[HN Gopher] Blur Busters Open Source Display Initiative - Refres...
___________________________________________________________________
Blur Busters Open Source Display Initiative - Refresh Cycle Shaders
Author : NavinF
Score : 77 points
Date : 2025-01-05 04:09 UTC (18 hours ago)
(HTM) web link (blurbusters.com)
(TXT) w3m dump (blurbusters.com)
| JonathanFly wrote:
| So this a new method that simulates a CRT and genuinely reduces
| motion blur on any type of higher framerate displays, starting a
| 120hz. But it doesn't dim the image like black frame insertion
| which is the only current method that comes close to the clarity
| of a CRT. But it also simulates other aspects of CRT displays,
| right?
|
| Can you use this method just to reduce blur without reducing
| brightness, on any game? They mention reducing blur for many
| things other than retro games in "Possible Use Cases of Refresh
| Cycle Shaders" but does reducing blur in a flight simulator also
| make it visually look like a CRT with phosphors?
| delusional wrote:
| They do mention that it does reduce brightness. The selling
| point compared to strobing sounds to be less eyestrain. I'd
| expect it to lose more brightness than strobing, considering
| the lower relative pixel on time.
| vlovich123 wrote:
| Does this mean that the original duck hunt gun might work again?
| AnthonBerg wrote:
| There's a really interesting discussion of precisely this in
| the comments under the article! Recommended. Might have to dig
| to see it.
| vlovich123 wrote:
| I don't see any comments under the article. Maybe have to be
| logged in?
| paulbgd wrote:
| It's on the article before this one. The tldr is that no,
| this doesn't reduce latency so there's no chance of making
| the original light guns work without modifying either them
| or the game.
| rasz wrote:
| If this is proven to work for 99% people and high refresh
| displays become cheap GPUs could optimize for rendering just a
| sliver of screen at a time.
| stevage wrote:
| I do not understand at all what this is talking about or why. Is
| it some elaborate joke?
|
| Don't visual effects people go to lots of effort to add motion
| blur? Why would you want to remove it?
|
| Why are they trying to simulate old CRT displays?
|
| Can someone explain what this is about?
| fishermanbill wrote:
| Yeah they are two different effects. Theres motion blur on
| individual objects that you want (as human eyes see/have) then
| there is full screen motion blur that is due to the display
| technology (lcd,oled etc) that you dont want (as human eyes
| dont see/have). CRTs dont have this motion blur as the screen
| is blank most of the time - see slo mo guys on youtube for crt
| displays.
| Springtime wrote:
| This is about improving motion clarity, so each displayed frame
| of moving content looks crisp rather than having blur
| (something that monitors can struggle with even at high refresh
| rates / high Hz).
|
| Most good monitor reviews of high Hz displays (eg: 120Hz+) take
| fast photographs of moving objects (typically from Blur
| Busters' 'Test UFO' web page) to demonstrate how good or poorly
| a monitor handles fast moving content.
|
| One technique of significantly improving motion clarity is
| inserting frames of pure black in the display output (aka BFI,
| black frame insertion). A downside is some are sensitive to
| this where it causes eyestrain.
|
| This CRT beam simulating shader is said to be similarly
| effective to BFI at improving motion clarity but with the
| benefit of reducing eyestrain. However from what I understand
| the current version is limited to simulating a lower Hz display
| and requires a higher Hz monitor.
|
| All this is distinct from the kind of in-media motion blur that
| can be enabled in games or seen in recorded video. It's instead
| about the monitor not being able to render fast moving content
| clearly enough which leads to non-crisp output frames.
| stevage wrote:
| Thank you, that's a really great explanation.
| noduerme wrote:
| What is the method used on newer TVs that attempts to double
| the framerate / interpolate frames / make everything shot on
| film look like an overlit soap opera? I find it impossible to
| watch; it destroys the lighting and the performances. My
| recollection of CRT TVs was that they had a lot of blur, both
| motion and spatial, and that was kind of what made them feel
| warmer and more analog / less overly crispy.
| Springtime wrote:
| That's typically called 'motion smoothing' and yeah that's
| trying to interpolate frames to manipulate lower framerate
| video (like 24FPS) into higher framerates in an attempt to
| make scenes like panning shots 'smoother' at the expense of
| a soap opera feel and interpolation artifacting.
