It is currently Sun Jan 19, 2020 10:39 pm



Reply to topic  [ 96 posts ]  Go to page 1, 2, 3  Next
 Motion Blur in Oculus Rift 
Author Message
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
does anyone else notice significant motion blur in the rift? it seems most noticable in Tuscany and any of the UDK maps (not so much in First Law , maybe just because there is so much black).

I have a pretty strong pc (8 core AMD, 8gb RAM, SSD) running a Gigabyte Radeon HD7970 3GB card, so my FPS should be completely fine but i get this motion blur that really takes away from the immersion and makes me pretty sick when looking around. I've never heard anyone mention this so i'm assuming it's something with my setup.

i tried playing around a little with the Catalyst settings (anti-aliasing, textures, etc.. ) but haven't found anything that makes it better. Currently running the newest driver available and Catalyst 13.2 (was on 13.1 same issue)

any ideas?? i know Palmer and team run NVidia GTX 680's , possibly a problem with Radeons ? anyone else with a Radeon 7970 or similar that can comment?

any help would be greatly appreciated (other than the motion blur , it's awesome!) . I will be testing with a Dell Precision laptop this weekend (NVideo Quadro card) so I'll see if it looks any better)

thanks
Mark


Fri May 10, 2013 7:45 am
Profile
Binocular Vision CONFIRMED!
User avatar

Joined: Sat Aug 11, 2012 4:41 am
Posts: 247
Reply with quote
It's not your PC. The motion blur is due to the slow lcd switch time of the screen used inside the oculus rift.


Fri May 10, 2013 7:54 am
Profile
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
If that is the case , it's really bad and i would almost consider it unusable. The field of view is great and tracking is awesome! I expected the resolution to be low and i'm fine with the basic resolution (looking straight ahead) , but the motion blur makes it impossible to use without getting sick.

I'll test with a few different machines and report back my findings


Fri May 10, 2013 8:04 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Apr 07, 2007 4:34 pm
Posts: 2910
Location: Sweden
Reply with quote
The Oculus rift is "a proof of concept" and in the current shape "for developers only". The main purpose was to make a hmd with a good fov+headtracking for a reasonable cost which they have succeeded with. For a device good on consumerlevel we have to wait a bit more. Also: Softwares needs to be developed to make this whole thing a reality. Meanwhile the hardware should also become cheaper/performance.

_________________
Mb: Asus P5W DH Deluxe
Cpu: C2D E6600
Gb: Nvidia 7900GT + 8800GTX
3D:100" passive projector polarized setup + 22" IZ3D
Image


Last edited by Likay on Fri May 10, 2013 8:13 am, edited 1 time in total.



Fri May 10, 2013 8:12 am
Profile
Diamond Eyed Freakazoid!
User avatar

Joined: Mon Jun 04, 2012 1:25 am
Posts: 790
Reply with quote
atreu69 wrote:
If that is the case , it's really bad and i would almost consider it unusable. The field of view is great and tracking is awesome! I expected the resolution to be low and i'm fine with the basic resolution (looking straight ahead) , but the motion blur makes it impossible to use without getting sick.

I'll test with a few different machines and report back my findings


if you want to sell it PM me. Atreu sounds french. I live in Germany.


Fri May 10, 2013 8:13 am
Profile
Diamond Eyed Freakazoid!
User avatar

Joined: Thu Oct 18, 2012 3:34 am
Posts: 733
Location: Brighton, UK
Reply with quote
atreu69 wrote:
If that is the case , it's really bad and i would almost consider it unusable. The field of view is great and tracking is awesome! I expected the resolution to be low and i'm fine with the basic resolution (looking straight ahead) , but the motion blur makes it impossible to use without getting sick.

I'll test with a few different machines and report back my findings

Different machines won't help. As Marbas said, it's caused by the LCD panel within the Rift having a poor pixel switching time. This is a physical hardware limitation, it has nothing to do with software. The best you can do in software is to ensure that you have a machine capable of putting out reliable high framerates. The blurring will be most pronounced when looking between areas of starkly different colours/brightnesses.

