That was a mechanical limitation. There are alternative options now for longer exposure like Tessive Time Filter. (thanks Tritosine5g for the Pm a long time ago)
The Great Simulated Optics Debate *spawn
- Thread starter L. Scofield
- Start date
Was going through the garage last night to take out the trash and as I passed the car, I thought...
How can a game, even with a futuristic eye tracking mechanism, determine if I am looking at the reflection of myself in the passenger window or looking through the window into the back seat?
Makes me believe that blurring should be limited to objects that move across/around the screen above a threshold speed and leave everything else to your vision's natural tendency to blur the periphery.
Blurring doesn't have to be perfect just good enough.
How can a game, even with a futuristic eye tracking mechanism, determine if I am looking at the reflection of myself in the passenger window or looking through the window into the back seat?
Makes me believe that blurring should be limited to objects that move across/around the screen above a threshold speed and leave everything else to your vision's natural tendency to blur the periphery.
Blurring doesn't have to be perfect just good enough.
Crysis 1 at 30fps in DX10 mode with full motion blur looks comparable to 60fps without MB. In some cases I prefer 30fps + MB over 60fps without. Of course 60fps with MB is ideal and looks to my very good human eyes like real life does in terms of temporal resolution on most cases. I think that's one of the reasons why some folks including myself still think Crysis is one of if not the best looking game of all time.
I understand it quite well since I already covered it prior to you repeating it...
The discussion was about eye tracking and whether or not you need it to simulate realistic motion blur as it applies to GAMES. In a game that does not require fast movement of the EYES you don't need eye tracking to mimic realistic MB. As an example...take a game where you control a character from a third person perspective compared to a character from a first person perspective. The former does not require eye movement based blur therefore does not require eye tracking while the latter does.
As for your static vs moving example...it's totally flawed and has nothing to do with motion or lack thereof. What you're describing is called focus. If I'm looking at object A which is static, it doesn't matter if object B in the background is moving or not. A will be clear and B will be blurred regardless. Additionally since B is blurred regardless whether or not it's moving you cannot say the blur is motion based.
Last edited by a moderator:
Eye tracking isn't fast. Fast eye movement is saccadic during which you are blind. Slow moving objects being tracked require motion blur applied to anything not on the same motion vector.
Take GTA. Pull up in a car at a junction as per my above real-life examples. If you look at the traffic lights, the cars moving across your FOV should have moblur applied. If you then switch to track a car pulling across your FOV, the traffic lights should have motion blur applied. At 60 fps and a moderate viewing distance, the absence of motion blur would probably make no difference, but scientifically it needs to be present to be accurate. Replace the moderate viewing distance with sitting close to a large screen, angular motion increases and the need for motion blur will increase. More so with lower framerates.
In real life when I'm at an intersection staring at the light, ALL cars get blurred not just cars that are driving by...this is due to peripheral blurring and or focus blurring. Since this is the case the fact there is also MB blur on the cars that are moving fast doesn't really change what you actually see.
It's like trying to read any eye chart that is moving AND out of focus.
Also I didn't say eye tracking was fast. I said in order to simulate MB caused by fast eye movement you would need eye tracking...because if the eyes were moving at the same speed as the object being observed by said eyes there would be no MB...so you would need to track the speed of the eyes relative to the speed of the object the eyes are focused on to do "realistic" MB.
It's like trying to read any eye chart that is moving AND out of focus.
Also I didn't say eye tracking was fast. I said in order to simulate MB caused by fast eye movement you would need eye tracking...because if the eyes were moving at the same speed as the object being observed by said eyes there would be no MB...so you would need to track the speed of the eyes relative to the speed of the object the eyes are focused on to do "realistic" MB.