|
| Whereas what Blur Busters (and similar enthusiasts) are
| focused on is how accurately frames are (perceptibly)
| displayed on the screen, so ideally each input frame is
| perfectly presented without any interference from prior
| frames (due to limits of panels in keeping up with changing
| the pixels from one frame to another, very rapidly, causing
| blurring).
|
| The ultimate goal, in a perfect scenario, is for input from
| say a video game running at 60 frames per second to have
| each frame perfectly rendered like individual screenshot
| stills, one after the other. In reality for most monitors
| displaying such content there's not enough distinct
| separation between frames, leading fast changing pixel
| content (like objects moving) to blend into each other,
| causing blurring at the monitor level.
|
| The BFI technique, by inserting alternating black frames in
| the output, mitigates the inter-frame blending issues since
| instead of the prior frame being various colors (ie: of the
| prior input frame) it's starting from pure black which
| dramatically reduces frame blending artifacts and
| perceptibly makes the motion clarity more distinct.
| CuriousSkeptic wrote:
| It's not that the frames blend in the screen. Screens are
| perfectly capable of switching the pixels fast enough.
| It's rather that each frame is displayed for to long.
|
| In a CRT the "pixels" start to fade immediately leaving
| the full screen mostly dark as the beam sweeps over the
| screen. It never shows a full frame.
|
| One could say that modern screens are more like slide
| shows, while BFI tries to make them more like
| stroboscopes.
|
| The blurring effect is more pronounced ate low refresh
| rates, its just that BFI requires at least 120hz to male
| sense at all.
| fulafel wrote:
| CRTs don't darken that fast, one way to observe this is
| that CRTs don't appear black in photos/video with shutter
| times << 1/60
| yuriks wrote:
| They do darken that fast (not fast enough you can't catch
| it in a high speed camera, but much faster than a frame).
| Most of the apparent persistence in the CRT comes from
| the retina/camera exposure, not the phosphor. A CRT has a
| sharp peak of light that quickly falls off, but the peak
| is bright enough that even though it is brief, when
| averaged out in an exposure in the camera it still
| appears bright enough to form an image.
| chowells wrote:
| You've never taken photos of a CRT, have you? Even at
| like 400 ISO equivalent, only about a third of the screen
| is illuminated.
| tigen wrote:
| Yes, the relevant blur here is in your retina, as it
| tracks a moving screen object, called "sample and hold"
| blur. 60 fps is not enough when the pixel persists for
| the full frame duration -- the pixels smear across your
| retina.
| thfuran wrote:
| >at the expense of a soap opera feel
|
| The "soap opera feel" is precisely the goal of motion
| interpolation on 24 fps source. It reminds people of soap
| operas because they were often broadcast 60i instead of
| 24p. The weird part is that many people somehow prefer
| the terrible 24 fps to higher film frame rates.
| 7734128 wrote:
| Because I hate it.
| haunter wrote:
| First thing I turn off in every single game is motion blur.
| It's only useful in racing sims to have more sense of speed but
| that's also a personal taste.
|
| Motion blur made a bit more sense on the 30fps Xbox 360 and PS3
| games.
| martini333 wrote:
| Why exactly do you think motion blur is added?
| cubefox wrote:
| Our eyes are constantly and mostly unconsciously tracking
| moving objects in our field of view in order to keep them still
| relative to our eyes. It's called Smooth pursuit:
| https://en.wikipedia.org/wiki/Smooth_pursuit
|
| This is because our retina has a very low "refresh rate", which
| means things can easily blur together. Smooth pursuit prevents
| that. However, modern sample-and-hold displays like LCD and
| OLED work against Smooth pursuit. If you watch anything moving
| on a screen (including "still" objects moving on screen due to
| camera movement), your eye will automatically track those
| objects if they are momentarily the focus of attention, which
| should make them be still relative to your eyes and thus appear
| sharp.
|
| However, since the tracked object is being still relative to
| your eyes and the individual frames on screen are being still
| relative to your screen, the frames _move_ (are not being
| still) relative to your eyes. Which means they appear blurry
| during smooth pursuit, when in reality they should be perfectly
| sharp.