_________________
Sometimes I sits and thinks, and sometimes I just sits.


Fri May 10, 2013 8:22 am
Profile WWW
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
I completely understand the rift is "proof of concept" and lower resolution than the production versions. The device is just about what i expected. The only reason for posting this is because although i heard many times about the lower resolution and i expected this, i didn't hear many posts talking about the motion blur.

I'm not in any way complaining about the rift itself, it's awesome! the field of view and motion tracking is amazing. I just want to try to determine if what i'm seeing is being caused by LCD pixel switching time limitation or if a video card configuration is causing this.

Thanks
Mark


Fri May 10, 2013 8:22 am
Profile
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
TheHolyChicken wrote:
atreu69 wrote:
If that is the case , it's really bad and i would almost consider it unusable. The field of view is great and tracking is awesome! I expected the resolution to be low and i'm fine with the basic resolution (looking straight ahead) , but the motion blur makes it impossible to use without getting sick.

I'll test with a few different machines and report back my findings

Different machines won't help. As Marbas said, it's caused by the LCD panel within the Rift having a poor pixel switching time. This is a physical hardware limitation, it has nothing to do with software. The best you can do in software is to ensure that you have a machine capable of putting out reliable high framerates. The blurring will be most pronounced when looking between areas of starkly different colours/brightnesses.


Thanks @TheHolyChicken , yeah i was afraid that this might be the case. that does make sense , since it's most apparent in Tuscany which is very bright.


Fri May 10, 2013 8:25 am
Profile
Binocular Vision CONFIRMED!

Joined: Tue Jun 26, 2012 2:50 am
Posts: 284
Reply with quote
The motion blur doesn't make you sick, it's the looking like you're moving when you arn't and no positional tracking which does it.

Perhaps you have a fault with your unit as I really don't see anything wrong with the blur, it's not really any worse than on my monitor yet I see people crop up stating how bad the blurring is that it's on the border of unusable :?

Do you know anyone else who has or is getting a devkit so you can check if the blur is the same or if you just have a problem with your unit.

While I can notice the blur it's miles away from being anywhere near unusable for me, and I refused to get an LCD monitor for a few years after they came out because the motion blur just annoyed me too much. I really think some people must have faulty panels if they consider the blur even in the ball park of unusable.


Fri May 10, 2013 8:27 am
Profile
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
Endothermic wrote:
The motion blur doesn't make you sick, it's the looking like you're moving when you arn't and no positional tracking which does it.

Perhaps you have a fault with your unit as I really don't see anything wrong with the blur, it's not really any worse than on my monitor yet I see people crop up stating how bad the blurring is that it's on the border of unusable :?

Do you know anyone else who has or is getting a devkit so you can check if the blur is the same or if you just have a problem with your unit.

While I can notice the blur it's miles away from being anywhere near unusable for me, and I refused to get an LCD monitor for a few years after they came out because the motion blur just annoyed me too much. I really think some people must have faulty panels if they consider the blur even in the ball park of unusable.


Thanks Endothermic - I have another Rift coming but probably not until the end of the month, so i could compare at that time. I wouldn't have mentioned anything if it wasn't pretty bad as i moved my head, so i was really hoping that i was just a software configuration issue.

I guess like you mentioned , the best test would be to try another Rift to eliminate the possibility of a hardware issue (can't do that until later this month), so for now i will try my rift with another pc/laptop and see if it looks any better , to eliminate the possibility of a software problem.

It is pretty awesome though, my favorite "by far" is First Law (the space sim demo) , and i'm not even a big space sim fan ..lol , but it's amazing!


Fri May 10, 2013 8:40 am
Profile
One Eyed Hopeful

Joined: Tue Jan 22, 2013 10:07 pm
Posts: 28
Reply with quote
Endothermic wrote:
While I can notice the blur it's miles away from being anywhere near unusable for me, and I refused to get an LCD monitor for a few years after they came out because the motion blur just annoyed me too much. I really think some people must have faulty panels if they consider the blur even in the ball park of unusable.


One persons usable is another persons unusable. Take vsync as another example. I have friends who barely even notice if a game has vsync disabled, yet what is perfectly fine for them is utterly unplayable for me. Same when it comes to framerate. I'll tolerate 30fps on console, simply because I can't do anything about it, yet on PC I have to have 60fps. Again, for some of my friends, whether it's 30 or 60 fps doesn't bother them at all.

