AMD have said that photorealism will be achieved in 5 years.

I'm also critical of the word photorealism. Because we don't play still photos. And one major uncanny valley problem with games is the animation.
Animation has little to do with photorealistic graphics. It's possible to have a photorealistic CG character whose expressions may be grotesque. I also don't think that the AMD guy said anything about animation.
 
Last edited by a moderator:
Animation has little to do with photorealistic graphics. It's possible to have a photorealistic CG character whose expressions may be grotesque. I also don't think that the AMD guy said anything about animation.

What I meant is if the goal is to create a scene that mimics reality, all aspects need to match up. A horse can't be pivoting on a center axis instead of turning and walking in a realistic manner. Pretty screenshots of extremely modded-up Skyrim come to mind. The actual game breaks any feeling of realism very quickly.
 
Animation has little to do with photorealistic graphics. It's possible to have a photorealistic CG character whose expressions may be grotesque. I also don't think that the AMD guy said anything about animation.

In that case AMD's statement is pretty pointless. I don't care about photo realism in games if the illusion breaks as soon as I touch the controller. It's also why I always find those "look at how awesome my modded Skyrim looks in screen shots" topics rather funny. The game looks downright awful in motion. Compared to a Dragon fight in Dragon's Dogma, a Dragon attack in Skyrim looks utterly laughable.
 
still won't look as real life. why. imperfections. real life isn't perfect. i dont think we will ever reach the point of graphics looking like real life. i will give you reason #2. REAL LIFE ISN'T MADE OUT OF PIXELS. thats why.
 
i dont think we will ever reach the point of graphics looking like real life.
We absolutely will, at least in some situations. We're getting very close as it is. Anything that doesn't look quite real just needs some DOF, chromatic aberration, and motion blur to hide all the perfections it hey presto, just like a photograph on a cheap, kid's plastic-lensed camera. :yep2:
 
Well...

Apollo11Comp_575px.jpg
 
We absolutely will, at least in some situations. We're getting very close as it is. Anything that doesn't look quite real just needs some DOF, chromatic aberration, and motion blur to hide all the perfections it hey presto, just like a photograph on a cheap, kid's plastic-lensed camera. :yep2:
What there needs to be is some sort of turing test for photorealism. Fool enough people who can't tell whether it's real or rendered and you can claim photorealism achieved.
 
What there needs to be is some sort of turing test for photorealism. Fool enough people who can't tell whether it's real or rendered and you can claim photorealism achieved.

Photorealism for a camera or a window ? Making something look like a camera took it is one thing, mimicking a window on to the would-be world outside would be another.
 
Photorealism for a camera or a window ?
Camera, as the names derived from photography. I suppose the human-eye view would be considered 'realism'. That's something not seen on a TV though, who's images are traditionally via a camera (the only exception being computer generated images), hence the target in realtime graphics to recreate on a TV what a (cinema)photograph captures.
 
what you see is basically an image ray tracing renderation. which means that the computer copies the picture by rendering it. most close-to photorealism nowdays is basically scanning real life and generating a picture. well. whats the fucking point if you're taking pictures. real life isn't made out of pixels. there are trillions of imperfections. millions of complex physics that humans probably don't even know how they work. so even if we achieve photorealism. you bet your ass you can tell it a part from a real picture. imperfections.
 
What I meant is if the goal is to create a scene that mimics reality, all aspects need to match up. A horse can't be pivoting on a center axis instead of turning and walking in a realistic manner. Pretty screenshots of extremely modded-up Skyrim come to mind. The actual game breaks any feeling of realism very quickly.
I agree.

In that case AMD's statement is pretty pointless. I don't care about photo realism in games if the illusion breaks as soon as I touch the controller. It's also why I always find those "look at how awesome my modded Skyrim looks in screen shots" topics rather funny. The game looks downright awful in motion. Compared to a Dragon fight in Dragon's Dogma, a Dragon attack in Skyrim looks utterly laughable.
While I concur with your statement, I think photorealism will push developers to have convincing animations, or the like. I can see them realising the uncanny valley.
 
Realtime graphics are not even remotely close to realism right now so I'm doubtful. Not even CG is close.

I'm also critical of the word photorealism. Because we don't play still photos. And one major uncanny valley problem with games is the animation.

This. I think most people misuse that word, AMD included.

Do we really want all games looking like Lego games with almost everything heavily submerged into Depth of field?
 
I wonder if games will make us feel some emotions as movies where you can see real people, like About a Boy or Twilight and so on and so forth.

http://www.pcr-online.biz/news/read/photorealistic-games-just-five-years-away/034735
]

Hmm videogames have genres just like films and novels and graphic novels and vocal...novels.

Movies do implement...some of them..clever writing to broadcast and even tug at the emotions of the viewer given that films average an hour and a half to two hours to set up the three different acts.

Videogames are far too interactive in comparison, plus you are given choices or decisions unless the game is heavily scripted to me a single linear story.

The better question would be is why haven't you as a videogame player felt emotion from playing videogame s?

Heavy Rain falls under that category as do a few others that don't degenerate into the usual "bromance emotions" of "not gonna leave you behind bro...nooooooo!!!!" Where most action shooters fall...or the passive novels type emotions of RPGs.

To me that reads like another claim after a bunch of previous claims of "marketing photorealistic/ism graphics" but never explaining why when given the chance to make hardware, they rushed things (any company)

I harken back to what Tim Sweeney said, he said 5000 teraflops is needed to get a good enough approximation of reality.