Last edited by a moderator:
MB and DOF blur are two different optical phenomena. Also DOF and peripheral vision blur within the local area of the fovea isn't all that significant when focussing far enough away. I cited the intersection example because it's noticeable exactly - traffic lights are clearly blurred in a motion vector when tracking a moving car with a luminescent streak, and likewise the cars are clearly blurred along the motion vector. DOF doesn't make any difference when focussed to 15-20 feet out swapping between cars and lights. Peripheral vision isn't so lacking in detail that you can't notice the motion-based blurring. eg. I can only see the letters I'm immediately typing/reading when reading, but the other letters and clearly letters/symbols, even quite a number of degrees from centre. I can see a blue circular light in the bottom corner around my monitor's power button. If this light were to suddenly escape and run off left, I'd see a blue streak.
Fast eye movement does not generate motion blur - you are blind during fast eye movement. Slow eye movement requires motion blur.
There would be no MB on the object being tracked, but there would its immediate surroundings and the rest of the screen.
To do MB correctly, as possible with discrete time displays, you'd need to track what part of the screen the player is looking at. You would then need to get the motion vector of the surface under that point (triangulating the two eyes should yield focus distance and enable determination of reflective surfaces for an ideal case, but that's clearly overkill!) and apply a corresponding relative inverse motion blur to every other part of the screen based on distance to the camera and relative velocity to the motion vector of the object being viewed.
Actually, you'd need to track the movement of the eye and match the motion vectors there, as it's possible to look at a point and not track it, and have that point motion blurred. So you'd need to track the eyes, determine motion vector, and apply the appropriate motion blur to all objects in the scene. Those moving along the same vector will have no motion blur whether being tracked or not, which would be correct, and of course tracked objects would have a corresponding vector so be unblurred.
That is fantastic!
by mesuring how cross eyed the user is.
about my blurry eye, microsoft have good explanation here
http://research.microsoft.com/apps/video/default.aspx?id=173013
basically the game dont need to render all area in high quality. Because naturally, eye cant see the whole screen clearly.
so the game load will be lower without needing to simulate blurry optics
http://research.microsoft.com/apps/video/default.aspx?id=173013
basically the game dont need to render all area in high quality. Because naturally, eye cant see the whole screen clearly.
so the game load will be lower without needing to simulate blurry optics
That's something that never occurred to me but makes much sense. Something tells me I would just disable this and buy a faster GPU if necessary but it could make sense for fixed hardware or people who don't care 
.
Also makes more and more sense as pixel density increases. At 4k on a normal sized monitor the difference would likely be nigh imperceptible but the performance savings would be tremendous. Just have to make sure aliasing is kept in check since that's something were are very good at detecting in the periphery.
.Also makes more and more sense as pixel density increases. At 4k on a normal sized monitor the difference would likely be nigh imperceptible but the performance savings would be tremendous. Just have to make sure aliasing is kept in check since that's something were are very good at detecting in the periphery.
Taken from Destiny Beta Eurogamer's article:
Again, VR can't come soon enough! It's like developers are afraid of realism or think their games won't sell if they are not cinematic enough.
Did people buy TLOU, or give the game awards, because of the excessive motion blur the game had?
Has the lack of motion blur in GTA5 prevented people buying the game? Quite the contrary because the lack of focus on costly GPU cinematic effects probably helped the developers to have better texture filtering, effects, framerate and more NPCs in the world they created.
The most commercially successful game of the decade indeed doesn't even use motion blur during gameplay! Taken from Eurogamer GTA5 face-off:
They even reduced DOF compared to GTA4, again, probably to better use GPU ressources for instance with decent non-blurring post AA.
For instance in Destiny if they remove the motion blur, the excessive chromatic aberrations, lens flares, all the cinematic effects: could they replace FXAA by SMAA?
Again, VR can't come soon enough! It's like developers are afraid of realism or think their games won't sell if they are not cinematic enough.
Did people buy TLOU, or give the game awards, because of the excessive motion blur the game had?