People's tolerances are different - simple as that. If you watch some of the videos where Oculus are demoing the Rift at various shows, the motion blur gets mentioned from time to time in those.


Fri May 10, 2013 8:45 am
Profile
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
Fluke wrote:
Endothermic wrote:
While I can notice the blur it's miles away from being anywhere near unusable for me, and I refused to get an LCD monitor for a few years after they came out because the motion blur just annoyed me too much. I really think some people must have faulty panels if they consider the blur even in the ball park of unusable.


One persons usable is another persons unusable. Take vsync as another example. I have friends who barely even notice if a game has vsync disabled, yet what is perfectly fine for them is utterly unplayable for me. Same when it comes to framerate. I'll tolerate 30fps on console, simply because I can't do anything about it, yet on PC I have to have 60fps. Again, for some of my friends, whether it's 30 or 60 fps doesn't bother them at all.

People's tolerances are different - simple as that. If you watch some of the videos where Oculus are demoing the Rift at various shows, the motion blur gets mentioned from time to time in those.



Maybe the term "Unusable" is a little harsh to describe it, but it's definitely annoying and really takes away from the immersion. Like someone mentioned , it's more noticable in brighter areas . My other concern is the UDK games resolution seems really bad, worse than the unity.

I guess the next step is for me to test with another system and report back. Thanks everyone for your comments! and sorry guys, i'm not selling! :)


Fri May 10, 2013 8:59 am
Profile
Certif-Eyed!

Joined: Sun Jan 27, 2013 6:15 pm
Posts: 555
Reply with quote
I have found that some demos it is more noticeable than in others.

But it's always there.

Until we can get a faster-switching LCD I think it will always be there.


Fri May 10, 2013 9:51 am
Profile
Cross Eyed!

Joined: Mon Aug 06, 2012 11:28 am
Posts: 108
Reply with quote
What I find much more annoying than the motion blur is the way the distortion shader reduces quality. Just seeing the undistorted view in the UDK editor (very detailed, even at the Rift's low resolution) and then seeing a shimmering, pixelated mess when running the game is really irritating.


Fri May 10, 2013 12:28 pm
Profile
Cross Eyed!

Joined: Sun Feb 17, 2013 10:34 am
Posts: 166
Reply with quote
Laserschwert wrote:
What I find much more annoying than the motion blur is the way the distortion shader reduces quality. Just seeing the undistorted view in the UDK editor (very detailed, even at the Rift's low resolution) and then seeing a shimmering, pixelated mess when running the game is really irritating.


Low quality optics may play a part in that, i'm sure things can be improved for the consumer version, but then there are also bound to be some limitations and trade offs that can only be mitigated at best... might just be what we will never get as good fidelity with warping and optics.


Fri May 10, 2013 12:53 pm
Profile
Cross Eyed!

Joined: Mon Aug 06, 2012 11:28 am
Posts: 108
Reply with quote
Machinima wrote:
Laserschwert wrote:
What I find much more annoying than the motion blur is the way the distortion shader reduces quality. Just seeing the undistorted view in the UDK editor (very detailed, even at the Rift's low resolution) and then seeing a shimmering, pixelated mess when running the game is really irritating.


Low quality optics may play a part in that, i'm sure things can be improved for the consumer version, but then there are also bound to be some limitations and trade offs that can only be mitigated at best... might just be what we will never get as good fidelity with warping and optics.

The optics aren't the problem here. Looking at the undistorted editor window through the Rift looks fine. It's the distortion filter that degrades image quality.


Fri May 10, 2013 1:07 pm
Profile
Cross Eyed!

Joined: Sun Feb 17, 2013 10:34 am
Posts: 166
Reply with quote
Laserschwert wrote:
Machinima wrote:
Laserschwert wrote:
What I find much more annoying than the motion blur is the way the distortion shader reduces quality. Just seeing the undistorted view in the UDK editor (very detailed, even at the Rift's low resolution) and then seeing a shimmering, pixelated mess when running the game is really irritating.


Low quality optics may play a part in that, i'm sure things can be improved for the consumer version, but then there are also bound to be some limitations and trade offs that can only be mitigated at best... might just be what we will never get as good fidelity with warping and optics.

