hmdq

[Rosebud47]

´hmdq´ is a small tool which reads out detailed values and figures of your HMD ( Head Mounted Display )

 

It could be downloaded for free:

https://github.com/risa2000/hmdq/releases

 

After downloading and unzipping, simply execute the application, while your headset is connected and running. ´hmdq´ now creates a log file.

 

Now open the Windows console ( Windows DOS-box ) navigate to the folder, where the hmdq.exe is and execute the hmdq.exe. It now shows in the DOS-box all relevant information of your HMD like in the screenshot attached ( which is the configuration of my P5k+ )

 

The interesting infos are the ´real´FOV settings of the HMD.

 

In next step I´ve copied the FOV values from hmdq to the stereo.lua in DCS and have overwritten the values for FOV left and right eye in the stereo.lua ( the ´stereo.lua´ located in

 

...\DCS\config\MonitorSetup\stereo.lua ) with the values from ´right eye head´ and ´left eye head´ of the hmdq dos-box.

 

Now, I don´t want to convince people - everyone has to trust his own eyes...

At this point I´m quite struggling to share my impression, but so far it seems that everything looked more right to me, with the exact FOV values. It´s difficult to describe. Most visible is the structure of the landscape, like small hills, big hills mountains, etc. Also I think the dimension of objects in relation to the distance from the viewing point is more correct... I would say it looks better with the real FOV values of your HMD, but again: everyone has to make an impression of his own.

 

Didn´t put it in the Pimax thread, as it works for every HMD and could be of interest especially for HP Reverb user, as the Reverb got an 1:1 ratio.

 

Edit: after editing the stereo.lua I would recommend to reboot your PC.

Edited August 11, 2019 by - Voight -

[JoeyJoJoJunior]

Can you be more specific exactly which lines you are editing. Not immediately obvious to me.

[Rosebud47]

Open with an editor in windows administrator mode the ´stereo.lua´ located in

 

...\DCS\config\MonitorSetup\stereo.lua

 

exchange the default values in red with the ones with the ones you get from hmdq:

 

_ = function(p) return p; end;

name = _('Stereo');

Description = 'Stereo render two viewports for left and right eye respectivelly'

Viewports =

{

Left =

{

x = 0;

y = 0;

width = screen.width * 0.5;

height = screen.height;

viewDx = 0;

viewDy = 0;

aspect = screen.aspect / 2;

eye_shift = -0.032;

--tans of side angles

projection_bounds = {

left = -1.742,

right = 1.346,

top = 1.270,

bottom = -1.270,

}

},

Right =

{

x = screen.width * 0.5;

y = 0;

width = screen.width * 0.5;

height = screen.height;

viewDx = 0;

viewDy = 0;

aspect = screen.aspect / 2;

eye_shift = 0.032;

--tans of side angles

projection_bounds = {

left = -1.346,

right = 1.742,

top = 1.270,

bottom = -1.270,

}

}

}

 

UIMainView = {

x = 0;

y = 0;

width = screen.width;

height = screen.height;

}

GU_MAIN_VIEWPORT = UIMainView

 

 

What I am wondering about is, that the default values are so small like 1.xxx and the real values are much bigger like 50.xxx ...

 

I think the default values are kind of one-size-fits-for-all, what might have worked for the very first generation of VR headsets, but especially for the wide FOV HMD these values are off and need to be customized to reach the potential of your device and for proper graphical fidelity.

 

Here´s the sensitive part in trying to adjust the values:

Personally I only trust hard facts, things which could be measured at any time, every place, reproduced any time, every place or followed up any time, any place - but at this point we´re only dealing with perception, what is more subject of personal impression, than common hard fact.

 

But beside the one-size-does-not-fit-to-all compromise, there must be a technical explanation for the FOV and how off values it result in perception - I don´t have this technical explanation, but would think of kind off effect of magnification glasses, if the FOV is off set, respectively narrower than the physical limitations. In VR you don´t perceive the magnification effect, as you don´t have a negative to compare to the positiv.

 

After putting in the values from hmdq, me got the impression that the perception of depth in the image has become much more accurate and ´widely´ - the first flight in DCS with these values reminded me instantly of some TV documentation, where people fly at 500 ft. filming the ground.

 

Also before, I always fiddled around with the IPD settings in DCS, as the cockpits appeared to tiny, like filigree toy cockpits, so I have had used the IPD setting as a world scale to by lowering the IPD setting to get a more accurate image ( which has also negative sideffects ).

 

Maybe with regular FOV HMDs the effect of correcting the values is not that much perceiveable, than with a wide FOV HMD, but at least it should be better than the one-size-fits-for-all values.

 

Would be interesting to read if you could see a difference at all.

 

Edit:

openXR solves this problem, when it is implemented in the driver of the headsets ( Valve, Microsoft and Oculus already stated, that they will implement openXR ) and implemented into DCS world. DCS world would then directly interacts with the characteristics and specification of each specific headset to set up the right values for rendering and adjusting the images in VR.

Edited August 13, 2019 by - Voight -

[kega76]

Where exactly in the hmdq file did you pick these values (which values?)? I get a lot of different values, but they doesn't directly translate to this LUA file.

