though I would surprised if you werent getting similar results out of Lego IJ. I dont see how this could be considered a taxing game by any sense (regardless of AA, resolution, or AF the AA is rather impressive though).

 

[Cyan]

The Xbox360 version is pretty much identical as far as framerate and pop ups issues.
Considering how average it looks, the game could very well be the worse optimised game yet. Or at least top 3.

This shows current consoles generation is one of the best ever. It may be coming to an end but there are a lot of games, good or bad -like this one- with decent technology.

I'm going to miss the millimetrical comparisons of PS3/360 games in this forum

 

[Statix]

He has clarified things a bit, though. Both versions are basically the same, they are so similar that the main feature used to distinguish the two games is that PS3's version drops to 25 frames occasionally, under heavy load conditions.

Google translator again....

I find that plainly WRONG, and quite suspicious on his part. If you would just watch the new XBOX 360 version videos at GamersYDE.com, you can clearly see that the XB360 version drops to around ~15 fps at some intense points.

I highly doubt that the final retail 360 version of the game will remain at 30 fps constant at ALL times, forever. Every game has significant frame drops during some very intense scenes... even COD4... If the worst the PS3 version drops down to is 25 fps, then I think PS3 fans have reason to rejoice.

 

[Cyan]

I find that plainly WRONG, and quite suspicious on his part. If you would just watch the new XBOX 360 version videos at GamersYDE.com, you can clearly see that the XB360 version drops to around ~15 fps at some intense points.

I highly doubt that the final retail 360 version of the game will remain at 30 fps constant at ALL times, forever. Every game has significant frame drops during some very intense scenes... even COD4... If the worst the PS3 version drops down to is 25 fps, then I think PS3 fans have reason to rejoice.

I've seen some videos and it seems to be the case. What I'm not so sure is if it's a console thing or just a problem with the video format.

The framerate drops to an indecent quantity of frames.

 

[GarretASUS]

You may as well ask why a whole host of PS3 cross-platform titles don't use AA. You aren't comparing like with like - Killzone 2 was developed for PS3 exclusively, Mirror's Edge uses an engine that seems to work better on Xbox 360. With regards the hit in performance, PS3 Mirror's Edge does indeed have marginally more screen tear than the 360 version in like for like scenes in the code as it stands, so introducing further processing may indeed impact the performance. 720p60 comparison vid is now up on my blog.

Just watched the video. Is it me or does the 360 version have occasion frame skipping issues? Very noticeable at the start with the pan down and over. And while the 360 version does have the benefit of AA, I find my eye is liking the lighting in the PS3 version. Texturing looks identical.

 

[GarretASUS]

For me the 360 version seems to have more fps dips and there are still jaggie issues, albeit not as much as PS3.

 

[grandmaster]

If you read the piece you'll find that frame rate analysis shows that both are pegged at 30fps, maintaining it if required by turning off v-lock. As it is, the PS3 version is marginally more prone to tearing frames, as seen in the intro sequence. If you're watching the streaming video version, that should in no way be taken as an indication of proper frame rate as Flash drops frames like a bastard. Also, both were running at default brightness and contrast but they can be tweaked to match if that's what you want. This sort of discussion should really be in the multiformat platform development thread to be honest.

 

[GarretASUS]

If you read the piece you'll find that frame rate analysis shows that both are pegged at 30fps, maintaining it if required by turning off v-lock. As it is, the PS3 version is marginally more prone to tearing frames, as seen in the intro sequence. If you're watching the streaming video version, that should in no way be taken as an indication of proper frame rate as Flash drops frames like a bastard. Also, both were running at default brightness and contrast but they can be tweaked to match if that's what you want. This sort of discussion should really be in the multiformat platform development thread to be honest.

Well I watched the high-def downloadable version. I understand the video can't be a total accurate assesment of frame rate, but why does it seem the 360 side is slower at times. Look especially at the part when she jumps from the far distance at 38 seconds in. As she slides runs down, the 360 side looks slower and less smooth.

 

[grandmaster]

Well I watched the high-def downloadable version. I understand the video can't be a total accurate assesment of frame rate, but why does it seem the 360 side is slower at times. Look especially at the part when she jumps from the far distance at 38 seconds in. As she slides runs down, the 360 side looks slower and less smooth.

Edit: yes there are some torn frames when sliding down the red thing, closer to 44-45 seconds - at 38 seconds everything is 30fps on both consoles. The PS3 performs similarly in terms of torn frames in other parts earlier on in that sequence.

 

[AlNom]

For me the 360 version seems to have more fps dips

Perception in this regard does not seem to be as reliable as the analyser tool grandmaster has at his disposal. If you read back through this FPS thread, you'll find that it is quite accurate.

10,875 frames (around 3 minutes of like-for-like video), only 178 frames were torn in the 360 version - a paltry 1.6%.
The PS3 version is more difficult to analyse as the vast majority of the torn frames are restricted to the first couple of lines of resolution - basically imperceptible to the human eye, most likely hidden in the overscan area of your HDTV, and therefore shouldn't be included in the analysis. So adjusting the detector to disregard those, the figure rises to 3.8%. I

In the grand scheme of things? Not a big difference. The PR spin would be that the PS3 has more than twice the torn frames as the 360 for the ~10.9K frames analysed. But that would a rather daft argument.