The optics aren't the problem here. Looking at the undistorted editor window through the Rift looks fine. It's the distortion filter that degrades image quality.


Perhaps, or perhaps the fidelity is 100% preserved in the warp and the quality of the optics are not getting us 100% of that quality back. Or do you not see that as an option?


Fri May 10, 2013 1:54 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 01, 2012 10:47 pm
Posts: 2708
Reply with quote
Laserschwert wrote:
Machinima wrote:
Laserschwert wrote:
What I find much more annoying than the motion blur is the way the distortion shader reduces quality. Just seeing the undistorted view in the UDK editor (very detailed, even at the Rift's low resolution) and then seeing a shimmering, pixelated mess when running the game is really irritating.
Low quality optics may play a part in that, i'm sure things can be improved for the consumer version, but then there are also bound to be some limitations and trade offs that can only be mitigated at best... might just be what we will never get as good fidelity with warping and optics.
The optics aren't the problem here. Looking at the undistorted editor window through the Rift looks fine. It's the distortion filter that degrades image quality.
I read in a research paper (somewhere) that the solution to that problem is to distort the object geometry BEFORE RENDERING, so that the final image looks correct with no added "pre-warp" pixel distortion. Unfortunately, that would have to be designed into the game engine, but their tested working hack was to create a virtual spherical mirror (or virtual fisheye lens) for their in-game virtual camera, and let that do the in-game warping. Most of their testing was done using POV-Ray, but other game engines should work too. Bouncing the camera off an in-game chrome sphere should work fine, and the FoV can be set by magnification (zoom) into the appropriate portion of the sphere. You need an invisible camera though, but in VR that is the normal situation.
:D

_________________
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Image


Fri May 10, 2013 1:56 pm
Profile
Binocular Vision CONFIRMED!
User avatar

Joined: Wed Oct 17, 2012 12:42 pm
Posts: 227
Reply with quote
Endothermic wrote:
The motion blur doesn't make you sick...
Anything that doesn't match your usual visual experience can make you feel sick.

geekmaster wrote:
I read in a research paper (somewhere)...
Any links?

_________________
This member owns things.


Fri May 10, 2013 2:06 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 01, 2012 10:47 pm
Posts: 2708
Reply with quote
Randomoneh wrote:
geekmaster wrote:
I read in a research paper (somewhere)...
Any links?
Okay, fine, I will google it for you...
...
There, done! But that great website seems to be down right now.

Direct rendering of warped fisheye views for spherical mirror projection:
http://paulbourke.net/dome/mirrorrender/
Quote:
There are a number of techniques that can be employed to render content for non standard displays such as domes. Only in rare cases does the rendering software support direct rendering that will take into account the necessary distortion of the projector and projection surface. In other cases one needs to create the correctly distorted image from simple perspective or orthographic projections supplied in all rendering packages. This document will describe one method that is rather sneaky and will work with "any" package although it does suffer from some problems.

The technique involves creating a mirror or lens as a geometric primitive that is placed in front of a camera. The mirror is designed precisely so that the captured image, when projected, will look correct on the intended projection screen.
And here was the source code:
http://paulbourke.net/dome/mirrorrender/source/

The images are the key, but Google did not cache them. A Google cache with no images is here:
http://webcache.googleusercontent.com/s ... rorrender/

I wish you could see the images, because they explained it all very clearly. You will have to rely on the text now. I hope that website comes back up soon...

_________________
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Image


Fri May 10, 2013 3:47 pm
Profile
Binocular Vision CONFIRMED!
User avatar

Joined: Wed Oct 17, 2012 12:42 pm
Posts: 227
Reply with quote
Thanks, Mr. Paul is awesome.
I very much enjoyed this when I read it sometime ago. Maybe you'll find it interesting too.

_________________
This member owns things.


Fri May 10, 2013 5:19 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 01, 2012 10:47 pm
Posts: 2708
Reply with quote
Randomoneh wrote:
Thanks, Mr. Paul is awesome.
I very much enjoyed this when I read it sometime ago. Maybe you'll find it interesting too.
I sent Paul Bourke an email about his website being down before I posted that. Perhaps his site was down for maintenance, or perhaps my email brought it to his attention. But it is good to see his site back in operation again.

