- 10,938
- 12,427
Continuation of this basically.
Or revision of these standards.
Long story short: Assuming that any movement that happened faster than a human could percieve happened within 2 ms is nonsense.
I will just quote the prior thread (See the OP of that thread for the full version):
Correlating that to movement is already somewhat of a stretch. The test is about being able to tell that any change is happening, not any proper perception.
Now, in the past thread, the following things were brought up as evidence that it is valid:
As such it is more a natural high end, than a necessary low end. Something that leaves the FOV below the FFT would be invisible. However, that an object that does not do so is visible is never said.
I will demonstrate in the following that treating blitzing generally as occurring in under 2ms (1/500s), as the standard suggests, is not scientifically proven nor do they generally appear reasonable.
First, I wish to direct attention back to the primary source that the 2ms idea is based on.
None of these factors are adequately handled by the current standard, as they are not handled at all. It's a constant number of 2ms that is not modified by any of these parameters.
Consider, furthermore, that some of these factors may be influenced in a movement scenario in a way different from a flickering light test. A moving object will have its light reach different parts of the eye's retina over time. The amount of light that a stationary object would all project on one part of the eye, is instead distributed on a greater area making the effective average illumination intensity less, as the same amount of light is spread more.
Given such considerations, I am confident to say that the source in itself would likely not agree with this usage of the value.
I will furthermore point out the the 2ms value is only produced in the study, only produced such high results with a specific setup: "The high spatial frequency image is first “bright on the left half of the frame and black on the right”, and then inverted. We observed the effect described in this paper whenever we displayed an image containing an edge and its inverse in rapid succession."
I.e. the specifically inverted the colours and specifically in an uneven way. Equalizing this to real life conditions, which may have edges but no inversion of said edges, seems faulty.
In general, I feel like the translation of FFT to blitzing is not scientifically validated. Human vision is highly complex and, as one has to consider, the brain does a lot of post-processing to every image. It for example pieces together several separate quick looks into a proper picture, one of which takes about 200ms alone. Speed based illusions are for instance also not fully explained by current state of science. It is a common idea that the brain's visual system simulates events before perception which can lead to illusions in movement that last longer than the mentioned 2ms. It is known that at least under certain circumstances the brain will decide to suppress blur for a better image, although that isn't the specific case.
The point is, scientifically hand waving the equalization between flickering light and movement is highly questionable due to the complexity of the subject.
Another problem comes from a translation of the FFT to a movement scenario.
If we are having a movement, how far of a movement would have to be completed for it to equal the length of one flicker?
Until they are out of the field of view? Until they reappear in the vision? I don't think so. Reasonably speaking, as long as you do not occupy the same place for longer than a flicker you should be below the threshold, as no part of the eye sees you longer than one flicker. It would be the equivalent of the left the of the image flickering below the threshold and right after the right side below the threshold. Considering that most humans are about half a meter wide, that severely would reduce the necessary speed. (half a meter in 2ms is subsonic+)
Of course, that is my argument as I see it as reasonable, not proper science. No scientific source describes how exactly one would apply flickering to invisibility via speed exactly, so that much is a given. But such uncertainties are exactly the problem.
Logistical problems aside, I believe the current state also is in conflict with real-life observations. I will say that while I will provide some videos in the following, since cameras with their shutter speed and videos with their FPS obviously screw with the observation, they are more for illustration purposes than that they prove anything.
First, let's talk about fans.
It is documented that when fans move fast enough, they turn practically invisible to, at minimum, some people.
Why is that a problem? Well, fan blades stay constantly in the field of vision. And they pass through a spot focussed on many times a second and for fans with thick blades they also occupy each point for a significant total percentage of the time. Furthermore, they are just subsonic. A standing fan is about 1300 to 2100 round per minute, which equates to 22 to 35 rounds a second. If we assume a blade takes up 1/6th of the total circle, then the time in which it passes through one spot in 1/132s to 1/210s, which is more than twice the time we have from FFT.
This appears not in line with the standard. The section assumes that a character can not stay invisible longer than said 1/500th of a second (otherwise, it would not make sense to conclude that a character that seemingly teleports must do so in that timeframe), but this example demonstrates that one can in principle stay invisible indefinitely.
Next, let's talk about guns.
A glock has a muzzle velocity of 375 m/s. That means that in 2ms the bullet passes a distance of 75 cm. In other words, by the current standard, we would assume that if someone shoots you with a Glock in the face from an arm's length away you can see the bullet emerge from the gun.
