On the Tekken 6/Sub-HD article, there's a mention that Tekken 6 may be based off the SC4 code. Wouldn't it be more accurate to say that they probably share code? Tekken 6 was out in arcades (and supposedly running on a very PS3-like board) more than 6 months before Soul Calibur was released on consoles.
Digital Foundry tech analysis channel at Eurogamer
- Thread starter FirewalkR
- Start date
- Status
Reading the lastest DF on Forza 3, he mentions that the 360's "relative paucity of the 360's texture cache" as a reason for the lack of AF.
Is there somewhere I can read more about that, or can someone give some insight? Besides the dithered shadows, my biggest gripe with the IQ of the 360 is the lack of texture filtering.
http://www.eurogamer.net/articles/digitalfoundry-tech-analysis-forza-3-demo
Is there somewhere I can read more about that, or can someone give some insight? Besides the dithered shadows, my biggest gripe with the IQ of the 360 is the lack of texture filtering.
http://www.eurogamer.net/articles/digitalfoundry-tech-analysis-forza-3-demo
Alucardx23
Regular
First of all I want to say hi because this is my first post on beyond 3D. I have been reading on this forum for a long time and as most of you do I like to learn how games are made instead of just playing then and I find beyond 3D Forums the best for that.
Ok my question goes to Grandmaster and is regarding your article on Digital Foundry “Tekken 6 PS3/360: The resolution game”, the issue comes from a discussion that I was having with a friend regarding the performance of the Tekken 6 engine on 360 and PS3. He says that because in your article you stated this, “Overall, although there's really nothing in it in the heat of gameplay, we'd take the PS3's blur-off 576p 2xMSAA picture as the best IQ option available across both versions”, the Tekken engine performs better on PS3.
The thing is that I find that if the 360 can manage to have the most stressful effect in the game with is “fulltime adjustable animation blur” and still has room to include some kind of texture filtering; (possible anisotropic filtering as stated in your article) isn’t this out of the 4 modes the one that requires more performance?
The other question is regarding image quality comment; in the article when you state “we'd take the PS3's blur-off 576p 2xMSAA picture as the best IQ option available across both versions” Is this and opinion or something that relates to the performance of the engine? I mean can I say that because I prefer the better texture filtering with motion blur on the 360 it gives a better image quality? Or would I be wrong for saying that?
And the last question is regarding the decision of Namco to get the 360 Tekken running at 1365x768 versus the 2X AA on the PS3. On your article you state this “Xbox 360's answer to AA is to supersample 1365x768 down to 720p and lower texture filtering.” Isn’t supersample the most stressful way of trying to remove aliasing when is compared with 2xMSAA? I understand that Namco chose the fastest way instead of trying to use tiling on 360 and because it made the same mistake as in Soul Calibur 4 of downscaling the image by software it doesn’t allow the higher resolution to be as good has it can.
I was telling my friend that because you said this:
“We suspect that fill-rate and pixel shading are the main issues in maintaining Halo 3 performance and it's here that we see much the same reasons why a selection of PlayStation 3 cross-platform games run at a lower resolution compared to their Xbox 360 counterparts.”
It is easier and less stressful for performance on the PS3 to include 2X AA instead of upping the resolution as on 360.
Anyway I also want to tell you that I’m a really big fan of Digital foundry when it comes to game performance analysis you are the best. I hope you can answer some of my questions, thanks in advance.
PS: Excuse my English because it’s not my native language. :smile:
Ok my question goes to Grandmaster and is regarding your article on Digital Foundry “Tekken 6 PS3/360: The resolution game”, the issue comes from a discussion that I was having with a friend regarding the performance of the Tekken 6 engine on 360 and PS3. He says that because in your article you stated this, “Overall, although there's really nothing in it in the heat of gameplay, we'd take the PS3's blur-off 576p 2xMSAA picture as the best IQ option available across both versions”, the Tekken engine performs better on PS3.
The thing is that I find that if the 360 can manage to have the most stressful effect in the game with is “fulltime adjustable animation blur” and still has room to include some kind of texture filtering; (possible anisotropic filtering as stated in your article) isn’t this out of the 4 modes the one that requires more performance?
