I play with the multimonitor options. It is good the new slider screen angle, but now it is a bite difficult to adjust to a real FOV.
So, you can calcule what FOV is what you seeing with simple math, knowing the distance from your eyes to the monitors, knowing the size of monitors, and assuming that the angle between them is correct.
In the last patch was easy to put the final FOV, but now I don't understand why you put the central FOV, but I think you have a good reason. It have some minor problems like if you change the config beetwen 5-1-5 to 1-1-1 the general FOV will change drastically
Also, I like more a Cylindical correction than a multiview, but I think it will consume a lot of FPS, so without multiview ( 0 - 1 - 0 monitors config) I have 74FPS in a grid start in pause, with 1 - 1 - 1 I have 63 FPS, and 39 FPS with 5 - 1 - 5 with the same total FOV.
I remember you said you would do a graph that is generated automatically to find the actual arrangement of the monitors relative to the FOV. I imagine that like an image I uploaded in the last patch topic (I upload again), where indicates the distances between visor and monitors (radius circle) relative to the FOV, and angle between them. LFS only need one more data to calcule, and is the width of each monitor.
It will very usefull for a very big multimonitor/multiprojector view to match the exact FOV, and for sure very easy. So the big majority of people don't know what really means FOV
I think you can use your TH2GO for the 3 center monitors, and use SoftTH to add up to 1 each side monitor... 1 + 3 + 1, with the SoftTH you can manage only 3 monitors, but the center monitor can have the width what you want, so you can do a config like 1024 + 3840 + 1024, and play it in fullscreen. Now I have TH2GO and 3 monitors, and I thinking about 5 monitor setup, thanks to the new multimonitor options of LFS
I don't know the reason. Are you using a wide screen mode? What is the effect of trying to use FlipCams... does anything happen at all? Is this the same for anyone who tries to use FlipCams with Z18?
No, all these updates are concerned only with the main view and I don't think I'll be adding any more features, in fact I've been trying to get off views and back to physics for at least two weeks now!
Thanks for the picture. I can imagine a 5 monitor setup using a future version of Eyefinity will be great, with each screen at 45 degrees, 5 sides of an octagon would be quite immersive. Even better would be a three projector setup...
That reminds me of something else I was thinking. This is another thing for the future but LFS could fully support a stereoscopic view, needing only a dual head card and two projectors. So a left eye view would be drawn on the left half of the render surface - and sent to the left projector. A right eye view would be drawn on the right half and sent to the right projector.
The two views would just need to be rendered from a slightly different view point (the distance between two eyes). The projectors would need polarising filters on the front and then you could see the full 3d view through polarising glasses. I haven't read up fully on this but it seems that you would not need any special "3d" stuff apart from the filters and glasses. A bit expensive with two projectors but apart from that it would be quite good.
Add to that a TrackIR but with the rotation disabled, so it just moves the view point left and right a bit when you move your head, that would complete the illusion very nicely.
The reason is because the FOV is calculated all based on the central screen's FOV. The number of screens, screen widths and monitor angles all affect the final FOV.
LFS calculates the values of each screen using an iterative process, working outwards from the centre, each screen's FOV and offset is worked out from the result of the previous one. Finally, from the result of the left and right screen edges, the final FOV is found by a simple subtraction.
So it is not easy with this new system, to set the final FOV as a user input and then somehow find the correct input variables to produce that output.
That is the one thing we lost in going from Z17 to Z18, but we gained a lot of flexibility.
I guess it could be possible to compute the main screen fov from the final FOV (as a user input) by another iterative process, taking a guess for the first iteration then repeatedly performing the original process I described until finally homing in on the best value, just like a user must do now, selecting an input value to produce the desired resulting FOV...
I did say say that was my plan but now I have run out of time (and space on the View Options page).
Are all the offsets applied correctly?
Does the mirror work in 3d?
What about the shadows? I remember someone posting a 3d thing before and the shadows were all offset as well, because the shadows themselves are also a rendered image and they were being offset by the same software that was creating the offset for the 3d.
Does that use alternate frames to render the left and right views?
