Image Quality and Framebuffer Speculations for WIP/alpha/beta/E3 games *Read the first post*

One of the guys here at Beyond 3D did some pixel counting of some of the released screens of BR when it first hit the arcades. I think it was in the Tekken 6 thread, though I can't remember for sure.

 

http://www.facebook.com/topic.php?uid=85571972977&topic=12513

 

I referred to that quote from the DF as possible poor wording, but I've explained that already. Care to note any differences in those screens? They seem identical to me sans the gamma output difference.

I'm curious, seriously. It can't be in-motion difference since the texture detail is blurred of course with the full motion blur.

 

Yes I do notice a difference, the textures look a little bit more blurry on the PS3, but this should not be a discussion about what I notice or not because I was no talking about that from the start, just taking the DF article as reference, if you are telling that the PS3 has equal texture filtering than the 360 version that’s your opinion.

 

The thread here is about facts and facts alone, which is the basis of your point, no? You're saying you're noticing a difference, I'm not and kindly asked if you can pinpoint any differences that I'm simply not seeing. Opinions have no bearing here.

If anyone else can note any differences I'm missing (as subtle as they are), please go ahead.

 

Looking at the pattern on the pants, the 360 with MB On has a sharper detail (more grain to the fabric), but that's if you look close enough.

 

Ok no prop, for me the difference is more noticeable in Heihachi pants; there you can compare the little details in the fabric. The diference is not much but if you change from one picture to the other I think you can notice.

 

I see it now in the finer detailed area, notably the waist because the gamma difference seems to affect other areas. Subtle enough, but it's there.

Though I'm not sure if anyone has brought up the difference in the model itself. Look at the right arm in all comparison shots. I thought that it was a stance difference initially but I suppose not.

 

You got me. Maybe he just works out the one deltoid more in the 360 universe.

 

TV/Console-related scaling discussion spun off here: http://forum.beyond3d.com/showthread.php?t=55386

 

A little update. You may be right since I'm hearing that the PS3 version is "better than arcade" from some fans who got their copies early. What little slowdown the arcade version had is nonexistent now.

So instead of optimizing their code after nearly 2 years, they simply chose to lower the resolution to offset the load on the GPU and implemented AA.

Given the DF's comparison, I think they made the right choice.

 

new ratchet demo is up, any word on the resolution? Seems like Tiny Black bar is gone. The game looks blurry to me, not sure if this is the side effect after playing Uncharted2.....

 

Heh, I remember a similar developer interview back in 2000 where they went to great lengths explaining how TTT sampled character-rendering at 120hz to generate motion-blur on characters during certain moves.

Unless we're privy to implementation details of Tekken 6 Motion blur, that's kinda hard to say.

 

The new Ratchet & Clank demo looks soft. A shame this game can't run at 1080p with 4x msaa so it'd look near Pixar quality like the bullshots.

 

I believe Quaz51 said something about it being 960x704.

http://forum.beyond3d.com/showpost.php?p=1333850&postcount=841

 

I'd actually be really interested if someone could further explain what he meant by the proprietary merge it's doing. I have no idea what he means by that and how it differs from other games that upscale.

 

is the R&CF2 still using the same method? I thought the tiny black bar from the top and bottom is gone.

 

Because the framebuffers is 1280x720p so it isn't exactly an upscale & it's definitely better to simply upscale . From what Quaz51 said, insomniac use this method to have the best edge of the 2xMSAA and probably to guarantee steady 60 fps, I guess.

 

The proprietary merge with the sub-samples has to do with the way RSX handles the information in a temporary buffer. This is why you might see 2560x720 as a rendertarget in certain presentations.

edit:essentially, they render to say 960x704 2x MSAA rendertarget and then expand that to 1920x704 and do the lighting calculations and then resolve back to 720p or some such... there are other steps there with transparencies perhaps.

 

Just want to know if this is 4xAA?



Seems too clean.