I saw your linked page before. In fact, I have read most stuff at his website, including the source code. :D

His new 2013 paper on space filling for procedural game assets is quite interesting, and his co-author lives practically within walking distance from me.

_________________
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Image


Fri May 10, 2013 5:24 pm
Profile
One Eyed Hopeful

Joined: Mon Mar 25, 2013 3:21 pm
Posts: 17
Reply with quote
Have you tried lowering the contrast and brightness on the Rift? I think that helped with the blur for me, or maybe I just got used to it.


Fri May 10, 2013 5:50 pm
Profile
One Eyed Hopeful

Joined: Thu Aug 23, 2012 7:22 am
Posts: 10
Reply with quote
kludd70 wrote:
Have you tried lowering the contrast and brightness on the Rift? I think that helped with the blur for me, or maybe I just got used to it.


I lowered the brightness all the way tonight based on one of the posts above and it seemed to help a little. I left contrast at 50, should I turn it all the way down?

Btw half-life 2 native that was just released is awesome ! Still blurry but awesome!

I didn't get a chance to test on a different system yet to see if nvidia performs any different than my radeon, but will test tomorrow


Fri May 10, 2013 8:20 pm
Profile
Two Eyed Hopeful

Joined: Fri Aug 17, 2012 12:51 am
Posts: 97
Reply with quote
The temporal aliasing Michael Abrash has talked about is also contributing to the motion blur. Faster pixel switching will reduce it but also introduce new problems which are just as jarring. A viable solution simply doesn't exist yet.

As for the distortion, the existing fragment shader technique shouldn't be lowering image quality to a noticeable degree provided your framebuffer is large enough. If the shader's sampling a 1280x800 image then yes, it's going to come out looking nasty. But increasing the size based on the distortion scale should result in a texel density that remains high enough to make any differences negligible.

The mirror/lens approach is definitely worth looking at if you're raytracing, but if not it'll still need to be done with a fragment shader, essentially giving you a less efficient version of the current method. I haven't tried using a geometry shader yet, but I expect it to require such dense meshes that speed would prevent it from being usable. Some of the stuff coming out of Siggraph in a couple of months may change that.


Sat May 11, 2013 12:34 am
Profile
One Eyed Hopeful

Joined: Thu Oct 18, 2012 10:45 am
Posts: 29
Location: Lithuania
Reply with quote
Confirming that you're not alone with blur being not just one 'annoying' aspect.

Have tested out dev.kit for about ~1hour at game dev. company way back in March,
my experiences in Tuscany:

* looking around very gently all looks deep and beautiful
* any faster head movement feels like a semi-transparent 'blinder'
is cast in front of me, makes me want to stay in one place instead
* the scenery is awesome, but it's beauty is cut-off sharply whenever I try to move bit faster
* noticed that I was moving from A to B only when absolutely necessary, ie.
to get to the next awesome spat and stay there, gently panning around and
enjoying the immersion

Unity's 'Angry Bots':

* strangely, moving robots didn't look blurry
* In this game i had to move a lot, with the
amount of blur I felt I much rather play it on LCD
* stereo sounds added good deal to the immersion

So, it wasn't a deal breaker, but was it's a big big handicap for
me when I tried it (restricts my desire to move). Still waiting for my own kit to arrive,
yet really really hoping Oculus finds a cure for blur in the consumer v.


Sat May 11, 2013 11:34 am
Profile
One Eyed Hopeful

Joined: Sun Oct 21, 2012 1:27 pm
Posts: 28
Reply with quote
Ben wrote:
The mirror/lens approach is definitely worth looking at if you're raytracing, but if not it'll still need to be done with a fragment shader, essentially giving you a less efficient version of the current method. I haven't tried using a geometry shader yet, but I expect it to require such dense meshes that speed would prevent it from being usable. Some of the stuff coming out of Siggraph in a couple of months may change that.

You could use a tesselation shader, like the one described here. The problem with that particular implementation is that it uses the same amount of tesselation for each triangle. I suspect you could make it adaptive by setting the tesselation factor for a triangle to a multiple of the area of the symmetric difference between the undistorted triangle and the triangle formed by distorting the original triangle's vertices, but I haven't tried it yet.


