IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Advanced display talk, display hackers, advanced game programmers, scientists, display researchers, display manufacturers, vision researchers & Advanced Display Articles on Blur Busters. The masters on Blur Busters.
Post Reply
thatoneguy
Posts: 181
Joined: 06 Aug 2015, 17:16

IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by thatoneguy » 05 Oct 2021, 08:15

Now I'm probably not the first one to have thought of this idea but what if instead of filming in 24fps we film at 1000fps and then add (a lot of)motion blur to simulate a 24fps look?
This could be done via digital methods in post-processing or in the source itself.

The benefit would obviously be to not have to deal with the limitations of 24fps playback.
The result = Original 24fps film look, pretty low motion blur(1ms additional blur from the display(and even less if using BFI/Strobing/Rolling Scan) + Film Motion Blur) , no more double/triple image effect(due to the fps and hz being matched).

Of course this wouldn't be feasible until 1000hz displays become a standard which aren't even a standard at cinemas and I doubt they would want to go through the hassle of manually converting 1000fps movies to ~24-60fps for home releases(if that would be even possible).
Then there's other things to worry about such as broadcast or home release(Blu-ray) and HDMI/Displayport bandwidth.
And of course there's the logistics of shooting 1000fps film(it's not easy and it could be pretty costly).

Regardless I think this is an idea worth thinking about for the future. I've seen some film enthusiasts gloom and doom about HFR coming to Cinema in the future but I think it could potentially be used to make 24fps a better experience(that is of course if the industry wants to use it that way instead of pushing pure HFR aggressively and deprecating 24fps entirely(i.e much like what happened to traditional film vs digital cinematography)).

EDIT: Maybe I should have posted this on the HFR section instead of Area 51. Feel free to move the thread if you deem it necessary.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by Chief Blur Buster » 07 Oct 2021, 15:50

True -- Valid Idea for Sample and Hold Displays

Yes, I already mention this in some Blur Busters pages:

This issue is already partially covered at Stroboscopic Effect of Finite Frame Rates as well as UItra High Frame Rate (HFR) video. For Ultra HFR 360-degree shutter I recommend bakes the motion blur into each frame. Both of these articles are part of the purple "Research" section at top of www.blurbusters.com

Now, 1000fps is not the final frontier, for say, a desktop monitor.
Ideally for the vanishing point of diminishing curve of returns -- you want to oversample temporally.
Say, 4000fps + baking in 1/4000sec motion blur.
= blurfree (even for ultrafast motion)
= strobscopics free (even for ultrafast motion)
= behaves like per-pixel variable refresh even for framerate drops (e.g. 4000-5000fps range).

Now, this also applies to video games, you do have to bake the GPU motion blur into each frame. The game of a retina refresh rate is make things look perfectly analog motion without a single hint of visible stroboscopics and visible motion blur.

Due to temporal aliasing factors (a temporal version of nyquist factor), you need to oversample the frame rate and refresh rate to compensate for two things (A) eliminating stutter; and (B) the extra blur added by GPU motion blurring needs to stay below human detectable thresholds.

So you need about a 2x-4x oversampling of what is thought as a traditional retina refresh rate. So for what was thought as 1000Hz retina refresh rate, you need 4000Hz if you're wanting to make it behave like per-pixel VRR (add a 2x oversample) and also add GPU motion blur (add a 2x oversample), so it is possible we need to oversample the retina refresh rate by approximately 4x to avoid a lot of side effects of various mitigations.

For framerate=Hz perfect synchronized motion that also has GPU motion blur added below human visible noisefloor, the retina refresh rate essentially depends roughly on screen FOV for any given spatially-retina angular resolution:

These are rough estimates on latest scientist/researcher experiments:
- Approximately 500Hz-1000Hz for a retina-resolution smartphopne
- Approximately 2000-4000Hz for a retina-resolution desktop monitor (e.g. 4K 24")
- Approximately 10,000Hz-20,000Hz for 180-degree 16K retina VR headset.

Interesting Approximate Guideline Of Estimating "Retina" Refresh Rate
Vanishing point of diminishing curve of returns is dictated by human eye tracking capabilities -- the screen resolution along the long axis of the screen tends to roughly resemble the retina refresh rate, because it's the fastest motion speed in pixels per second that still stays on the screen for 1 second (long enough to identify 1 pixel of motion blurring from sample and hold refresh rate limitation). As long as the resolution is not far beyond human vision acuity (aka retina resolution). Usually that is somewhat oversampled already, so that already gives an approximate defacto oversample of retina refresh rate.

