Motorstorm, F1, GTHD - AA and HDR

[Neb]

Nebula..just looking at single screenshots will not give you an answer: HDR rendering does not mean you're going to see shiny stuff everywhere, it just means you High Dynamic Range rendering..if your scene pictures a rainy day or simply doesn't not have (for some reason) a wide luminance range you won't be ablet to tell the difference with some LDR rendering, especially cause you can't see exposure changing over time from a static screenshot

Of course not thats why I have posted the link to the video and also watched it myself. Neverthless I never stated this game didn't have HDr but I said "F1 has what seems to be 4xAA and the lightning seems to be non HDr"! :smile:

Neverthless the lighting is a disappointment if it's HDr.

F1 video review.
http://www.gametrailers.com/downloadnew.php?id=17617&type=wmv&f=

 

[nAo]

Neverthless the lighting is a disappointment if it's HDr.
http://www.gametrailers.com/downloadnew.php?id=17617&type=wmv&f=

this is the same argument expressed with different words, even if lighting disappoints you it does not mean there is some correlation with their engine employing HDR rendering.

 

[Neb]

this is the same argument expressed with different words, even if lighting disappoints you it does not mean there is some correlation with their engine employing HDR rendering.

Haven't said that becouse the lightning disappoints me, it means it's not HDr. No I was watching the video and seeing how the lightning didn't affect the car and sorroundings as would be expected from HDr. And neverthless the lighting is still a disappointment for being HDr and I just noticed the banding on the sky to, low precision HDr?

But watch the video for yourself (previous post).

 

[Todd33]

As I said in my previous post, the no AA image has jaggies with hard edges, the 2xAA has the transition between the jaggies smoothed out but still not enough to hide them.

Which is the argument made in the previous pages with pics like this:

http://forum.beyond3d.com/showpost.php?p=982944&postcount=20

IMO 2x AA looks a lot like no AA if you are zooming in on edges without a reference 0x picture. Only 4x AA is obvious from any pic and any angle.

This is 2x AA, but I bet many people would say it's 0x if it were a blind test...

 

[nAo]

I just noticed the banding on the sky to, low precision HDr?

quantization due to video comrpession?

 

[deepbrown]

quantization due to video comrpession?

nAo...can it be outright said that F1 has AA and HDR...you are close with Cambridge SCE I'm sure...or is that under NDA's?

What do you think of Motorstorm, and GTHD...this technology interests me greatly.

 

[rekator]

quantization due to video comrpession?

Yes, it's the BIG problem with HD Games the quality of video on Internet…
Some have correct transcription of colours, brightness, contrast and visual quality but a lot are bad…

 

[Mintmaster]

Well done, shows what I have been saying so far, lack of AA. And no it's not shader aliasing!

By the way the old ss was only to show the aliasing as the others have blur/bloomish light to cover up and blur out jaggies!

Nebula, that definately is 2xAA. Blurs would not give you that kind of pattern.

My guess is the lack of AA on the truck roof is due to using the alpha channel for HDR as some sort of scale, allowing them to stick with the 8 bit per channel backbuffer. The AA resolve kind of screws up the AA, though, because this isn't a linear format. There's also a few (sometimes irrelevant) limitations with transparency.

It's a pretty decent hack, though, and should have been used more in PC games instead of leaving non-FP-blending cards in the dark. The Source engine probably does something similar.

8-bit and especially FP10 formats provide plenty of range and precision for good HDR. The former just occasionally has artifacts due to mathematical incorrectness when used with hardware built to treat the format as LDR.

 

[chris1515]

"True HDR" makes my cry

Yes I think many people don't understand HDR and many says that NAO32 is false HDR. Maybe people can understand with some explanation of HDR encoding by Greg Ward:

http://www.anyhere.com/gward/hdrenc/hdr_encodings.html

 

[Neb]

Which is the argument made in the previous pages with pics like this:

http://forum.beyond3d.com/showpost.php?p=982944&postcount=20

IMO 2x AA looks a lot like no AA if you are zooming in on edges without a reference 0x picture. Only 4x AA is obvious from any pic and any angle.