[Rosebud47]

The values are not in the log file. You have to execute hmdq.exe via an I/O system like windows console and then hmdq returns the specific values of your headset into the console ( in former times we named it the Dos-Box of Windows ).

 

I could imagine that nowatimes people are not used to handle console inputs.

 

If you are still interested in this and google couldn´t help, please give a sign and I´ll try to support you getting the values of your headset, but you have to find and identify the windows console first to browse to the directory, where you have copied the hmdq folder and its .exe.

[kega76]

Hi and thanks for helping. I did do the cmd and got the values. However I got several values that wasn’t even close to the ones in the LUA. Also there were several for left and then for right. However, I did write in what I thought was correct, so for left eye (left, right, to and bottom) and same for right. I had like left -43.xx right 43.xx etc.

 

I fired up DCS and did a test run. Didn’t see a immediate difference but I didn’t test it for long. I have the Reverb and have had a sensation of eyes swimming a bit in DCS, not X-plane, and maybe this might fix it. Also it stated that the IPD was 65.

 

Am I in the ballpark here?

[Rosebud47]

Yes, me too wondering why the lua values are so much different. Have you restarted your system after you´ve edited the .lua? I think it´s necessary, as windows mixed reality for steamvr need to be restarted, respectively openVR of steamVR, which connects to openVR within DCS.

 

Yes, it´s the values for left eye: left, right bottom, top and the same for the right eye - I´ve took the values from hmdq, right eye head and left eye head. Don´t know if there´s a difference to the values hmdq, left eye raw/right eye raw.

 

Yeah, me too didn´t long time test and back and forth test etc. but got the impression that these values work much better for me with the P5k+ in normal FOV mode ( approx 150° horizontal FOV ).

 

If you don´t see a difference on first view, first means that there´s nothing gone completely wrong by editing.

 

The first thing which came to my eyes was a less flat landscape - mountains appeared more massive ( and more real ). Take a flight close to the caucasus mountain area and especially if your observation into the depth of the picture has been slightly changed. Maybe it does not at all, maybe only so slightly that it is not so much perceivable with a narrow FOV, maybe I´m nuts and talking s***.

 

So far I like these new settings - looks better to me, but need to do some more flights and have a look for a viewpoint, which obviously changed to a better more natural perspective from the viewport settings before.

 

Would be glad, if you would share your impression after some more flights even, when there has nothing changed.

[kega76]

Hi, so now I've done several flights after a restart. It's hard to really say that something is improved, but it didn't degrade anything for sure. I don't have that eye strain for sure, but that could have been a steam or WMR software upgrade.

 

I will definately stick With this now for a while and test it some more. With the Reverb it's just crisp, crystal clear now, I love it! I'm pasting my settings below just for reference to others. Thanks again for posting this!

 

Left =

{

x = 0;

y = 0;

width = screen.width * 0.5;

height = screen.height;

viewDx = 0;

viewDy = 0;

aspect = screen.aspect / 2;

eye_shift = -0.032;

--tans of side angles

projection_bounds = {

left = -47.38,

right = 43.62,

top = 45.42,

bottom = -44.82,

}

},

Right =

{

x = screen.width * 0.5;

y = 0;

width = screen.width * 0.5;

height = screen.height;

viewDx = 0;

viewDy = 0;

aspect = screen.aspect / 2;

eye_shift = 0.032;

--tans of side angles

projection_bounds = {

left = -43.60,

right = 47.63,

top = 45.24,

bottom = -45.13,

[Rosebud47]

That´s great! Also got the impression that the SDE had become a little more visible - I usually don´t see any SDE in the P5k+, but after your comment, it could be also an indication of a more clear and crisp image... need to put more attention to this.

 

When I got more time, I will make some before/after screenshot with the steamVR mirror.

 

After every DCS update, the edited stereo.lua will be overwritten with the default stereo.lua.

Having a copy of the edited values to put in again after every DCS update is recommendable.

[Baldrick33]

The figures in the lua file are tangents. The figures out of hmdq are in degrees. Hence the difference.

 

 

I tied entering the values from vrcmd which I think are the same as the lua but didn't see any difference

[Rosebud47]

@Baldrick33

Thanks for sharing your impression and solving the relation of different figures.

 

There surely need to be done proper testing, which I want to, when I find half a day space to do so.

So far we are just sharing and discussing impressions, which is not science, but at least makes us understand the whole thing a bit more and maybe we could get to individualy better results by try and error.

[kega76]

Just an update after todays update. When I was about to change the Stereo.lua again today I came over a file under disk/user/savedgame/DCS.openbeta SAmsung Windows Mixed Reality 800ZBA.LUA

 

This one contains the following:

EYE_LEFT = {

L_OUT =1.230279,

L_IN =1.045660,

L_TOP =1.437487,

L_BTM =1.418807,

}

EYE_RGHT = {

R_OUT =1.230279,

R_IN =1.045660,

R_TOP =1.437487,

R_BTM =1.418807,

}

These numbers are more familiar with the ones we extract from your tool. I've added them now to the Stereo.lua but haven't had time to test it. I'll report back when that is done...

 

Update: have tested now and I don’t really see any difference. Seems like I can put whatever in the Stereo.lua?

Edited August 14, 2019 by kega76

[Rosebud47]

Yeah, this is a dead end.