The other question is regarding image quality comment; in the article when you state “we'd take the PS3's blur-off 576p 2xMSAA picture as the best IQ option available across both versions” Is this and opinion or something that relates to the performance of the engine? I mean can I say that because I prefer the better texture filtering with motion blur on the 360 it gives a better image quality? Or would I be wrong for saying that?
And the last question is regarding the decision of Namco to get the 360 Tekken running at 1365x768 versus the 2X AA on the PS3. On your article you state this “Xbox 360's answer to AA is to supersample 1365x768 down to 720p and lower texture filtering.” Isn’t supersample the most stressful way of trying to remove aliasing when is compared with 2xMSAA? I understand that Namco chose the fastest way instead of trying to use tiling on 360 and because it made the same mistake as in Soul Calibur 4 of downscaling the image by software it doesn’t allow the higher resolution to be as good has it can.
I was telling my friend that because you said this:
“We suspect that fill-rate and pixel shading are the main issues in maintaining Halo 3 performance and it's here that we see much the same reasons why a selection of PlayStation 3 cross-platform games run at a lower resolution compared to their Xbox 360 counterparts.”
It is easier and less stressful for performance on the PS3 to include 2X AA instead of upping the resolution as on 360.
Anyway I also want to tell you that I’m a really big fan of Digital foundry when it comes to game performance analysis you are the best. I hope you can answer some of my questions, thanks in advance.
PS: Excuse my English because it’s not my native language. :smile:
I'm fairly certain it's not AF, but a negative LOD bias that they employ at lower resolution to sharpen the details. Just think about the angles and mip maps for this scenario...
Yes, rendering more pixels will be more stressful than sub-pixel sampling (MSAA) in that whatever pixel shaders they are using are run per-pixel. Apparently the game is not particulary pixel processing limited. I would be interested to see how high they could go if they built the engine around the 360. The limited environments and characters does not convince me that tiling, and thus vertex processing, is a problem. :s
Hard to say. 1024x576 with 2xMSAA takes up more memory (and bandwidth) than 1365x768. On RSX 2xMSAA would have less Z-fillrate performance even, but... who knows. The only differences, of course, are pixel shading/vertex shading. *shrug*
Yes, rendering more pixels will be more stressful than sub-pixel sampling (MSAA) in that whatever pixel shaders they are using are run per-pixel. Apparently the game is not particulary pixel processing limited. I would be interested to see how high they could go if they built the engine around the 360. The limited environments and characters does not convince me that tiling, and thus vertex processing, is a problem. :s
Hard to say. 1024x576 with 2xMSAA takes up more memory (and bandwidth) than 1365x768. On RSX 2xMSAA would have less Z-fillrate performance even, but... who knows. The only differences, of course, are pixel shading/vertex shading. *shrug*
Alucardx23
Regular
Thanks for answering some of my doubts AlStrong. Grandmaster seems to be pretty busy right now to answer my question so can someone else point me in the right direction. Where do you assume the Tekken 6 engine has better performance and regarding the image quality question is it something subjective that has to do with taste or when they say in the Digital Foundry article "we'd take the PS3's blur-off 576p 2xMSAA picture as the best IQ option available across both versions" is related to performance?
Can I say that I prefer this IQ. Motion Blur on?
Instead of the IQ that is produced by motion blur off.
As far as I know, the game is 60fps across all settings and platforms. I believe they are referring to the MSAA and sharper textures in their preference there. I haven't seen the motion blur... in motion, but my own impression is that both of those benefits being applied at all times is of more value to overall image quality than something that is infrequent and perhaps not really needed in a 60fps game.
MAG interview is up:
http://www.eurogamer.net/articles/digitalfoundry-mag-factor-article
Great job Grand
http://www.eurogamer.net/articles/digitalfoundry-mag-factor-article
Great job Grand
Alucardx23
Regular
What I understand is that the 360 version is the one that has better textures when the motion blur is turned on and the PS3 one has 2X AA when motion blur is turned off. That’s why I was making the question about image quality, better texture filtering and motion blur vs 2xAA.
