Widescreen Gaming Forum

25-30 ... :?

id guess probably 20-30.

It's not a "mod". It's a customizable feature of the game that was added in a patch by the developers.

i was under the assumption that "mod" was short for "modification". my bad.
here is a screenshot that undeniably demonstrates the uselessness of this customizable feature. you would have to set your monitors at 90 degrees and sit directly in between the outers to get any use out of it. but like i said... to each his own.

Since rFactor is an open platform, there are literally hundreds of "mods", made by third-party rFactor enthusiasts. Semantics, but generally in the rFactor community a "mod" is not something that is released officially by the developers.

Looks like you're running a pretty large FOV in that screenshot. Mine's definitely less than that, but since I use TrackIR I have no good way of telling what it may be. Based on how your shot looks, I'm guessing smaller is better with the multiview option enabled.

Hi Jkeefe,

I have had a chance to check out rFactor now. It rocks! It's also just what I am looking for in good looking/playing games trying out different display architecture. Funny, I wasn't a petrol head before and certainly had no passion for F1 but now... y'know, rFactor could be good for the planet :)

As an implementation it is a bit of a black box thing and its hard wired (well, to be fair, not exposed) variables are a bit of a shame. It does indeed look a lot like they have implemented three cameras with the outer two having different headings. The artefacts in images others have posted along with some below are typical of this. I would think it unlikely that they go to all the work of doing image warping to allude to multiple cameras when you could just use them.

The minimum variables I would look to controlling would be the horizontal field of view and the rotational transform of the cameras (assuming an auto vfov). In my setup it is clear that the cameras are configured for a desktop monitor situation with possibly quite a tight angle on the side ones (tighter than typical TH2Go setups). This would make sense as ideally you'd make a mini cave around yourself.



First image is normal single camera opened up wide and the next with the multi screen flag checked. I highlighted the horizon artefacts on the right hand side.

My opinion here would be to go for the perspective distortion you get from a single camera (if that was all you could choose from). In the heat of the action you are very focussed on the centre (head tracking is another thread I think).

I have tried to buy this great great game (did I mention how great I think it is?) but the third party group handling sales have a serious restriction to online commerce and licensing (potential grumble for another thread in another forum on another website). So I am stuck playing the demo... with all those big trucks yet to race :(

Now what I am not showing is what can happen to fix the artefacts caused by this fix to wide angle cameras. I was hoping to specify the variables mentioned above to match my system which would be: 56 degree HFOV, 45 VFOV, 50 degree offsets for side cameras and all on the same vertical axis. This would give me an effective HFOV of 150 degrees as I use a 3 degree overlap on either side with the other channels. Once to this point, a spherical distortion applied to the finished frame, either in the gpu or in the projector helps alleviate that mismatched horizon/vanishing point effect. If I could get the game to conform I would post the images but turning on the distortion the way it is configured now is quite unpleasant.

How this distortion works is a bit involved. I don't have any documents of my own I can make public but this link can give an introduction to it.
http://orihalcon.jp/projdesigner/

Hopefully Image Space Inc. will make the necessaries through their api so we can give it a go or better still, believe us that they need to be exposed.

God its good though

keltie:

rFactor has been available on CD for a while now, but the last time I looked it, the seller shipped only within the continental US. You might investigate the current state of things, or perhaps you could even have someone up here buy it for you and send it down (where it would languish in customs for quite a while I know, but...).

gogamer.com

Thanks for posting that link of yours too--looks useful.


Christopher Snow

As to this discussion:


The chief advantage of "Subviews=2," as I see it lies in it's potential use for projection of the images onto a SPHERICAL type screen. A screen which is curved in both direction, planetarium style.

First:

If Subviews is left at "=0," it becomes technically impossible to project a "true" image onto a hemispherical (180 degree surround) screen without distortion--the corners stretch out to infinity.

I will admit that you could probably live with the distortion in actual practice--it would be far off to the sides--but it's not quite ideal.

With Subviews=2, however, it does seem to me it should be possible to emulate a true surround view pretty accurately, and without huge expense.

The screen keltie has is curved only horizontally, and if I had to guess based on the two images I see close above, I would guess the overall "sweep" would encompass something like 90 degrees if you stood in the theoretical center of the curve, rather than a true 180.

Doesn't mean you can't help prove the idea though, and I think you can, if you are willing to experiment a bit more, keltie. :)

--------

Assuming I am correct on the overall sweep (feel free to adjust please as required by your actual setup), it seems to me you effectively have 3 stiched-together screens, each showing a horizontal sweep of ~30 degrees if Subviews=2.

In rFactor, FOV is set up in terms of vertical FOV, so if we assume each of your projectors is displaying something like a conventional 4:3 image, the vertical FOV in rFactor should therefore be set to about 24 degrees or so.

I THINK, if you do this carefully....perhaps draw it up on paper in order to derive the exact v-FOV (based on your actual setup), the horizontal "kink" you see in the guardail/grandstands/building (where the images overlap) should--theoretically--disappear.

I should note too, and again given that your screen is curved only in one direction, it seems to me the "kink" would actually disappear for you only at the actual horizon (the kink should gradually "reappear" the further you look both above and below the horizon)...but if I'm right it might help to prove the concept and potential worth of even having "Subviews=2" as an option.

As for me? I haven't seen it in action here yet, but I do believe I've already convinced myself it will be a necessary part of any "true" scaled surround view system.


Christopher Snow

Hi Christopher,

There are warping standards based on spherical displays however for displays with a limited VFOV it is practical to use a cylindrical screen and do your image warping as is if you were viewing 'through' a spherical screen to the 'flat' dimension of the cylindrical one. Up to 40 degrees VFOV the flat aspect of a cylindrical screen seems to be accepted. Between 40 and 60 degrees you see what is a section of a doughnut being deployed, either as a hard screen but often as a soft one with a vacuum/suction behind it. Displays attempting larger VFOV that 60 degrees really need to be getting into being a partial dome. A side effect of this warping is that the VFOV gets both clipped and cut off so a larger angle is required to achieve the same effective result of actual VFOV. Where you have multiple channels/cameras this warping happens on each one based on their properties.

In my case I have a physical 40 degrees VFOV based on the sweet spot dictated by the screen radius. The actual HFOV is 150 degrees. Its actually not an idea I am trying to prove so much as deploying a simulation industry standard to the issue of this thread.

My own applications (those targeted at that particular display device) are written to use 3 x 56 degree cameras giving a total effective HFOV of 150 degrees (recall the 50 degree offsets and 3 degree overlaps above which I use for projector blending) and a VFOV of 45 degrees giving an effective actual 40 degrees.

Even with these variables being exposed by the application (camera perspective properties and rotational offsets matching whichever reality your display system has) you will still have the issue of proximity manifesting itself in an obscure scale of the HUD or car body in this case and there will need to be very specific image warping to smooth that horizon line (and all things deviated from it) between the channels. This second part I have control over and as soon as I can get into the guts/API of rFactor I will post what I am talking about in pictures.

Cheers

Garry