Additional detail

[Mystify]

One comment that stuck out to me from all the videos was that you noticed a lack of details more in VR. For instance, you notice that the wall is just flat and textured, instead of being modeled.
So if we assume that we need more detail, at least close up, what types of techniques do you think would prove useful? Displacement mapping could give the appropriate detail for a wall, for instance, though it is expensive. Maybe we need techniques designed for 3d? for instance, a bump map that alters how a surface is rendered in 3d to give it the illusion of more depth? Or a more sophisticated form of parallax mapping? Techniques that don't add too much of an extra load on the content producers would be great.
I'm not really a graphics guy(at least not to this level of sophistication) so my understanding of such techniques is limited.

[geekmaster]

Actually, displacement mapping can be very cheap, especially when pipelined directly into the assembly-optimized blitter kernel. It worked well back when I implemented it on a 386 with no GPU (in about 1991). My graphics library was used in automated vending kiosks using (DOS based) textured graphics and sculpted touchscreen buttons, located in shopping malls scattered around the USA for many years.

Just because something may not be easy for a GPU does not mean it is beyond a software-only implementation. Of course, mixing software and GPU techniques may be difficult.

EDIT: I just realized that all the computers at work (and in the kiosks) were only 286 CPU systems. I had a 386 system at home, but not at work. At least I had 21-inch monitors at both locations, back when 13-inch was the norm. 640x480 256-color VGA was state-of-the-art then, and the Paradise card was about $300. Perhaps driving 1280x800 would need a faster CPU, like a 486. Which reminds me, Castle Wolfenstein was brand new, and I replaced the pictures of Der Fuhrer on all the walls with pictures of my boss. That was popular at the office.

[Laserschwert]

I think parallex mapping could already do wonders here (and is rather cheap for the GPU). But the next generation of graphics cards will probably support hardware based displacement, so I guess it'll just get better in time.

[Libertine]

Bump and displacement mapping work great in 3D. Texturing is still king of the hill imo though, especially when combined with specular and reflection maps.

Heres a video i made of Metro 2033, which had all of the above. Lighting helps too. Theres another one on my channel in 3D if interested.
[youtube]http://www.youtube.com/watch?v=BYOMk6VSH2k[/youtube]

This lighting and bloom is exactly what The Hobbit was missing IMO, that and mist.

[snorelab]

One comment that stuck out to me from all the videos was that you noticed a lack of details more in VR...So if we assume that we need more detail...

That's probably a great assumption for most games.

The flip side to this, though, is that you could go with less detail. Since the human eye is always going to be drawn to the bit of detail that isn't right it might make for a very compelling experience to do away with complex textures. I think indie games that do this may be very compelling simply because you don't have as many visual cues constantly telling your brain that you are in a computer generated world.

[Pingles]

Actually, I enjoyed Anandtech's first discussion of the Oculus Rift on their CES podcast.

While they were certainly impressed they also discussed some concerns.

One was the level of detail. They mentioned that it is distractingly blurry when viewing objects from such a close view.
Another concern was the placement of the view, remarking that in most current FPS the camera is really in the chest of the player.

They noted that both of those concerns had nothing to do with the Rift itself but were concerns for porting over older games to the Rift.

[Mystify]

One comment that stuck out to me from all the videos was that you noticed a lack of details more in VR...So if we assume that we need more detail...

That's probably a great assumption for most games.

I think even in most games that they take a lot of shortcuts with detail. I'm not just idly theorizing about it, I'm looking at the guys who tried VR and said "Things need more detail"

The flip side to this, though, is that you could go with less detail. Since the human eye is always going to be drawn to the bit of detail that isn't right it might make for a very compelling experience to do away with complex textures. I think indie games that do this may be very compelling simply because you don't have as many visual cues constantly telling your brain that you are in a computer generated world.

I won't argue with this.

[snorelab]

I'm not just idly theorizing about it

Leave that to me.

[2EyeGuy]

640x480 256-color VGA was state-of-the-art then,

There's no such thing (at least I think there isn't, you might have to check with Michael Abrash, you can do some higher unintended resolutions with ModeX hacking).

