Bringing stereoscopy closer to the screen

[Dr_Invisibrow]

I just started messing around with Dolphin's stereoscopic features and overall I'm extremely impressed. I'm using side-by-side on a 3D projector.

Some games are breathtaking right off the bat like Punch Out and Xenoblade. But in others -- like Mario Galaxy -- I feel the action occurs too far away from the camera, preventing anything from popping out. Increasing depth pushes everything very far back and increasing convergence doesn't seem to do anything at all.

Is there a way to adjust the point from the game's camera at which objects start coming out of the screen? I guess that would be the point between depth and convergence?

[wowfunhappy]

I just started messing around with Dolphin's stereoscopic features and overall I'm extremely impressed. I'm using side-by-side on a 3D projector.

Some games are breathtaking right off the bat like Punch Out and Xenoblade. But in others -- like Mario Galaxy -- I feel the action occurs too far away from the camera, preventing anything from popping out. Increasing depth pushes everything very far back and increasing convergence doesn't seem to do anything at all.

Is there a way to adjust the point from the game's camera at which objects start coming out of the screen? I guess that would be the point between depth and convergence?

Yeah, this is called convergence, ie the point where the game's 3D space intersects with your actual screen. You want to increase the convergence to bring stuff closer.

[Dr_Invisibrow]

Yeah... that's what convergence was supposed to mean...! And it does seem to function that way in the aforementioned games that look great. But why does the convergence slider not seem to do anything at all in some games like Mario Galaxy?

[mbc07]

Some games needs a high convergence value before it becomes noticeable. In other words, right click the game in Dolphin's game list and increase "Convergence Minimum" value until you find a value that properly reacts when you move the Convergence Slider from Graphics Settings...

[Dr_Invisibrow]

Thanks for the reply Jhonn!

Can anyone verify that this works in Mario Galaxy? I tried changing the convergence value in the game's properties, I've tried adding a [Video_Stereoscopy] and StereoConvergenceMinimum value in the config, and everything else I could think of. These settings work fine in other games. No matter what I set these values to, the 3D in this game is always set far into the screen with nothing popping out.

I'm on 4.0-8721.

[Dr_Invisibrow]

Hey, so I decided to try this stuff out in the Ishiiruka build. Everything works dandy there. Looks like there's a bit of a problem with the main dolphin build + stereoscopy.

[Armada]

To get stuff close to the screen you should decrease the convergence distance, not increase it. If you need to move stuff even closer, you need to input a negative convergence distance in the game properties.

Also StereoConvergenceMinimum has been removed, it has now been replaced by StereoConvergence.

[masterotaku]

To get stuff close to the screen you should decrease the convergence distance, not increase it. If you need to move stuff even closer, you need to input a negative convergence distance in the game properties.

Also StereoConvergenceMinimum has been removed, it has now been replaced by StereoConvergence.

What? Unless convergence works the opposite way in non 3D Vision stereoscopic modes, increasing convergence makes the screen depth point be further into the scene, making things look nearer to you (and pop out if they move out of the screen in your direction). At 0 convergence, everything is flat and at infinite depth, while at a very high convergence you leave the game behind you (and you see either the horizon skybox or nothing).

There are games like Metroid Prime where just 1 unit of convergence in Dolphin is too high, while zero is flat with the scene pushed at full depth. Would decimal values be a future possibility? As an opposite example, other games have very little range of convergence ingame with the hotkeys. This is what OP is complaining about.

This leads me to another question: what makes convergence need so different values for the same visual convergence? This applies to PC games too with 3D Vision. The only idea I have is that models may be at different scales, like in X game the character is 100 meters tall while in Y game the character is 1 meter tall (in videogame making distance measurements), but for us it looks the same because everything is scaled, and thus convergence needs to be scaled in the same proportion.

[Armada]

What? Unless convergence works the opposite way in non 3D Vision stereoscopic modes, increasing convergence makes the screen depth point be further into the scene, making things look nearer to you (and pop out if they move out of the screen in your direction). At 0 convergence, everything is flat and at infinite depth, while at a very high convergence you leave the game behind you (and you see either the horizon skybox or nothing).

You're right ofcourse, if you want more out-of-screen effects you need to increase the convergence distance. I was incorrectly thinking about bringing all objects closer to the screen, which means making it flat.

There are games like Metroid Prime where just 1 unit of convergence in Dolphin is too high, while zero is flat with the scene pushed at full depth. Would decimal values be a future possibility? As an opposite example, other games have very little range of convergence ingame with the hotkeys. This is what OP is complaining about.

I will either support decimal values or scale settings. Scale settings are still not correctly implemented, so I may implement decimal support for the 5.0 stable.

This leads me to another question: what makes convergence need so different values for the same visual convergence? This applies to PC games too with 3D Vision. The only idea I have is that models may be at different scales, like in X game the character is 100 meters tall while in Y game the character is 1 meter tall (in videogame making distance measurements), but for us it looks the same because everything is scaled, and thus convergence needs to be scaled in the same proportion.

Many games render at different scales and Dolphin has no way of knowing which scale the game is rendering at. So we will have to build a database in the GameINIs that define a scale for each game based on user testing.

[eckso]

what makes convergence need so different values for the same visual convergence? This applies to PC games too with 3D Vision. The only idea I have is that models may be at different scales, like in X game the character is 100 meters tall while in Y game the character is 1 meter tall (in videogame making distance measurements), but for us it looks the same because everything is scaled, and thus convergence needs to be scaled in the same proportion.

Simple, developer choices.

Some decided that 1 grid unit represented 1 meter, others 1 cm, or something else.
That's not all, camera settings, focal length, etc.

It's basically how developers implemented things. That's the same story behind aspect ratios. Different developers understood sample scaling in different ways, maybe they even didn't understand it at all and hence why we see so many games slightly stretched (even with the PR 3472 in line).