While we're on the view settings and cameras, would it be possible to have a camera height in Shift+U mode level with the ground (0.0 m height instead of minimum of 0.4 m that it is now)? Since telemetry export of "ride height left" is based on suspension and not body height from the ground, it would be usefull to have a camera on the ground level to check for body bottoming out.
We've had problems when body sustained frontal damage and weren't sure if it's actually going to scrape the ground (the sound wasn't there during full throttle), but on braking it would scrape and unload the front tires - which would then lead to huge flatspots and tire blowing 10 laps early.
I was thinking the very same thing, but it would need some serious horsepower I think. I get 150 - 180 FPS with three screen / three viewports and about 130 FPS with four viewports but it drops to 10 FPS when I stretch to the fourth monitor (all monitors set to 1680x1050, I got 30 FPS by reducing the resolution).
Running two graphics cards, SoftTH with the secondary card running two monitors and the TH2G from the main card might just work, and I'm awefully tempted to give it a go
As an interim fix, you can use LFS TV Director to set the shift + u camera to any height, from 0 (usually under the ground) to 700m upwards if you really want to.
I'm not sure it would be very artistic, being able to see through the ground. That is the problem with allowing the camera to go to zero height. The bottom half of the image is underground, you can see edge of the clipping plane plus distant objects through the "ground" and it's one of those things we try to avoid as game developers. There's no cars or world there at all, just a load of textured triangles but we try to avoid showing that.
You're right... it would be great to have some good artistic shots from this angle but cropped out the ground. But from the developer view, I'd agree to keep it on the 0.4, it doesn't look good in those shots and some people may get confused.
Hmm, you could go lower (see attached) but then you have to keep the camera horizontal as if you tilt it down a bit, you start to see some untextured stuff.
No, because the autocross objects are saved in each track. So all tracks would have to be updated. Making a special autocross objects file which is available on all tracks would be a nice idea, also many updates I would like to do to the autocross editor but will not go into them now.
Now is all about : fixing any bad bugs left in Z18 so I can get back to physics. I have massive lists of things to do, surrounding me in my office and I am not looking for more things to add to the lists, because I will be swamped by paper.
Or what about increasing the maximum height? Sometimes you want to view a big part of the track. Is it not implemented also because of the unnecessary untextured spaces?
Having this feature in LFS itself would be great. It makes you feel much more there when not 100% glued to the car. Suspension movements and track unevenness become much more apparent than the current view system could ever hope of achieving to show. I think this would be a very good addition coupled with the new tyre physics and/or Rockingham.
The funny thing is, at least through InSim the basic functionality is a dead simple thing to code as I've found when implementing my own version. All you do is instead of instantly applying the car's pitch/roll to the view, you average the values over the last X physics iterations (say a queue/buffer size of 20) and apply that instead, coupled with the automatic camera smoothing of InSim. Then add a proper over-/underflow handling (so the view doesn't go beserk when doing a barrel roll) and you're done. No need for actually simulating the driver's head in any way.
The only issue I can see for a proper implementation is what to do when the user presses the look buttons or uses TrackIR (= does not look into the same direction as the car is pointed), and for both occasions simply disabling the free-float or fading the effect out the farther you look away from where the car is pointing would work as a preliminary solution. The real solution would of course be a proper translation of the car vector to the look vector and all the calculations that go with that... I can see how a proper solution could be too much work for now.
Now personally I have no problem with just continuing to use InSim for that, but I think a lot of people miss out on this great little feature unless it's implemented in LFS itself.
But sorry for OT, this has really nothing to do with the test patch.
I'd like to thanks Scawen for his very good work.
The feature I wanted (triple screen external views) is back, and now we can even adjust screen angles... very nice options
Just a little problem though: it seems that the said "Option to force external views to use full width"isn't saved.
I use SoftTh with 3 screens so I ran into this matter when I changed the number of screens. When LFS is set with 1 - 1 - 0 screens or the opposite, menus are moved to one side and buttons don't seem to work.
But in fact, it is possible to click the buttons. For example when everything was moved to the right I had to click more to the right, and below too. You just have to click randomly until you find one No need to delete anything, but I reckon it's not really comfortable.