VGA could only do 320x200 256-colour, except a few ModeX games that did 320x240 256 colours. Or it could do 640x480 16 colour, but normally only Windows used that, it was rare in games.

It's funny when everyone keeps calling 640x480 "VGA". For almost everybody who used VGA, VGA meant 320x200. EGA and CGA also meant 320x200, it was THE resolution for games for a long time, even though EGA could do 640x350 in 16 colour, and CGA could do 640x200 in 2 colour or in text mode.

Super-VGA introduced higher resolutions and more colours, but it wasn't used much in games.

[geekmaster]

640x480 256-color VGA was state-of-the-art then,

There's no such thing (at least I think there isn't, you might have to check with Michael Abrash, you can do some higher unintended resolutions with ModeX hacking).

VGA could only do 320x200 256-colour, except a few ModeX games that did 320x240 256 colours. Or it could do 640x480 16 colour, but normally only Windows used that, it was rare in games.

It's funny when everyone keeps calling 640x480 "VGA". For almost everybody who used VGA, VGA meant 320x200. EGA and CGA also meant 320x200, it was THE resolution for games for a long time, even though EGA could do 640x350 in 16 colour, and CGA could do 640x200 in 2 colour or in text mode.

Super-VGA introduced higher resolutions and more colours, but it wasn't used much in games.

My copy of Michael Abrash's "Power Graphics Programming" book was dog-eared from constant use as a primary reference book. He popularized Mode-X, which was 360x480 256-colors, and was the maximum that the original IBM VGA cards could do back then. But other vendors supported higher resolutions. Of course, video game programmers needed to support the lowest-common-denominator, which was the IBM VGA card with its maximum 360x480 256-color Mode-X, or more commonly just the standard 320x240 256-colors.

As I recall, 320x240 (and 360x480) were just video RAM limitations of the original IBM VGA cards, but other VGA cards we used could do 640x480 256 colors, and that is what we used. Some cards required plugging additional memory chips into the unused sockets to use those extended resolution modes, however. And 640x480 was called "VGA", not "Super VGA" as you suggest. Super VGA (SVGA) was 800x600. As I recall, 320x240 just duplicated pixels horizontally and vertically, but the electrical signal at the VGA connector was essentially 640x480.

I added Mode-X support for IBM VGA cards (and other low-end cards) to the graphics library I wrote back then, but even anti-aliased tiny text was too limited for our needs on that lower resolution, so for our main application we required a newer video card that supported 640x480 256-colors that our app required.

Michael Abrash was my hero back then, and I loved his later GPBB ("Graphics Programming Black Book", now available as a free download):
http://nondot.org/~sabre/Mirrored/Graph ... BlackBook/
I have a number of his books in my collection, but I think I read his "Zen of Code Optimization" the most (especially the part on Pentium pipelines). It is good to hear that he is interested in the Oculus Rift:

EDIT: I have been laying awake thinking about this. That happens whenever my facts are challenged, requiring more research before I can trust my memory again. My chain of thought on this subject is as follows: We know that there WERE 640x480 256-color cards at some point in time, and that they preceeded 800x600 256-color SVGA video cards. The question is WHEN? I worked on the stuff back when Window 3 came out (followed by Windows for Workgroups), but our app predated that and was DOS-based. Amongst the various VGA card chipsets I supported were the Tseng ET4000 and the STB chipsets which WERE available at that time, and both of those chipsets supported graphics mode 0x2E, which is listed as 640x480 256-color. I am certain we used that video mode for our application. So perhaps my memory was correct after all. Here is a list of video modes (including mode 0x2E):
http://www.columbia.edu/~em36/wpdos/videomodes.txt

So it looks like 640x480 256-color was indeed supported by many cards back then just as I remember, as according to this chart, 640x480 resolution was called "VGA":
http://en.wikipedia.org/wiki/Display_resolution.

Of course, for extended modes, each graphics chipset needed to be initialized differently, so my code had to probe the card to determine how to initialize it. Luckily, that information was supplied for popular graphics chipsets in my dog-eared copy of Michael Abrash's book.