Page 1 of 1

Looking For New SHOCT Formula - Does it Exist?

Posted: Fri Aug 23, 2013 5:28 pm
by Neil
Hi Guys!

I'm putting some thought into the next round of SHOCT. I have some new ideas I'd like to try out, but I need to confirm (or discover) a formula:

When dealing with stereoscopic 3D images, the deeper you are in the scene, the more objects in the distance separate (objects in the left eye go left, objects in the right eye go right).

Is there a basic formula for determining the relationship between IPD and how far the left and right image separate in relation to distance?

What I'm looking for is a formula that will fill X:

With an IPD of 64mm, objects will separate X amount at 5 feet, X amount at 10 feet, X amount at 15 feet, etc. etc. It's ok if the separation is measured in mm instead of pixels. I'm just trying to find a firm relationship or assumption that is consistent.

Regards,
Neil

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sat Oct 05, 2013 7:02 am
by virtualvision
I'm assuming you mean the distance between the object on the left half of the display and the right half of the display as one single plane, rather than the distance between the coordinates of the object in the left eye and the coordinates of the object in the right eye.

Well, disregarding optical distortion, which I imagine you can probably reuse the code to handle...

You still need the distance from the viewer to the screen. I believe in the Rift that's something like 1.5in - 2.5in depending on the eyecups and how the viewer has it adjusted.

I whipped this formula together:
VIRTUAL REALITY STEREOSCOPY FORMULA
Point Q = A point inside virtual reality which is equidistant to the center of the left and right eyeballs.

DLR = IPD x (DV - DSE)/DV

DLR = The distance between the left-eye representation of Point Q and the right-eye representation of Point Q.
DV = The distance from Point Q to the viewer.
DSE = The distance from the screen to the center of the viewer’s eyeball.
IPD = Interpupillary Distance
Note: you can calculate a point that is not equidistant to the left and right eye, but the formula will be a lot less pretty. Basically, you would just use one distance (from the point to the right eye or left eye) and set up a ratio between the distance from the point to the screen and the from the screen to the eye at the same angle. Then multiply that by the IPD.. but you would probably want to take note of the offset as well.

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sat Oct 05, 2013 8:56 am
by Neil
Image

Thank you for putting this together!

One of the big challenges with VR drivers is getting an accurate World Scale in relation to IPD. On paper, we think we nailed the process. Just one of several experiments running in Vireio Perception Labs!

Thanks again!

Regards,
Neil

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sun Oct 06, 2013 5:17 pm
by Fredz
Ideally if you want a correct world scale the equation should also take into account the scale of the game, ie. the relation between units in the virtual world to units in the physical world.

Most of the time these values can be found on the Web for the most popular games/engines (ex. in Quake III 1 unit = 1 foot and players are 7 feet high, in COD4 1 unit = 1 inch and players are 6 feet high, etc.).

The game should still enforce a consistent scale for all the objects in a scene though, which isn't necessarily always the case (guns in FPS and cockpits in some space or racing games for example).

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sun Oct 06, 2013 8:36 pm
by Neil
We have methods in place to get correct World Scale. We'll have a quick and dirty feature in the first Vireio Perception build, but we have a much more accurate and easier to use version to follow - if the coding works out.

The one thing that is difficult is height. For example, the objects might be to scale, the doors and walls, the whole world in fact - but the camera placement for the eyes could be off! I was doing some testing this evening with Left4Dead, for example. Why does my character seem shorter than everybody else in the game?

So I compared to the game and the crosshair placement without a 3D driver, and the perspective was the same. More than that, if you look at L4D and see that there are different characters to play, all of whom are different heights...how is it that the perspective is the same every time?

I think getting accurate character height is going to be a trade-off with VR drivers. Though, do you think native VR games are going to take that measure of considering the height and unique camera placement of different characters? We'll have to wait and see. I don't know if people started considering that kind of thing. Would that negatively impact the game's dynamic and balancing somehow?

Regards,
Neil

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sun Oct 06, 2013 11:21 pm
by virtualvision
The main thing that bothers me in Vireio is not really perspective stuff but more the menus. I can't really play a lot of games like Skyrim because I can't see all of the menu options. If Vireio fixed the menu to a certain point in the dimensional space, and I could look up/down/left/right at the menu, that would be the difference between me actually being able to use it to PLAY a game instead of just using it as a sort-of demonstration.

Even if it's just Skyrim. I think literally 80%+ of people who download Vireio do it mainly for Skyrim.

Also, I have to use disable all of the shadows to use it :-(

Re: Looking For New SHOCT Formula - Does it Exist?

Posted: Sun Oct 06, 2013 11:55 pm
by Neil
Did you see the Vireio demo on the front page?

Regards,
Neil