I dont really know if this is realistic at all or if it has any real meaning, it just seems a suitably faraway post to aim at.

If you quad crossfired/SLI or whatever, 4 of the top PC cards currently, I guess you'd have about 22 teraflops (R 9 290x=5.6 TF). There dont seem to be massive advances currently though, stuck on 28nm.

If current consoles are 1.3 and 1.8 TF, maybe we will see 10-20 TF next gen. Even 10 seems like a lot currently, no single card can achieve.

I think about the technology presentation of the Sony PlayStation 3's E3 2005 where Ken Kutaragi and nVidia's CEO were speaking of engineering visions that were misinterpreted as statements.

Back then there was a comparison of the 2TF theoretical performance of PS3 (CellBE+G70 based RSX at 90nm) versus the required theoretical power of 10TF used by contemporary workstations.

However in hindsight if we were to pre-plan last gen consoles to have been launched later with 2007 or 2008 technology and engineering nodes then the theoretical vision may have had to be revised because say:

a 65nm CellBE or Xenon would have packed more transistors and higher clocks.

At 55nm, a G92 or RV670/RV770 GPU would have more transistors and higher clock speeds yielding over 2.5 ttimes the raw rendering performance not to mention evolved architectures.

Higher ram ammounts and overall throughput and bandwidth.

That would have placed them at closer to 4TF theoretical performance.

But currently there were different decisions at work that made what we have now.

This tech is meh. There's absolutely no non-diffuse information stored about the environment. No chance for the surface to interact with dynamic lighting conditions, no translucency. Mixing some renders with video footage looks impressive but overall it's going to looks even more past gen than other stuff from Farm 51. I love those guys but it's all smoke and mirrors.

The E3 2005 presentation used that "nVidia Luna Technology demo" where a lot of that tech you mentioned is shown.

I ran that tech demo in a PC with an nVidia 6800GT along with the "Mad Mod Mike tech demo".

Problem is that those were as stated, "simple tech demos". Videogames however which feature Artificial intelligence, collision detection, polygon counts, etc and etc effects and then resolution and frame rate. ..(not to mention required sales) are just much more complex.

Hence the "photo realism" really depends on marketing and viewer perception. The latter is a huge problem for me because I don't want to argue how the average gamer kept telling me back in 2009 that CoD was "like real life"

Videogames do kinda require smoke and mirrors effects much like movies. I suspect that even if a real console were to exist with a GPU capable of 12TF, that even at 1080p smoke and mirrors would still be used.

still won't look as real life. why. imperfections. real life isn't perfect. i dont think we will ever reach the point of graphics looking like real life. i will give you reason #2. REAL LIFE ISN'T MADE OUT OF PIXELS. thats why.

In science that would be a debate...obviously life isn't pixels but there is a yt video on the resolution of the human eye...but that's a science discussion.

This. I think most people misuse that word, AMD included.

Do we really want all games looking like Lego games with almost everything heavily submerged into Depth of field?

Indeed, as I implied about the "smoke and mirrors" what games, even dramatic videogames would have is "hyper realism" as even movies are similar because reality would be too unexciting.

Increased costs for assets, animation, engine license ecc

Last console generation has been a massacre

Not to mention how only certain genres dominated sales last gen and how technically acclaimed games that were highly recommended were not given the coverage (by gaming sites) and sales and instead faced a lot of dramatic opinions or dislike by gamers who imho were disconnected by being too much into certain genres.

The nVidia GTX 980 technology demo is a very simple one that although animated, has a main objective on lighting by global light source and it's ray casting and bouncing effects as.well as shadows.

We are way past volumetric smoke, lighting, shadows which last gen could bring consoles to limits fast, I suspect current gen is going up against that wall but through clever use of "smoke and mirrors" the graphics will get close to the general interpretation of "photo realism" depending on the genre.

There is a bit of a problem though because of how most games imho last gen became kind of apologetic or hand holding.

Few games were set in total darkness with using a light source and many games had night settings where you could clearly see everything but shadows. I suspect it's a side effect of games being too easy for fear that gamers will fear the learning curve and not buy...or blame the game...not like Demons Souls was an instant hit, much less Dark Souls but it's part of a pattern going on because "photo realism" is greatly dependent on lighting and shadows and even smoke and mirrors literally specially as pointed out in the nVidia tech demo where light source gets discussed.
 
BTW any news about this game ?

https://www.youtube.com/watch?v=1vXYUYC3WsM
https://www.youtube.com/watch?v=nZ1-BLhYocg

looked promising in the photorealistic backgrounds department

Thing is, when you view this, your brain is getting overwhelmed by the high frequency detail to miss the underlying simplicity of the geometry (and the total lack of any variation in lighting - eg, no specular, reflections, etc). It's sortof the opposite of typical rendering - the small details are all really good (being photo reprojection) but the whole is an inconsistent mess. It helps enormously they are showing a dusty room capture (little specular) - the rundown look helps hide the inaccuracy in their 3D capture too (all the flat surfaces appear to be quite blobby when you look close).

It looks real because a lot of those shots are probably very close to the sampling points, so what you're seeing is basically a reprojected photo.

To completely break the illusion, just keep your eyes focused on the windows in the first video. Once you look at it that way, it'll look really awful. :mrgreen:
Notice how they massively cranked up the post effects in the second video? That helps hide these problems (when in doubt, blur it out!). The windows are bloomed out, for example. They are still demonstrating it with a smoothly moving cinematic camera too.
 
Back
Top