This is because higher resolutions and wider FOV amplify refresh rate limitations, due to the Vicious Cycle Effect section in the 1000Hz article on Blur Busters. Eventually it maxes out at the threshold of retina resolution -- so you can only increase FOV in order to increase the retina refresh rate, because you need enough time for a human to track eyes on moving objects, in order to discover motion is stepped/strobscopic'd or blurred instead of stepless/blurless.

For other readers -- don't understand why we need 10,000 Hz in certain cases? Even at 10,000 Hz, moving a mouse at 40,000 pixels per second (fast swipe of mouse at high pointer velocities on a 8K display), can still create a stepping effect. That's why motion blur needs to be pre-baked to eliminate stroboscopics.

Image

This sort of effect also becomes visible in games (see www.blurbusters.com/stroboscopics ...) and some humans can detact high stroboscopic thresholds, including from past lighting studies (e.g. PWM/AC-driven lighting).

Image

Stroboscopic stepping effects and wagonwheel effects are still visible at ultra high refresh rates, e.g. Motion moving 3000 pixels per second at 1000Hz still shows stroboscopic stepping of 3 pixels, unless you intentionally add 1ms of GPU motion blur (3 pixels of motion blur for 3000 pixels/sec computer generated material) or intentionally add 1/1000sec camera shutter (3 millimeters of motion blur for 3000 millimeters/sec motion).

Adding baked motion blur fixes the stroboscopics. But also slightly raises the retina refresh rate due to the intentional blur added. It's a temporal equivalent of spatial antialiasing, essentially. Now, you will want to oversample it by about 2x to get beyond various nyquist factors (antialiasing can still have side effects such as beading along slightly diagonal lines -- www.testufo.com/aliasing-visibility ...) so you gotta oversample somewhat and improve the antialiasing. Retina refresh rate techniques is a temporal refresh rate equivalent of of the spatial antialiasing techniques. And you just described one important technique.

So instead of 1000Hz, you may need 2000fps+2000Hz with baked blur to get the same amount motion clarity as 1000fps+1000Hz non-baked-blur. Oversampling becomes needed.

Still, for people sensitive to stroboscopics, adding GPU motion blur is a recommended technique once refresh rates are sufficiently high enough that the added GPU blur is practically below human detection threshold. Think about it -- at 10,000Hz, you're only adding 0.1ms of GPU blurring or camera shutter blurring (1/10,000sec of intentionally added motion blur) -- at that point, GPU motion blur isn't evil anymore, but solves a stroboscopics problem to make motion look completely analog framerateless exactly like real life. Very useful for VR and simulation. No display blur, no stroboscopics. Retina resolution, retina refresh rate, and both of them sufficiently oversampled, so you have simultaneous perfect blur-free AND strobscopics-free.

Newcomers seeing this thread, jawdropped at high retina refresh rates, need to read textbook material at www.blurbusters.com/area51 and know that I'm already credited in more than 20 peer reviewed research papers.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

thatoneguy
Posts: 181
Joined: 06 Aug 2015, 17:16

Re: IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by thatoneguy » 07 Oct 2021, 20:57

Well, when we reach 1000hz as a standard I'd imagine 2000hz wouldn't be far away but even with 1000hz you could just use cheat by using a little BFI.
This would be quite a mission to convince Hollywood and other filmmakers to adapt this approach(for many reasons).

But nonetheless the basic idea is to shoot at ridiculous HFR as a way of eliminating much of the additional display blur(with little or no flicker/strobing) and to avoid dealing with the multiple image effect.

As resolutions start to get higher and higher we'll see 8K or even 16K movies so 24fps starts to get really unrealistic at that point(320 pixels of motion blur for 8K or 640 pixels of motion blur for 16K...actually even more than that considering that cinema aspect ratios are generally wider) and even moreso when it comes with emerging technologies like OLED and MicroLED which have near-instant response time which means the frames don't smear into each-other like they do with LCD TV's.
So a method like this is gonna be needed if filmmakers want to continue shooting 24fps.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by Chief Blur Buster » 08 Oct 2021, 00:47

There's a thread here too:
Cinematography of 2030s: Ultra HFR (1000fps at 1000Hz!)

