3D Artists/Experts help me spot the diferences :p (Uncharted render)

[Metalmurphy]

In the model it self, ignoring the obvious better lightning and other eye candy effects. Do you see any differences? Mainly polygon and texture wise?

A friend and i are having a "friendly" debate, he says that its clearly obvious both models are different, i however think that both models are exactly the same, even the textures, and only the lightning and some other post proc effects are different.

What do you guys think?


P.S.- Not sure if this topic should have been in this forum.

 

[tongue_of_colicab]

The left guys chin looks a bit wider, you got the ''line'' next to the nose wings wich is not as big as on the other. Also the right guy has a line in his cheek the other doesnt seem to have. The right guy also seem to have more wrinckles in his forehead. But those things might be because of the things you mentioned. I think they are different. The left one has more poly's and I think some subtile differenaces in the model along with other materials. Though I'm no expert.

 

[Paul_G]

The left image has much better eyes (don't think that, you know what I mean) The eye itself looks better, but the right models eyelid looks more angular

That said they are very close

 

[macabre]

I´m no expert either but I guess the right one is a normal mapped version of the left one with less poligons.

 

[deepbrown]

Better texture, better lighting, better eye specular. Huge amount of AA...but otherwise they are very close...even polygon count doesn't seem too far off. I expect the eyes to be better in the final game, since this is something that appears to come later in a game making process (eg. Heavenly Sword). Well it'll be very impressive no doubt, especially with all the ingame facial animations....just watch the trailer and just watch his face...his facial expressions change throughout

 

[luke76]

Better texture, better lighting, better eye specular. Huge amount of AA...but otherwise they are very close...even polygon count doesn't seem too far off. I expect the eyes to be better in the final game, since this is something that appears to come later in a game making process (eg. Heavenly Sword). Well it'll be very impressive no doubt, especially with all the ingame facial animations....just watch the trailer and just watch his face...his facial expressions change throughout

man i can't wait....but 1up when show up direct feed image seen in the last egm ?

 

[deepbrown]

 

[Cheezdoodles]

Do you see any differences? Mainly polygon and texture wise?.

Right picture is a normal-mapped version of the left picture. Left picture has higher polygon count, and well, pretty much better quality everything. Just take a look at the hair.

 

[ManuVlad3.0]

I'm not a techie, but the screen above is in the same level IMO and it is realtime.

 

[Metalmurphy]

Right picture is a normal-mapped version of the left picture. Left picture has higher polygon count, and well, pretty much better quality everything. Just take a look at the hair.

Tbh the hair seems the same apart from the lightning conditions. How can you tell its a normal-mapped version btw? I honestly dont see a difference, except maybe for the hears.

 

[Tap In]

from NeoGaf

 

[ManuVlad3.0]

The realtime model has hepatitis

 

[archangelmorph]

Poly count seems consistent..

The only visible [technical] differences I can make out are better shaders (skin, eye and hair shaders mainly.. better translucency, SSS etc..), possibly better texture resolution and a much more powerful lighting model (it's amazing how much lighting can affect what you're looking at)..

 

[Laa-Yosh]

Could we please stop comparing the polygon counts and such when one of the images is clearly not a 100% screenshot? Either it's been edited, resized and filtered otherwise, or it's a magazine scan, or it's been photographed from a TV/monitor - I can't really decide, but it's not a direct feed image for sure.

This also obviously effects image quality, a lot of the realtime rendering artifacts (lack of AA, low quality texture filtering, shading precision etc) are blured out.

I'd also suggest looking at this rendered image as the compression is obviously better and it shows a lot more details:

Some things making the realtime image inferior that you can still see:
- ingame version uses a more simple shader, simple additive speculars resulting in orangish skin highlights, instead of HDR reflection, making the surface look metallic
- lack of SSS (subtle, but important effect) look at the right ear of the rendered face
- problematic self-shadows in the insides of the nostrils and the ears


They've clearly gone to great lenghts to smooth out the silhouettes of the highest LOD face model for the hero character, so that you don't catch poly edges even in the close views. We've built 1100-1300 poly faces, those can get a bit edgy in closeups, adding about 500-800 polygons should do the trick for a realtime image when combined with normal mapping.
The actual difference between the poly counts is probably at least an order of magnitude, if not two. This can be spotted in the fidelity of the shading but we'd need a clearer image of the realtime version for that. Until we get such, the entire topic is quite useless...

 

[Metalmurphy]

No need to be so harsh

Its a screencap from a video, but i would think it really wouldn't matter. I was trying to compare the 3D modeling it self, not the lightning, not the shaders, not the shadowing, etc...

 

[archangelmorph]

The actual difference between the poly counts is probably at least an order of magnitude, if not two. This can be spotted in the fidelity of the shading but we'd need a clearer image of the realtime version for that. Until we get such, the entire topic is quite useless...

That doesn't make sense?

How can you say the polycount is higher just by looking at the shading fidelity?

Or aren't you aware that pixel shaders don't take into account the number of local vertices when processing an output and therefore you would only be able to see a visible difference in polycount by looking at the edges of the model and comparing the smoothness of the contours..

The fidelity of the shaders are likely to relate directly from inputs such as relevant [powerful] lighting models and much higher resolution textures and maps for the face, without considering any vertex information throughout the computation..

And looking at the contours of the face for both models i'd say the poly counts are very close if not identical which is rather impressive to say the least..

 

[_xxx_]

I'm with Laa-Yosh there on polycount, the left pic in the first post is at least an order of magnitude higher, the right one looks like a normal-mapped low(er)-poly version.

 

[archangelmorph]

I'm with Laa-Yosh there on polycount, the left pic in the first post is at least an order of magnitude higher, the right one looks like a normal-mapped low(er)-poly version.

How can you tell?

What aspects of the images are you using to draw such a conclusion?

I agree the high fidelity model could have a higher poly count but to say an order of magnitude higher begs at least some form of logical justification because as I stated before it's not immediately obvious either way..

 

[StefanS]

I agree the high fidelity model could have a higher poly count but to say an order of magnitude higher begs at least some form of logical justification because as I stated before it's not immediately obvious either way..

Just look at the ears. There's definitely quite a difference in poly count. Though, I wouldn't exactly stick a number to it.

 

[archangelmorph]

Just look at the ears. There's definitely quite a difference in poly count. Though, I wouldn't exactly stick a number to it.

I'm beginning to think I have a sight problem..

The only differences I see with the ears are the inclusion of SSS and better lighting illuminating them...

Judging by the contours I can't see any visual difference...

I think it's too difficult to tell whether there's any polygonal difference unless we had either a zoomed in profile of each or could see them rotate and examine properly..

Or even better, some wireframe shots!!