next »

[William Broom]

Or what if you encode two separate images, so the game must be played with one eye closed and the viewer may 'switch worlds' by switching which eye is open?  (How could the two worlds interact?)

This is what I would have done if I had been more ambitious. What I am doing (with michael) is still going to be cool though.

[Cymon]

It is also possible that the stereoscopic effect could be abused in interesting ways.  For example, what if the stereoscopic depth cues convey different or conflicting information from the other cues in the image?  Or what if you encode two separate images, so the game must be played with one eye closed and the viewer may 'switch worlds' by switching which eye is open?  (How could the two worlds interact?)

Both are exactly what I was thinking. Remember the old side scrolling jumping games? What if hidden walls could be seen with 3D glasses on because they were a different depth? Maybe even make the 3D glasses an option that you can put on whenever you wanted to avoid the whole headache thing. How would having 3D depth cues work if they were still providing information in games where depth cues are intentionally messed with like echocrome or Fez?

The two world thing could be like playing a game on two sides of a plane, either by flipping your character at certain points or simultaneously playing both sides like a puzzle. With both eyes open you'd occasionally get bad information thinking you were in danger when you were actually safe or visa-versa, but limiting your vision can help you succeed. The only problem, some of us can't wink both eyes as well.

[increpare]

This winking effect could be much more easily got by having a normal display and a 'change view' button on one's control-device...

[Zaphos]

This winking effect could be much more easily got by having a normal display and a 'change view' button on one's control-device...

True to some extent, thought not if the two worlds interact through some stereoscopic effect.

[Crackerblocks]

Can you thread a needle or swat a fly with one eye closed?  Stereoscopy as a depth cue gives different details than other depth cues, and it's very important for some tasks. Can any of those be made in to a game?

Excellent examples!

It is also possible that the stereoscopic effect could be abused in interesting ways.  For example, what if the stereoscopic depth cues convey different or conflicting information from the other cues in the image?

I have done some experimentation... I think your brain tries to resolve the world as a whole. So either you get a wonderful 3-D world, or you get a headache of red and blue pixels. If you give conflicting depth information, your brain gives up and says "this isn't a 3-D world, it's just a mess of color on a flat screen".

I'll keep experimenting. Just having trouble getting my right brain revved up.

Incidently, does anyone want some bare bones C++ D3D9 code for setting up stereoscopic rendering?

[mikeaka]

I have also found that you cannot change the depth to quickly otherwise the brain catches on and the images won't merge.

[bigpants]

My "research" has shown that you cannot have too much depth going on at once. Specifically, you can't have stuff way in the back and stuff way in the front AT THE SAME TIME. The Depth delta needs to be contained otherwise the user will get a headache from constantly switching his eyeball focus between background and foreground elements. With a contained range, the user can see all the background and foreground images at the same time (like real life).

This isn't the same as going cross-eyed. It's FINE to have a ton of stuff in the back. It's also FINE to have a ton of stuff in the front (causing the crossed eyes). However, it's headache inducing if you have both going on at the same time. You have to watch the range of the Z index that the red and blue delta produces.

Through trial and error, I've determined that the red and blue images should never be more than 60 pixels apart. Specifically, the delta of the read and blue of the most foreground image can't be more than 60 pixels width different than the delta of the red and blue of the most background image. Another way to put it, red and blue can only differ by 0 - 60 pixels for ALL images. The range of 60 pixels can take place entirely behind the monitor (-60 to 0), in front of the monitor (0 to 60), or a mixture of the two (some a bit behind -30, and some a bit in front +30). As long as the depth is contained to 60 pixels delta, you're good.

On a related note, my tests have showed that an infinite depth background / vanishing point works, but it means you can't have stuff coming out of the screen without causing a headache (same problem as above).

[muku]

Through trial and error, I've determined that the red and blue images should never be more than 60 pixels apart.

I'd expect this to be more a function of actual on-screen distance in centimeters than pixels.