Sat May 11, 2013 1:45 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Jun 08, 2012 8:18 pm
Posts: 1329
Reply with quote
While I know a significant cause of motion blur is due to low pixel switching times, I'm curious if there's a noticeable difference between V-sync enabled/disabled on a beefy rig for those people that seem particularly sensitive to the blurring.

I also wonder if problems could be further mitigated by using predictive tracking and skewing the image in the frame buffer relative to the angular rate sensed by the head tracker.


Sat May 11, 2013 2:57 pm
Profile
Two Eyed Hopeful

Joined: Mon Aug 20, 2012 3:49 am
Posts: 89
Reply with quote
i would like to see it recorded on high speed camera just to make sure it is pixel switching. i would also like to confirm whether the sdk is adding the blur artificially to prevent another artifact.

the blur is so bad it seems almost deliberate. and i agree. it made me want to not look around because of the bad feeling it gave me.


Sat May 11, 2013 6:22 pm
Profile
Golden Eyed Wiseman! (or woman!)

Joined: Fri Jun 08, 2012 8:18 pm
Posts: 1329
Reply with quote
LukePoga wrote:
i would like to see it recorded on high speed camera just to make sure it is pixel switching. i would also like to confirm whether the sdk is adding the blur artificially to prevent another artifact.

the blur is so bad it seems almost deliberate. and i agree. it made me want to not look around because of the bad feeling it gave me.



I haven't looked at the SDK, but I've heard there is no deliberately implemented motion blur.


Sat May 11, 2013 6:31 pm
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 01, 2012 10:47 pm
Posts: 2708
Reply with quote
MSat wrote:
LukePoga wrote:
i would like to see it recorded on high speed camera just to make sure it is pixel switching. i would also like to confirm whether the sdk is adding the blur artificially to prevent another artifact.

the blur is so bad it seems almost deliberate. and i agree. it made me want to not look around because of the bad feeling it gave me.
I haven't looked at the SDK, but I've heard there is no deliberately implemented motion blur.
John Carmack recommended ADDING software motion blur to minimize stroboscopic motion ghosts, similar to how animated motion pictures pay a premium in render time to ADD motion blur. This reduces the flicker effect from stroboscopic ghosting (individual frames perceived at different points in space while your head is moving). Blurring those ghosts together can be much less annoying, but it seems to be a personal preference. You can still see the ghost images, but less annoying than if they appeared sharp and distinct. I like the build-in motion blur, despite its source (hardware or software).

Motion blur can also be selectively enabled in some games and/or video drivers, and in fact was a selling point for GPUs at one point. People used to pay extra just so they could do motion blur, and it is still used a lot in the motion picture industry. This argument reminds me about how some people argued vehemently AGAINST color TV when it first came out, preferring black and white over color. But then, color TVs still had round picture tubes, and sacrificed a lot of the peripheral display, while black and white tubes had nearly rectangular display surfaces.

Progress will continue to move forward despite these objections, and motion blur was added along the line of progression, and is here to stay (whether from a hardware artifact or a software feature).

_________________
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License. Image


Sat May 11, 2013 7:18 pm
Profile
Two Eyed Hopeful

Joined: Mon Aug 20, 2012 3:49 am
Posts: 89
Reply with quote
whatever is happening it doesn't look the same as good motion blur. its a blur but thats about as close to what you are describing gm. many games have motion blur, but it doesnt look like that.


Sat May 11, 2013 7:38 pm
Profile
One Eyed Hopeful

Joined: Sat May 11, 2013 8:08 pm
Posts: 2
Reply with quote
I spent about half an hour yesterday trying to find pixel switching times for some lcds and I couldn't even find one. Funny how you can find draw latency time on lcd monitors and tvs, but not for tablets or phones.


Sat May 11, 2013 8:14 pm
Profile
One Eyed Hopeful

Joined: Sat Sep 01, 2012 6:30 pm
Posts: 38
Reply with quote
I would like to hear something about the variability of the panels themselves. I tried the prototype 5.6in rift at SXSW and thought the motion blur was quite bad. I was very surprised to see how little blur there was on my devkit, I was expecting much more. Its not bad on my unit at all, in fact its not very noticeable at all unless there is something very bright in front of a very dark field, then its obvious but far from unusable. I suspect there is some variability in the screens themselves. I would like some other numbers than average switching time they found when doing quality control at the factory.


