Is this Gen of Consoles just WEAK?

[Tachikoma]

d21lewis said:

I guess my ass could use a little schooling.  If it's not too much to ask, can you help me appreciate the leap? 

But of course my boy.

First we need to address the primary issue - When comparing the new generation to last generation, people automatically do so with the newest games of last generation, these games have benefitted from nearly a decades worth of development experience, SDK updates and shared development community expertise and workarounds, to get the clear picture of the technological leap, you have to strip back the consoles to their launch offerings, in which case the best we can do here is with comparing Resistance : Fall of man to Shadowfall, people tend to romantisize the past and remember thinkings looking better than they actually were, as is the case with retro games and people fast discovering that they looked a lot worse than they remember.

The best way to see this shift in development is to compare a launch title on PS3 with a much later game such as beyond two souls or the last of us.

Beyond the obvious differences in texture quality, mesh complexity, bump mapping and shadows have all taken a considerable step in improving, the biggest difference however is the ability to use multi layer textures, shaders and subsurface level shaders to produce multiple material lighting and coloring effects, but these effects show very little in static screenshots as shown above, the real meat of the meal comes from the game in motion.

Compare the lighting/shader effects used in Resistance: Fall of man here : https://www.youtube.com/watch?v=n79Efwn22w0   (select 720p), noting that the gun is primarilly flat textures with poor active lighting and a rough lightmap.

With the lighting/shader effects used on the gun in KZ:SF here : https://www.youtube.com/watch?v=tTrFMof-bDo The lightmap is significantly better and not only in resolution but in refracting colors from the environment, as well as the colored light from the weapons fire itself.

These same reflective and refractive properties are echos throughout the games texture collection, adjusted for each material type, a common occurance for games on all three of the newer systems.

It isn't that these features were impossible on the last gen systems, because they were, it just required so much power from the paultry hardware that it was rarely used properly and often faked using other effects for a less than satisfactory result.

So what we end up with is:

 

• Improvements across the board to lighting and lightmap resolution
• Improvements across the board to texture resolution thanks to higher memory capacity
• Extra GPU features and power allows for complex shader and subsurface calculations for lighting and texture
• Additional power allows for more objects to have tesselation applied, and for mipmap, bumpmap and displacement maps to be applied and with a high resolution
• Additional power allows for per pixel motion blue and accurate depth of field, rather than bluring the entire screen.
• Integrated solutions for anti aliasing result in better overall IQ with little impact to performance
• Expanded memory pool available to GPU allows for more expansive assortment of textures instead of tiling the same texture multiple times across a scene.

Textures, Shaders and raytracing for purposes beyond lighting.
One of the major advantages behind having the capacity to handle gpu compute and by extension, complex shader math is using engine elements beyond their original intended purpose, if we take for example this particular gears of war mission in which it's raining heavilly, the solution used for handling rain is simply to apply an animated displacement map to the flat textures of rain ripples, and a second animated texture of "dripping" rain to all vertical surfaces, while it works, roughly, it presents issues such as the animated textures play out on surfaces that should not be wet such as underneath sheltered areas, the particle effects volume that stops the raindrop effect does not effect the ground textures, because the textures used are hardcoded to have the water effect applied, it's only when you enter a solid structure that has textures designed to be always dry that the rain effect ceases.

With complex shader math you can apply calculations to emit rays from above at a set of variable direction downwards, the resulting lightmap it generates on the textures can then be used, as a "wet" map, wth predefined calculations of how to animate the drying process or wetting process, and to handle "dripping" on sharp edges of watness where raytracing stops abruptly, taking away much of the calculaton and design time and putting it into a dynamic shader.

It doesn't need to be just a raytracer used for input either, just like in the forrest map of killzone shadowfall, a particular part of the rocky area of the map has a cove where the torrent of water splashes up occasionally, the same particle effects emitter handling the splash effect, is actually used by the defered rendering engine to apple the wetness effect to the surrounding ground texture.

As we move further in to the new generation of systems, solutions such as these and other, new techniques will be used more extensively to build better and more living, breathing worlds which enhance both the believability and draw of these environments, these small details will play an extremely intricate part of making VR work better.

The xbox 360 and PS3 did occasionally use ore advanced textures too, but the power required to do so often meant that they were used extremely sparingly and in controlled environments where minimizing performance hits by reducing the complexity of onscreen detail (indoor scenes, for example) allowed for it.

On a more technical level, often times when developing games for the Xbox 360 and PS3, the most common multiples used for textures were between 256 and 1024 resolution, varying in combination based on the object shape or purpose, most titles tend to use 512x512, 256x512, 512x1024 or in some cases, 1024x1024, in a few cases, cutscenes and character creator screens would use up to 2048x2048 but drop back to lower resolution sets when playing actual game, it's worth noting that this doesnt mean that the path on the back of a player characters shirt is 1024x1024, it means that the entire torso and often legs/hands of the character are all fit within that resolution, with the head usually contained within another file, which would also contain the hair, teeth, mouth orafice texture and any additional accents.

With the bump in memory, most games being developed for xbox one and PS4 are now using on average at least twice the resolution as before, in many cases even higher, many of the textures in killzone shadowfall for example are beyond 2048x2048 ingame, with the added benefit of having textures of equal resolution applied through shaders - the lightmap resolution has also taken a big jump with this additional memory, allowing for much cleaner shadow resolution and much more intricate shadow representation on player characters and dynamic objects.

Often people argue that it's simply a case of high resolution and nothing more beyond that, it's a very stupid and narrow minded approach, because much of the changes aren't the resolution but instead how the materials are representated and how they are effected by static and dynamic light, as well as how the materials interact with the world around them, and therein lies the biggest jump graphically, it just doesn't translate well in the screenshots people use to compare because that is the very nature of dyamic materials.

Beyond textures and lighting effects, one of the other big changes is he ability to use progressive screenspace effects without salughtering performance, one of the best examples thus far this generation is with Alien Isolation in the subtle but accrate depth of field effect applied when using the motion tracker:

While creating a very gradual and soft transition between focus and out of focus, the effect also tweaks light calculation to generate realistic halo glimmer around bright distant objects, and an expanding effect on extremely bright light sources which would occur in the real world when adjusting your focus, the result is a believable shift in player focus when using the motion tracker, at which point your eye is drawn to the motion trackers screen and you are given reason to put it away to better see the path ahead, just as you would in real life.

Effects such as these are by no means new, but the accuracy and quality of the effects possible on the new hardware, with minimal impact to performance means they can be included in games to enhance the player experience and not just used as set pieces or cutscenes.

The overall fidelity of games has taken a notable step forward even over the last of the titles from the previous generation, and will continue to push forwards and improve as the generation continues.

So again, saying there's little to no difference this generation over the last, is just plain wrong and shows a complete lack of understanding, or comprehension at how the changes in shaders, textures, lighting and particle effects have progressed with the advent of more powerful hardware.