Personally, I feel 24fps will endure. There is a subsegment of population that will get motion sick from hypersmooth motion, but 1000fps is less nauseating than 48fps and 120fps. The good news is is easily temporally

However, 48fps, 60fps and 120fps HFR will be obsoleted by 1000fps UltraHFR.

So there would be two standards -- 24fps and whatever Ultra HFR frame rates is made available.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

thatoneguy
Posts: 181
Joined: 06 Aug 2015, 17:16

Re: IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by thatoneguy » 09 Oct 2021, 15:20

Chief Blur Buster wrote:
08 Oct 2021, 00:47
Personally, I feel 24fps will endure. There is a subsegment of population that will get motion sick from hypersmooth motion, but 1000fps is less nauseating than 48fps and 120fps. The good news is is easily temporally

However, 48fps, 60fps and 120fps HFR will be obsoleted by 1000fps UltraHFR.

So there would be two standards -- 24fps and whatever Ultra HFR frame rates is made available.
I don't see how with the rising resolutions tbh. At some point it becomes too much.
With OLED there's plenty of people already who complain about 24fps motion being too stuttery due to near-instant pixel response time without interpolation or BFI and that's just 4K.
With the 1000fps + motion blur(to approximate a 24fps look) method you get your cake and eat it too, or at least that's the theory...and of course no double image effect, no need for pulldown methods etc.

Personally I don't understand how 48fps would be nauseating but 1000fps wouldn't be. I don't find 48fps nauseating at all though it does make movies look cheap/fake and like a soap opera or somebody recording a stage play. If 48fps had more added motion blur maybe it would look more like 24fps film?
But with 1000fps you have much more headroom for that sort of thing.

Anyhow that's my 2 cents.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: IDEA: 1000fps + Baked-in Motion Blur as a repacement for traditional 24fps cinema

Post by Chief Blur Buster » 22 Oct 2021, 10:03

We are not talking about you - but average population percentage getting motion sick at movies. Or games. Or VR. Or others. It can come from flicker, or from judder, or from blur, or from another attribute, or vertigo, or all.

Different people find motion blur nauseating. The 24fps judder compensates to ground a person but sample-and-hold 48fps and 120fps just look like blur with less or no judder to distract you from blur nausea.

People who don’t get nauseated at judder but nauseated at motion blur, are the individuals most affected by the uncanny valley of intermediate blurry sample-and-hold frame rates.

Not everyone gets motion blur nausea (but it’s the reason why VR had to go for mandatory strobing)

Fuller five-sigmaing of human population for ergonomic motion comfort on a screen for real life simulation comes blurless sample and hold - but requires insane frame rates and vertigo sync.

Even without vertigo sync (6dof VR), low persistence (blurless) sample and hold is the closest you can get to real life (flickerless, plus no extra blur above and beyond normal human vision to confuse/nauseate some brains). But blurless sample and hold mandatorily requires ultra HFR.

There is a plan allowing 8K 1000fps UltraHFR using today’s technology. Weak links were identified and the means to work around them, at least for simulator venues.

As you already know, staying ergonomically flicker-free without adding motion blur (that nauseates people) = UltraHFR mandatory.

Some of us are bothered by motion blur. Or flicker. Or stutter. Not all. But others are bothered by two or three in varying combinations.

Also some are bothered by the camera blur less than the sample-and-hold blur (since that blur varies in stationary eyes vs moving eyes as demo’d at www.testufo.com/eyetracking …. It’s subtler in cinematic material but also yet anither nausea contributor in some)

Ultra low frame rates (15fps-24fps) on sample-and-hold sometimes helps these people (as long as no stutter nausea) because ultra low frame rates veers into the stutter region of the stutter-to-blur continuum (demo’d by www.testufo.com/vrr …). This creates the 24fps-ish-and-below safe territory, whereas raising frame rates goes into motion blur territory until UltraHFR territory. Thus, the motion uncanny valley problem affecting a segment of human population (not necessarily you).

You cannot judge a human population just because of only your own eyes — many people perceive motion very differently than you, just as everyone can see colors differently (degrees of colorblindness or color ultra sensitivity etc)
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

Post Reply