"Removing the motion blur filter frees up a lot of resources, and Namco has chosen to deploy these in two different ways on each console. Xbox 360 gets an enormous resolution boost to 1365x768, while the PS3 gets 2x multi-sampling anti-aliasing. In terms of overall image quality across the two modes and two consoles, the PS3 gets the nod in "blur off" mode thanks to decent enough upscaling based on an anti-aliased image, while in default mode with the motion blur active, the 360's enhanced texture filtering gives clear image quality advantages. "
I was trying to find something about the framerate differences and I found this comment from a recognized Tekken player, these comments are from the E3 2009 build. This slowdown also happens in the Arcade version of the game.
MarkMan
“They are BOTH good, the 360 one ran just as good if not better than the PS3 one. I noticed slowdown during rage effect/floor break on PS3 version, but never noticed the same thing on 360.”
http://sdtekken.com/2009/06/02/ign-e3-tekken-6-report/
In this video you can see the deference between motion blur on and off.
http://www.youtube.com/watch?v=R5pbREO3bdg&feature=player_embedded
That report is kind of old today and the 360 didnt have the blur effect back then. Some things might have changed since then
grandmaster
Veteran
So, I think it's high-time we did a calibration test of the Xbox 360's video output. The Uncharted 2 Photo Mode proved fairly conclusively that full-range RGB output on PS3 is match for the framebuffer, but I am not sure it will be the same for Xbox 360.
With few developers calibrating to the output of the 360, and presumably relying on the 1s and 0s in the framebuffer, I'm curious about what settings are required to get the closest possible match.
Is there any way to get a test pattern on the 360 that doesn't involve the HD-DVD player, the DVD player or the JPEG viewer? Ideally a test pattern in-game would be good. Failing that I guess a screenshot would do... but a screenshot of what, and from where?
With few developers calibrating to the output of the 360, and presumably relying on the 1s and 0s in the framebuffer, I'm curious about what settings are required to get the closest possible match.
Is there any way to get a test pattern on the 360 that doesn't involve the HD-DVD player, the DVD player or the JPEG viewer? Ideally a test pattern in-game would be good. Failing that I guess a screenshot would do... but a screenshot of what, and from where?
Wouldn't the best solution be a dirt-simple XNA application? You could load it up with whatever test patterns you wanted then.
grandmaster
Veteran
OK, would any one want to oblige with such an app? Is it easy enough to run someone else's XNA app on my 360? And would it work on a test kit, so I can take the direct framebuffer dump?
Alucardx23
Regular
The arcade version of Tekken 6 which is based on PS3 hardware has slowdown doing the same thing as the PS3 version, so it would be nice to know if they fix that. How do you know that the 360 version did not have motion blur in the E3 version?
You'll need at least a trial Creator's Club Membership to test the app on a 360. I got mine from the recent Dream Build Play 2009 competition. But that's over & they're are no longer offering them. Check the footnotes here to find out if additional places are still available. If you can't find any free, then you'll need to purchase a Premium membership (4 months for $49 or $99 for 12 months).
BTW, All the tools are meant to work with retail boxes. So I don't think it will give you anything more if you have a test kit.
http://creators.xna.com/en-US/faq
Tommy McClain
That was what was reported back then.
It doesnt matter if the arcade version had slowdown. This is something I am familiar already since t was noted in the same article.
Since then the slowdown may have been fixed.
Sounds like Indie Games will be the cheaper option though.
Alucardx23
Regular
Yeah I was trying to find info about that, when you wrote that the 360 version did not had motion blur I thought that you had it confirmed from somewhere. The thing is that neither of us knows if they have fixed the slowdown, because it may still be there or not. I’m just taking the last confirmation about slowdown on the PS3 version as reference until Digital Foundry does a framerate analysis.
grandmaster
Veteran
This looks ideal, but I very much doubt I'll be able to run it on the test kit. I do have a retail unit of course, but the idea here will be to compare the framebuffer with the HDMI capture.
With the test kit I can dump the framebuffer and simultaneously capture the HDMI output.
- Status
Similar threads