[bigpants]

Muku, you are... A GENIUS!
My 60 pixel measurement was arrived at using a 17 Inch LCD monitor, 1280 x 1024 resolution. A lower resolution or a larger monitor would render my 60 pixels flaccid (I won't mention what projecting the game would do).

---

Looking at Kokoromi's Gamma post, they're already well aware of this:
"your game must include controls for decreasing and increasing the interpupilar distance (the distance of the red and blue images from the center point).  The controls should be bound to the angle bracket keys: < for decrease and > for increase."

Since I'm faking the 3D, I misunderstood this requirement when I read it originally. I thought they wanted to be able to bring the entire game playfield in front of or behind the projection screen. How wrong I was. They're trying to ensure the depth range perfected on the PC is respected by the projection (by making the depth range adjustable). Hence, less headaches.

In my case, I'm sticking with my 60 pixel rule of thumb, and getting the > and < to increase and decrease that amount.

More importantly...
how many people are calling their game interpupilar distance?

[mikeaka]

Anyone know how to use additive blending on your 3D objects but at the same time make them not see through to other 3D objects?

[muku]

Muku, you are... A GENIUS!

Ahem, *cough cough*, yes, yes I am. You are right.

[Crackerblocks]

Anyone know how to use additive blending on your 3D objects but at the same time make them not see through to other 3D objects?

Doesn't additive blending mean see-through by definition.

I'm confused by your confusion 

[Zaknafein]

What he means is that if you render stuff in additive blending, far-away objects blend with near object when they overlap.

What I'd do is use render targets. Render everything normally to two FBO/pbuffer/RenderTargets, and then blend then both together with additive blending.

Is that what everyone's doing?

[Cymon]

How did I miss this before:

http://www.neave.com/anaglyph/

[Crackerblocks]

What I'd do is use render targets. Render everything normally to two FBO/pbuffer/RenderTargets, and then blend then both together with additive blending.

Is that what everyone's doing?

That's what I'm doing. But using glColorMask or D3DRS_COLORWRITEENABLE would probably work just as well.

How did I miss this before:

http://www.neave.com/anaglyph/

very cool indeed. I wish they used something other than a blank white background though.

[Zaknafein]

But using glColorMask or D3DRS_COLORWRITEENABLE would probably work just as well.

That doesn't allow you to use fancy channel combining stuff like optimized anaglyphs. That's why I use a shader and 2 RTs.

[[SS] Olivil]

May sound wierd but is there an age limit? I assume it involves drinking thus making it 18+? I would freaking like to attend. 

[bigpants]

3 Red, No White, and Blue Things

• I started off using a white background (as opposed to black). I'm getting good 3D depth, and even managed some basic colouring. Does a black background provide a better experience? It turns out that due to the way I'm rendering, switching to a black background requires quite a few changes. I'll switch if people think it's the way to go. I'm very worried about the white background box taking away from the experience when projected.
  
  
• I'm using Hue-Saturation-Value value to set the colours (as opposed to RGB). I've discovered that the Saturation and Value need to be in sync for the red and blue. You can darken/lighten them (to better match the glasses), as long as the levels are the same for each. If they're out of sync, you get a shimmering effect as your brain resolves the lighter and darker colors to something inbetween. The end result is the image looks less solid. i.e. A dark blue combined with a light red will give a 3D effect, but the image won't look as solid as a medium blue combined with a medium red. My graphics are fairly basic, so this might not be applicable for you.
  
  
• My glasses are cardboard with foldable arms, so I can choose which colour covers which eye. I folded the arms so that Red covers the Left eye, and Blue covers the Right eye. Based on http://en.wikipedia.org/wiki/Anaglyph_image, and some people I've messaged, this is correct. Everyone needs to follow the same standard, because switching them will cause the background to become the foreground. What threw me off was the photos in this forum thread of people wearing the opposite configuration - turns out the camera snapshot reversed things. Phil > You need to add this standard to the Gamma3D contest pages.

[Alex May]

[fish]

it's red and BLUE, you nincompoop!

deadline is tomorrow by the way!