Sat May 11, 2013 9:18 pm
Profile
One Eyed Hopeful

Joined: Sun Apr 07, 2013 3:15 pm
Posts: 24
Reply with quote
What do you guys think about in addition to a faster refresh rate, adding a "smoothing effect" to the panel? Kind of how newer LCD's have an option to enable smoothing. It kind of makes movies shot at 24fps look like soap operas, or almost hyper real.


Sun May 12, 2013 1:32 am
Profile
Petrif-Eyed
User avatar

Joined: Sat Sep 17, 2011 9:23 pm
Posts: 2220
Location: Menlo Park, CA
Reply with quote
Syke408 wrote:
What do you guys think about in addition to a faster refresh rate, adding a "smoothing effect" to the panel? Kind of how newer LCD's have an option to enable smoothing. It kind of makes movies shot at 24fps look like soap operas, or almost hyper real.


That inter-frame interpolation ("smoothing") effect is only good for bringing low frame rate content up to the refresh rate of the panel, but at the expense of additional overhead, latency, and artifacting. You are much better off just rendering at the refresh rate of the panel.

Maybe if you had some hypothetical panel that refreshed at 120Hz or 240Hz and were unable to push those frames due to card and I/O limitations, then it's possible that on-board interpolation might be justifiable. But you would need to weigh the benefits of smoother animation versus added latency.


Sun May 12, 2013 1:42 am
Profile
One Eyed Hopeful

Joined: Thu Oct 18, 2012 10:45 am
Posts: 29
Location: Lithuania
Reply with quote
geekmaster wrote:
People used to pay extra just so they could do motion blur, and it is still used a lot in the motion picture industry.


it was chiefly used to blur fast moving objects such as race cars zooming by, don't
remember any game that deliberately blurred the entire scene when user pans around,
unless to simulate player's death or damage points

I still chase bots in Quake3 semi-regularly - there's a lot of fast panning to locate
the enemy, it is smooth and very crisp even on my 60Hz Dell, bought Razer mouse
which further improved smoothness (took away jerkyness of panning). Game would
become unplayable quick if the blur was added to the user's panning action

Just did a fast walk around my apartment spinning my head around while i walk,
i do perceive blur during very jerky head-movement, but otherwise i see
none at all, so I guess that's what my brains expect from Rift

How heavy blur in Rift could be likable is beyond me, it's single biggest offender in my otherwise
breathtaking Rift-experience


Last edited by VRus on Sun May 12, 2013 2:22 am, edited 1 time in total.



Sun May 12, 2013 2:11 am
Profile
Two Eyed Hopeful

Joined: Fri Aug 17, 2012 12:51 am
Posts: 97
Reply with quote
stinkvis wrote:
You could use a tesselation shader, like the one described here. The problem with that particular implementation is that it uses the same amount of tesselation for each triangle. I suspect you could make it adaptive by setting the tesselation factor for a triangle to a multiple of the area of the symmetric difference between the undistorted triangle and the triangle formed by distorting the original triangle's vertices, but I haven't tried it yet.
Thanks for the link. I'll probably experiment with the code there to get some image quality comparisons. I think speed would still be an issue even with adaptive tessellation. In my experience rasterization and shading are just too slow when dealing with a whole screen of pixel sized triangles. I haven't had the opportunity to try it on the latest and greatest graphics cards though.

MSat wrote:
I also wonder if problems could be further mitigated by using predictive tracking and skewing the image in the frame buffer relative to the angular rate sensed by the head tracker.
The screen would still only be updating 60 times a second, so the temporal aliasing which is perceived as motion blur would remain. Perceived is the important word there. If you load up a demo and compare the amount of motion blur when you move your head to the amount when you keep your head still and move the mouse, you should see considerably less with the mouse.

The image you see when you move your head isn't actually more blurry, but the act of moving your head causes your eyes to counter-rotate at the same speed. It's an automatic reflex, and it's actually to prevent you from perceiving motion blur in reality. The problem is that the view your eyes get from the screen in the Rift doesn't act like reality. It only updates 60 times a second, so even if the pixels switched instantly they'd still only be 'correct' at the moment they updated.

