It's a bit of a shame that stereoscopic 3D consoles, like the Virtual Boy (or this "Video Boy" here), are no longer a thing outside of VR. And even VR is looking bleak with Meta recently closing their game studios due to huge losses.
I guess people don't want to wear things on their faces. And autostereoscopic screens, which don't require glasses, don't work so well for stationary TVs. The Nintendo 3DS was the only successful system with (auto)stereoscopic 3D so far. Unfortunately its first hardware iteration wasn't quite there yet, and the generally low resolution was an issue.
But I think it could easily have been an optional feature of the Nintendo Switch 2, if they had built in a movable lenticular lens array and a head tracking camera, just like in the "New Nintendo 3DS" (the original 3DS used a simpler but worse system).
To get to the original 1080p resolution, they would also have needed a screen with double the horizontal resolution (3840x1080 rather than 1920x1080), since autostereoscopic screens effectively halve the usable resolution. Games that don't support 3D would just show the same image for both eyes.
It could have been an optional feature even for games that support it, with a choice between a 30FPS 3D mode and a 60 FPS 2D mode, which comes down to the same amount of rendered pixels. An ML system similar to Nvidia's DLSS might even generate the 3D effect (left and right frame) from a single rendered frame by using the depth buffer. For a smaller performance cost.
But I guess the additional hardware cost doesn't justify a cool feature like that.
The thing is, the 3DS was a success mostly despite the 3D, rather than because of it. No one was excited about it or all that impressed - remember that initially the 3DS was such a flop at launch that Nintendo needed to pull a Hail Mary, cutting the price and launching the Ambassador Program. Or consider how successful the 2DS was.
The 3D on it is a very neat trick, but it's mostly a distraction and there's something "uncanny valley"-like and unpleasant about it. I almost always play with the 3D slider off; it's easy for me to see why Nintendo gave up on it.
> The second revision with the better “head tracking” screen (I don’t remember how it worked) was much better than the launch model.
It worked with a sheet of lenticular lenses on top of the screen, which separated the odd and even pixel columns for the left and the right eye. The sheet could also be moved a tiny amount by a motor, to change the "center" of the separated light, depending on whether the head was tracked more to the left or more to the right.
> But mostly, you’d just turn 3D off. It was a gimmick, cut the resolution in half, and doubled processing requirements.
Well, it did cut the resolution in half, but this meant the GPU cost stayed the same. They could instead double the horizontal resolution of the screen to keep the same effective resolution, in which case the GPU performance cost would indeed double. But as I said, in future systems this performance cost could likely be reduced with a form of frame generation by a machine learning model.
> remember that initially the 3DS was such a flop at launch that Nintendo needed to pull a Hail Mary
I agree people weren't very excited about it, but the lack of interest affected the PS Vita (direct competitor without stereoscopic 3D) even worse.
I'm pretty sure the fact that 3DS/PSVita were much less successful than DS/PSP was caused by the rise of smartphones and app stores, which led to a flood of highly successful cheap mobile games that didn't require buying a separate handheld console.
I would add, I've always felt the addition of the 3D screen also held it back because the touch screen became relegated to always being secondary (so that the main display of the game could be 3D). Many games that were sequels to touch-focused DS games did this, they had you touching the bottom screen to interact with something on the top screen and made those games feel a lot worse to me.
> The 3D on it is a very neat trick, but it's mostly a distraction and there's something "uncanny valley"-like and unpleasant about it.
I recently picked up a used New 3DS XL to give this a try for the first time, and I actually really like it. With the eye tracking, it's pretty good imo. Did you also try the later iterations of the hardware?
IIRC they required all games to work without 3D (because some people couldn't see it? or it gave them motion sickness?), so it could never be a core mechanic and thus doomed to be a "neat trick".
But that's a thing people have to put on their faces, which, as we have seen, most people really don't like. It also costs extra money, unlike a "3D mode", which would just be enabled in the game menu.
Yes but game developers can't rely on optional add-ons because most people won't have them. So there is little reason for the developers to support them. That's why such optional add-ons were rarely successful in the past.
I guess people don't want to wear things on their faces. And autostereoscopic screens, which don't require glasses, don't work so well for stationary TVs. The Nintendo 3DS was the only successful system with (auto)stereoscopic 3D so far. Unfortunately its first hardware iteration wasn't quite there yet, and the generally low resolution was an issue.
But I think it could easily have been an optional feature of the Nintendo Switch 2, if they had built in a movable lenticular lens array and a head tracking camera, just like in the "New Nintendo 3DS" (the original 3DS used a simpler but worse system).
To get to the original 1080p resolution, they would also have needed a screen with double the horizontal resolution (3840x1080 rather than 1920x1080), since autostereoscopic screens effectively halve the usable resolution. Games that don't support 3D would just show the same image for both eyes.
It could have been an optional feature even for games that support it, with a choice between a 30FPS 3D mode and a 60 FPS 2D mode, which comes down to the same amount of rendered pixels. An ML system similar to Nvidia's DLSS might even generate the 3D effect (left and right frame) from a single rendered frame by using the depth buffer. For a smaller performance cost.
But I guess the additional hardware cost doesn't justify a cool feature like that.