This is 2x AA, but I bet many people would say it's 0x if it were a blind test...

I can't see the picture becouse the link is broken. But if you say that many would think it has no AA when it in reality has 2xAA then it would mean the difference is minimal between no AA and 2xAA, yes?

Then the reverse would be true to, no AA could be misstaken for 2xAA and then the question why devs would use 2xAA if most would not be able do it apart from no use of AA!

No there is a difference between no AA nad 2xAA as seen on the Oblivion screenshot and zoom it in to see the difference (see the extra half translucent "steps"?) or see below. :smile:

First two ss are from FEAR and the last one from Hitman: Blood Money.

 

[Neb]

Nebula, that definately is 2xAA. Blurs would not give you that kind of pattern.

My guess is the lack of AA on the truck roof is due to using the alpha channel for HDR as some sort of scale, allowing them to stick with the 8 bit per channel backbuffer. The AA resolve kind of screws up the AA, though, because this isn't a linear format. There's also a few (sometimes irrelevant) limitations with transparency.

It's a pretty decent hack, though, and should have been used more in PC games instead of leaving non-FP-blending cards in the dark. The Source engine probably does something similar.

8-bit and especially FP10 formats provide plenty of range and precision for good HDR. The former just occasionally has artifacts due to mathematical incorrectness when used with hardware built to treat the format as LDR.

Yes the roof top lacks AA (probably due to what you explained) and some other parts look a bit strange but others looks to have 2xAA. But I will try the same track I tested before (canyon level) and then others to see if it changes AA depending on level and/or amount of cars becouse the aliasing on the canyon level was as visible as in games with no AA. :smile:

 

[Todd33]

Then the reverse would be true to, no AA could be misstaken for 2xAA and then the question why devs would use 2xAA if most would not be able do it apart from no use of AA!

That doesn't follow, people see "jaggies" and think there is no AA. I think devs use it for the same reason PC people do, it's almost free and sometimes 4x cost to many FPS. 2x cleans up the image mildly, but then people can still zoom pics and argue about aliasing and here we are.

No there is a difference between no AA nad 2xAA as seen on the Oblivion screenshot and zoom it in to see the difference (see the extra half translucent "steps"?) or see below. :smile:

Those same translucent steps appear in the screen shots Mmmkay posted too, that was our point. I'll try to re-post my image. If everyone agree that some edges are smoothed out even while not in motion, then some form of AA is at work and we are arguing for no reason.

 

[Neb]

That doesn't follow, people see "jaggies" and think there is no AA. I think devs use it for the same reason PC people do, it's almost free and sometimes 4x cost to many FPS. 2x cleans up the image mildly, but then people can still zoom pics and argue about aliasing and here we are.

It cost a bit to have 2xAA both perfomance wise and memory wise otherwise it would be enabled in all games. 2xAA isn't enough to hide jaggies (unless very high resolutions vs size of screen). Still the difference between 2xAA and no AA is notable, especially when in motion.

Those same translucent steps appear in the screen shots Mmmkay posted too, that was our point. I'll try to re-post my image. If everyone agree that some edges are smoothed out even while not in motion, then some form of AA is at work and we are arguing for no reason

See my post above yours, dskneo's comment made me think about it (I only played the canyon track). :smile:

it appears to be track dependent... some tracks its really obvious/annoying, others its really good.
My copy is European.

anyway, the game was made to look good in movement.

 

[Nick Laslett]

Although these technical debates have some interest, the focusing on a few graphical elements of a game would appear to completely miss the point.

There is a similar thread here at B3D where Gears of War's graphical failings are torn to shreds.

From my perspective, Gears looks amazing, but even more important as a whole package including framerate, animation, textures, art direction, etc the game is excellent.

It is the same with Motorstorm. This game looks and plays very well with good art direction, framerate, textures etc. I could care less how a specific effect was achieved or whether it is even present.

