MS in full damage control mode over 1080p

Status
Not open for further replies.
http://www.majornelson.com/archive/2006/10/22/More-on-1080p.aspx

I got an email last night from Bruce Dawson. Some of you may remember Bruce as the co-author of the console comparison that I posted after E3 2005. Bruce was thinking about 720p vs 1080p in ways that only a super-smart engineer can.

index.gif


:)

His first post got totally owned..

http://ozymandias.com/archive/2006/...rence-Between-1080i-and-1080p-for-Movies.aspx

In fact, I'll stick my neck out and predict that that you won't see any 1080"x" games for the PS3 this year.

So now he went to this..

http://www.ozymandias.com/archive/2...oughts-on-High-Definition-Game-Rendering.aspx

I was talking to Bruce Dawson, one of our senior software design engineers here, about some questions I had around 1080i and 1080p. Frankly, I was particularly curious about why Sony has continued harping on 1080p as being "TrueHD", especially since the 360 has enabled 1080p output as well (coming soon to homes near you!) I was trying to figure out if I was just missing something, and his emailed answer was particularly clear and helpful to me, and since there's nothing confidential here I thought I'd share it with you.

The really interesting statistic that popped for me is how much less time a game console has to render a 1920x1080 scene versus a 1280x720 scene. (Remember this is on the same console, whichever one you like. This is not a comparison of different console's rendering capabilities to each other.) Simply put, for a 1080i/p game the console has 55% less time per pixel to render any special effects, anti-aliasing, illumination, etc. than for a 720p game. Yes, even Resistance has fallen off the bandwagon and admitted they can't hit 1080i/p as previously claimed. (It also helps explain why Gran Turismo HD is so underwhelming.)

* 2.25x: that’s how many more pixels there are in 1920x1080 compared to 1280x720
* 55.5%: that’s how much less time you have to spend on each pixel when rendering 1920x1080 compared to 1280x720—the point being that at higher resolutions you have more pixels, but they necessarily can’t look as good
* 1.0x: that’s how much harder it is for a game engine to render a game in 1080p as compared to 1080i—the number of pixels is identical so the cost is identical
There is no such thing as a 1080p frame buffer. The frame buffer is 1080 pixels tall (and presumably 1920 wide) regardless of whether it is ultimately sent to the TV as an interlaced or as a progressive signal.
* 1280x720 with 4x AA will generally look better than 1920x1080 with no anti-aliasing (there are more total samples).
 
Wow eh Microsoft's PR's get there hands dirty lol alot of that is PR crap. I saw a nother article on IGN which had microsoft said the 360 downright owns PS3 and had graphs it was quite amusing how there tryin to battle. unfortunatly not much of it is working- espiecially after seeing the Gamerday videos. the PS3 looks rocking
 
Wow eh Microsoft's PR's get there hands dirty lol alot of that is PR crap. I saw a nother article on IGN which had microsoft said the 360 downright owns PS3 and had graphs it was quite amusing how there tryin to battle. unfortunatly not much of it is working- espiecially after seeing the Gamerday videos. the PS3 looks rocking

MS PR has nothing to do with this. Besides, there is nothing particularly wrong about what he said.
 
They shouldn't be none of the 1080P games look great IMO. It is a waste I would rather have more effects at 720p instead of going for a number. I hope that no games accept xbox live arcade games are 1080p for the 360.
 
* 2.25x: that’s how many more pixels there are in 1920x1080 compared to 1280x720

True. Interestingly 1280x1024 is about 40% more pixels than 720p which is quite a bump when going from a 16:9 to 5:4 aspect.

* 55.5%: that’s how much less time you have to spend on each pixel when rendering 1920x1080 compared to 1280x720—the point being that at higher resolutions you have more pixels, but they necessarily can’t look as good

True, but with the catch that you are GPU bound and, as I believe nAo pointed out, that due to the way GPUs work on quads that in some situations that the cost is not linear.

* 1.0x: that’s how much harder it is for a game engine to render a game in 1080p as compared to 1080i—the number of pixels is identical so the cost is identical

In general true as most are not expecting halfheight framebuffers.

* 1280x720 with 4x AA will generally look better than 1920x1080 with no anti-aliasing (there are more total samples).

Only more samples for edges, but in general:

It depends.

An RTS may be better at 1080p, as well as games not in demand for 60fps. On the other hand I would argue that there are cases where a lower resolution than 720p may even result in better image quality (something Carmack expressed angst at in 2005 and I agree with him).

Basically: Use the resolution that best fits the job. Personally, and this is just me, I would take 720p60 with 4xMSAA over 1080p with no MSAA, but some people really prefer the resolution (although at some point pixel density will be more than we can see (4kx4k? although I think they eye is more sensative to horizontal resolution or vice versa), at which point the only real concern will be per pixel quality). YMMV.
 
This forum has already had many a one indepth discussions about the advantages and drawbacks of 720p vs 1080p. The only thing different about this thread is that its flamebait.

Request for lock, good sirs.
 
Last edited by a moderator:
Ha! Caught a sneaky point in the chart... "Total Memory System Bandwidth" instead of "Total System Memory Bandwidth". In the former one could argue that the "memory system" in the 360 does indeed have 278GB/s; whereas "system memory bandwidth" would tend to indicate bandwidth available to the entire system, and while the eDRAM does alleviate system bandwidth usage from the main memory pool this is more misleading as the entire system cannot use it (although the entire arguement is apples-bananas anyhow because both Sony and MS want to direct consumers attention to worse/best case scenarios and have no interest or gain in accurately represent work flow and variables... pure PR, discussed here long ago).
 
this should be fun
No, this should be locked. Acert93 said everything that needs to be said in his first post, and the same arguments have been made by non-partisans in every thread on this topic so far. I don't believe any further technical discussion would yield any new results, so threads like this one can only serve as flamebait.

(Also, that chart is - without a lot of further explanations - intentionally misleading)
 
Status
Not open for further replies.
Back
Top