|
| For example, your eyes track a sign that moves on the screen
| due to camera movement. Say it moves 10 pixels per frame
| horizontally. This means you will see a 10 pixel wide
| horizontal blur on this sign. Which could make it unreadable.
| In reality (without screen with a real sign) the sign would
| appear perfectly clear.
|
| On CRT screens this doesn't happen (to the same extent) because
| the frame is not displayed for the entire frame time (e.g.
| 1/60th of a second) but much shorter. The CRT just very quickly
| flashes the frames and is dark in between. Strobing/flickering
| basically. So if the tracked object moves 10 pixels per frame,
| the frame might only be (say) visible for 1/5th of that frame
| time, which means it moves only 2 pixel while the frame is
| actually on screen. So you get only 2 pixel blur, which is much
| less.
|
| Of course at 60 FPS you might instead get some degree of
| perceptible flicker (computer CRTs therefore often ran higher
| than 60) and in general the overall achievable screen
| brightness will be darker, since the screen is black most of
| each frame time. CRTs had a low maximum brightness. But they
| had very little of the "persistence blur" which plagues sample-
| and-hold screens like OLED and LCD.
|
| The motion blur intentionally introduced by video games is
| there to make moving objects appear smoother that are _not_
| tracked by our eyes. In that case motion blur is natural (since
| smooth pursuit doesn 't try to remove it). So some forms of
| motion blur are undesirable and others are desirable.
|
| The optimal solution would be to run games (and videos content
| in general) at an extremely high frame rate (like 1000 FPS)
| which would introduce natural perceptible motion blur where it
| naturally occurs and remove it where it doesn't naturally occur
| (during smooth pursuit). But obviously that would be
| computationally an extremely inefficient way to render games.
|
| By the way, if you have a screen with 120+ Hz you can test the
| above via this black frame insertion demo, which emulates how
| CRTs work:
|
| https://testufo.com/blackframes
|
| On my 120 Hz OLED screen, the 40 FPS (1 frame + 2 black frames)
| UFO looks as clear as the native 120 Hz UFO. A real 60 or even
| 80 Hz CRT screen would be even better in terms of motion
| clarity. Perhaps better than a 240 or even 480 Hz OLED.
| saagarjha wrote:
| Past discussion: https://news.ycombinator.com/item?id=42506211
| fishermanbill wrote:
| We need display manufacturers to provide a refresh cycle that is
| agnostic of the incoming signal hz sent down the cable AND to
| either provide shader support (ideally) at the displays hz OR to
| implement this shader.
|
| There really is no need for an expensive RetroTink if we had
| this. Some manufacturer must be able to do it and the rest would
| follow.
| nopurpose wrote:
| With about half of the screen is black, can it also boost FPS by
| not spending GPU time on pixels in those areas if integrated deep
| into engine?
| kevingadd wrote:
| You could definitely do this, but a lot of modern rendering
| techniques rely on having full copies of previous frames lying
| around, like TXAA, screen-space reflections, etc.
| rzzzt wrote:
| The images are briefly persisted and averaged in the back of
| the viewer's eye.
| Scene_Cast2 wrote:
| I don't think this is how it works.
|
| The technique is for when you have X fps content and Y fps
| capable monitor, where Y > X. In games, you'll still render at
| your old FPS cap, but this shader is for relatively cheaply
| generating extra frames that will make the content look
| smoother / sharper.
| fishermanbill wrote:
| Does all this shader really get around is the problem of display
| inversion?
|
| From Gemini: "Display inversion is the process of alternating the
| voltage between positive and negative for each pixel on an LCD
| screen to prevent damage. This process is called polarity
| inversion."
|
| If display manufacturers knocked that on the head for certain
| scenarios then surely we could just have a simple block of
| horizontal screen scrolling down the display at high refresh
| rates?
|
| Phosphor fall off as far as can be seen in Slo Mo Guys is quite a
| small effect not on the scale of this shader.
| parski wrote:
| I'd buy a new TV or monitor for this feature alone.
| P_I_Staker wrote:
| Cycle refresh shaders where someting my last team really nailed.
| The key challenge was during the day there is a lot of sun.
| Adjustments can be made to the location, it really pays
| dividents.
___________________________________________________________________
(page generated 2025-01-05 23:01 UTC)