When you have games like Spiderman 3 polluting nextgen machines it seems especially sad to be nitpicking stuff like Motorstorm. Spidey3 has poor framerate, poor art direction, low poly count, dodgy animation and yet they expect us to pay the same price for this as Motorstorm.

I know these are very different types of games, but even so can't we put more effort into damning those games and then praising the others that deserve it?

 

[deepbrown]

Although these technical debates have some interest, the focusing on a few graphical elements of a game would appear to completely miss the point.

There is a similar thread here at B3D where Gears of War's graphical failings are torn to shreds.

From my perspective, Gears looks amazing, but even more important as a whole package including framerate, animation, textures, art direction, etc the game is excellent.

It is the same with Motorstorm. This game looks and plays very well with good art direction, framerate, textures etc. I could care less how a specific effect was achieved or whether it is even present.

When you have games like Spiderman 3 polluting nextgen machines it seems especially sad to be nitpicking stuff like Motorstorm. Spidey3 has poor framerate, poor art direction, low poly count, dodgy animation and yet they expect us to pay the same price for this as Motorstorm.

I know these are very different types of games, but even so can't we put more effort into damning those games and then praising the others that deserve it?

ABsolutely agreed....if people can't tell either way, ie. that it looks so good they think it has AA (i think it does have 2xAA) then the game's art has won.

But still...this post was more interested in how far these first party devs have got into the ps3, and what's possible on it.

 

[Todd33]

This is the pic I posted from Oblivion yesterday. The point was when you zoom in on a 2x AA pics many will think it has no AA. Of course if you have a 2x and 0x side by side it's pretty easy tom tell the differences, too bad we don't have that for Motorstorm.

 

[Obakan]

I think the game uses 2xaa on some maps and no aa on other maps.

 

[kyleb]

http://img87.imageshack.us/img87/7153/oblivot5.jpg

This is the pic I posted from Oblivion yesterday. The point was when you zoom in on a 2x AA pics many will think it has no AA. Of course if you have a 2x and 0x side by side it's pretty easy tom tell the differences, too bad we don't have that for Motorstorm.

It really doesn't take any comparison to recognize the use of AA in that shot. Without AA there would be no blending between those gray pillars and the blue sky behind them. Where as higher levels of AA can provide more subtle transitions along long edges, the shot obviously shows 2xAA as there is only one stage of blending between what is in the foreground and the background.

The confusion around Motorstorm is apparently because while it has 2xAA on many edges, it lacks any AA at all on plenty of others. People like myself who consider completely non-AAed jaggies are quick to notice them in Motorstorm and proclaim the game has no AA. Hearing others talk of the game using AA, I figured they were just being tricked by some edge selective blurring and motion blur. However, after taking a good look at the game it is obvious that there is a 2x multisampling going on, but what looks like something with the lighting canceling that on many edges.

I think the game uses 2xaa on some maps and no aa on other maps.

So what maps do you think use no AA?

 

[Todd33]

It really doesn't take any comparison to recognize the use of AA in that shot. Without AA there would be no blending between those gray pillars and the blue sky behind them. Where as higher levels of AA can provide more subtle transitions along long edges, the shot obviously shows 2xAA as there is only one stage of blending between what is in the foreground and the background.

The confusion around Motorstorm is apparently because while it has 2xAA on many edges, it lacks any AA at all on plenty of others. People like myself who consider completely non-AAed jaggies are quick to notice them in Motorstorm and proclaim the game has no AA. Hearing others talk of the game using AA, I figured they were just being tricked by some edge selective blurring and motion blur. However, after taking a good look at the game it is obvious that there is a 2x multisampling going on, but what looks like something with the lighting canceling that on many edges.


So what maps do you think use no AA?

To the trained eye it's obvious, I was speaking from a off the street gamer POV. I agree with your analysis, the AA if used is somehow selective or adaptive. The game looks great regardless of this microscope analysis.