Has the lack of motion blur in GTA5 prevented people buying the game? Quite the contrary because the lack of focus on costly GPU cinematic effects probably helped the developers to have better texture filtering, effects, framerate and more NPCs in the world they created.
The most commercially successful game of the decade indeed doesn't even use motion blur during gameplay! Taken from Eurogamer GTA5 face-off:
They even reduced DOF compared to GTA4, again, probably to better use GPU ressources for instance with decent non-blurring post AA.
For instance in Destiny if they remove the motion blur, the excessive chromatic aberrations, lens flares, all the cinematic effects: could they replace FXAA by SMAA?
Taken from NeoGaf thread about Destiny DF article, It's about the useless* chromatic aberration used in the PS4 version, which apparently won't be used in the 1080p XB1 version (making the XB1 defacto the best version):
Remember when I was posting somewhere in this forum that giving more GPU ressources to devs can have detrimental effect on the image quality.
Similar to MGS5 Ground zeroes without motion blur but with "room to spare" VS MGS5 TPP + heavy motion blur using the "room to spare" GPU ressources.
What will they do to the image quality when they'll have 20 Tflops machine at their disposal? I wonder.
useless* -> I can now safely use that word because if it's not used in the Xbox 1 version, they indirectly confirm us it's not an essential feature of the game.
Remember when I was posting somewhere in this forum that giving more GPU ressources to devs can have detrimental effect on the image quality.
Similar to MGS5 Ground zeroes without motion blur but with "room to spare" VS MGS5 TPP + heavy motion blur using the "room to spare" GPU ressources.
What will they do to the image quality when they'll have 20 Tflops machine at their disposal? I wonder.
useless* -> I can now safely use that word because if it's not used in the Xbox 1 version, they indirectly confirm us it's not an essential feature of the game.
well, the Xbox One version only runs at 720P so when playing the Xbox One version on a 1080P tv you, in a way, have the simulated optics of having a minus .75 dioptre eye handicap
motion blur in 60fps games is wrong!
After comparing extensively two 60fps games, one with motion blur, TLOUR, and one without motion blur, Metro 2033 redux, I have come to the conclusion that, on my TV at least, motion blur only brings a depth of field blurry effect when you span the camera. It doesn't bring any smoother motion, even perceptual.
I can understand that, at 30fps, motion blur can bring some illusory smoother motion (at the cost of image clarity). But not at 60fps, definitely. The motion of Metro 2033 without motion blur is even smoother in my book (better clarity) than the motion of TLOUR.
Just a reminder: Did someone ever complain that Metro Redux was not smooth enough because it lacked full screen motion blur? Did someone ever ask for a smoother, motion blurred, MGSV ground zeroes? On the contrary, I have only seen praises about how smooth both games, without motion blur, were.
After comparing extensively two 60fps games, one with motion blur, TLOUR, and one without motion blur, Metro 2033 redux, I have come to the conclusion that, on my TV at least, motion blur only brings a depth of field blurry effect when you span the camera. It doesn't bring any smoother motion, even perceptual.
I can understand that, at 30fps, motion blur can bring some illusory smoother motion (at the cost of image clarity). But not at 60fps, definitely. The motion of Metro 2033 without motion blur is even smoother in my book (better clarity) than the motion of TLOUR.
Just a reminder: Did someone ever complain that Metro Redux was not smooth enough because it lacked full screen motion blur? Did someone ever ask for a smoother, motion blurred, MGSV ground zeroes? On the contrary, I have only seen praises about how smooth both games, without motion blur, were.
Motion blur at 60 fps can absolutely make things appear to move more smoothly. It depends on the movement of the object relative to the camera. But "smoother" is not necessarily worth the cost of reduced clarity and artificially imposed effects that can actually act against how the eye would work in real life.
It depends on whether you're trying to mimic the eye, or camera, I guess. I know that I generally prefer less motion blur rather than more.
It depends on whether you're trying to mimic the eye, or camera, I guess. I know that I generally prefer less motion blur rather than more.