For the 16ms between that time and the next update the image would be moving with your head instead of smoothly transitioning from where it was to where it will be. After 16ms a new image would appear in the correct position, but because our retinas take a fraction of a second to 'update' we'd perceive the new and old frames at the same time, giving a stroboscopic-like effect, appearing as a trail of 'ghosts'. Technically we're already seeing this, it's just that the pixels don't switch instantly so the trail appears blurred instead of a sharp set of duplicate images. I actually prefer the blur to more well defined ghosts, but I don't expect everyone to agree. Ideally we wouldn't need to have either.

So, how do we get rid of both? I don't know. It's a tough problem. Faster pixel switching would reduce motion blur, but increase the visibility of the duplicate images. A panel capable of updating 120 times a second would help with both, I think. But it would also require a computer capable of actually rendering content at that speed, and even then it wouldn't alleviate the problem completely. Michael Abrash said you'd need 1000-2000 frames per second to do that. I'm not sure how serious he was.

I'm not a hardware person, so this is likely to be a ridiculed as impossible by those that are, but the best solution I can come up with is using motors to physically rotate the lenses (or mirrors, or move the actual panel, or something else) within the headset so that the pixels themselves counter-rotate in the same way as your eyes. If the movement was accurate, and appropriately catered for by the in-game camera, it should remove the perceived duplicates/motion blur caused by temporal aliasing despite the image itself updating considerably less frequently.

Alternatively you could just decouple the screen from the headset, enlarge it, attach a stand, and set it down on your desk. I'm not sure there's much of a market for something like that though.

LukePoga wrote:
i would also like to confirm whether the sdk is adding the blur artificially to prevent another artifact.
There's definitely no blurring happening in any of the code from the samples in the SDK. It might be worth checking to make sure it's been disabled in the UDK though.

memristor1 wrote:
I spent about half an hour yesterday trying to find pixel switching times for some lcds and I couldn't even find one. Funny how you can find draw latency time on lcd monitors and tvs, but not for tablets or phones.
If you wanted the specs for the panel in the Rift (or at least the one I've seen the insides of) it's apparently 15ms on and 20ms off. I say apparently because when I looked it up it wasn't clear where the numbers had came from. Manufacturers typically all use the same formula for their specs instead of actually measuring it, and it's pretty rare for those numbers to match up with the reality of the panel.


Sun May 12, 2013 2:22 am
Profile
Two Eyed Hopeful

Joined: Mon Aug 20, 2012 3:49 am
Posts: 89
Reply with quote
for measuring latency: you can display a timer on a crt and put the rift in front of the crt and take a photo of both to see how many ms apart the displays are.

alas i disposed of my last Sony CRT some time ago.


Sun May 12, 2013 3:17 am
Profile
Two Eyed Hopeful
User avatar

Joined: Wed Sep 05, 2012 5:04 pm
Posts: 69
Reply with quote
Is "pixel switching time" the same thing as "response time" ? Most panel specs only mention the response time :?: :

Prototype 5.6 inch Rift :
http://www.panelook.com/HV056WX1-100_HYDIS_5.6_LCM_overview_3594.html

Response Rime : 30ms (Typ.)(Tr+Td)

Rift DevKit (panel from the same series, should be very similar) :
http://www.panelook.com/HJ070IA-02C_Chimei%20Innolux_7.0_LCM_overview_16677.html

Response Rime : 15/20ms (Typ.)(Tr/Td)


So the panel in the DevKit is indeed better in almost every way compared with the display of the 5.6 inch prototype. Response time of other displays :

iPhone 5 : 23 ms
Lumia 920 : 9 ms
Samsung Galaxy SIV : 0.01 ms :shock:

Ideally the consumer Rift would have a 120hz panel with a very low response time to tackle the motion blur. But even then you would still not have solved the "motion judder" problem Michael Abrash was talking about. This would only go away at enormous frame rates like 1000/2000 FPS. Or by using other tricks that would also come with a trade-off.


Sun May 12, 2013 5:52 am
Profile
Display posts from previous:  Sort by  
Reply to topic   [ 96 posts ]  Go to page 1, 2, 3  Next

Who is online

Users browsing this forum: Tbot and 3 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Jump to:  
Powered by phpBB® Forum Software © phpBB Group
Designed by STSoftware.