And... that is wrong. Even from a greater distance, you don't simply see bullets. (given bullets slow down with distance and if you stand far away they appear to move slower, so I don't think it's per se impossible to see a bullet fly under the right conditions)
Now, bullets are smaller than humans, but at less than a meter away one can still clearly see them. Now, I have no doubt that size does play a role in whether or not you can see a fast object, even in the realms where you can see the object when it stands still. Consider at that, that humans are not a regular size and the apparent size of a person lowers with distance. Meaning distance is also a factor.
However, notice that the standards do not account for that in any way. In fact, the flicker test does not account for size. (And there might be a difference between the relevance of the size of a light source and size of a passively illuminated object)
And, when translating from the flicker test results to actual movement... well, there it would need to be considered as well.
In total what we see by this example is that either 1/500s is not enough or the value at least needs a component to correct for size.
Furthermore, I believe it to be plausible that such a method needs to account for the possibility of a character being invisible by speed, without leaving the field of view and for extended periods of time. A theory that does not account for that appears to be incomplete to me.
As such I propose to revert things to the prior standards on the matter.
Staff Votes:
Or revision of these standards.
Long story short: Assuming that any movement that happened faster than a human could percieve happened within 2 ms is nonsense.
The Arguments
Let's start with where the number comes from.I will just quote the prior thread (See the OP of that thread for the full version):
I.e. it comes from the timeframe in which a light can turn on and off repeatedly, while the human can still see that the light flickers.What are the speed limits of what we see?
Under normal conditions, our vision has certain limitations that can change depending on different conditions and circumstances. To understand these limitations, researchers conduct experiments under controlled conditions, often involving flickering lights.
When a light flickers very quickly, there is a minimum speed at which we no longer perceive the flickering and the light looks steady instead. This specific minimum flickering speed is called the critical flicker fusion frequency (cFFF) or Flicker Fusion Threshold (FFT). At this point, our eye receptors can't detect the individual changes in what we see anymore. Instead, they merge together, creating a steady signal that our brain interprets as continuous light.
What Does Flickering Lights Have To Do With Perceiving Motion?
People with higher Flicker Fusion thresholds (the ability to no longer see high speed flickering in flickering lights but instead see them as steady light) tend to have better accuracy in what they perceive in general. In other words, people who can see fast “flickering lights” at a higher speed than others, can see fast moving objects at a higher speed than others. [1]
This concept is also applied to TVs, cinemas, and other similar forms of media. That’s why frame-by-frame viewing needs to happen at frequencies close to the limit at which we can perceive “flickering lights” so we can view motion on TV fluidly. At a higher speed than our limit, things would look like they’re skipping frames. Animals that have better vision than humans need to “see fast flickering lights” better to survive in the wild. According to this study, Birds that have a great ability to see “flicker lights” can see fast-moving objects better than humans, especially when these objects could potentially collide with them in the air. [2]
Correlating that to movement is already somewhat of a stretch. The test is about being able to tell that any change is happening, not any proper perception.
Now, in the past thread, the following things were brought up as evidence that it is valid:
What this neglects is that, while FFT is of importance to perception, at no point is it identified as the unique deciding factor. There is correlation, perhaps, but it is never identified as the actual timeframe of perception of a moving object.Certain neurons in the visual system are specialized to detect sudden changes in luminance, and they play a fundamental role in the initial stages of processing visual motion. These neurons are part of a hierarchical system that progressively analyzes direction, speed, and overall motion velocities. So like… together, these processes contribute to our perception of objects in motion.
Hence why Flicker Fusion Frequency (FFF) is important for identifying moving objects too since that studies the literal first stage of perception — rapid changes in luminance. The other stages happen in and around that same timeframe of 1/50th to 1/90th of a second.
Oh and the study linked in the OP also implies FFT is important for animals to identify approaching targets fast enough.
As such it is more a natural high end, than a necessary low end. Something that leaves the FOV below the FFT would be invisible. However, that an object that does not do so is visible is never said.
I will demonstrate in the following that treating blitzing generally as occurring in under 2ms (1/500s), as the standard suggests, is not scientifically proven nor do they generally appear reasonable.
First, I wish to direct attention back to the primary source that the 2ms idea is based on.
This paragraph identifies a total of 19 variables which influence the FFT and hence the 2ms. Patterns and movement behaviour is pointed out for the case of hunting prey, but also things like frequency or precise position on the retina, and colour.The ability to detect flicker fusion is dependent on: (1) frequency of the modulation, (2) the amplitude of the modulation, (3) the average illumination intensity, (4) the position on the retina at which the stimulus occurs, (5) the wavelength or colour of the LED, (6) the intensity of ambient light [3,5,6] or (7) the viewing distance and (8) size of the stimulus [7]. Moreover, there are also internal factors of individuals that can affect CFF measures: age, sex, personality traits, fatigue, circadian variation in brain activity [4] and cognitive functions like visual integration, visuomotor skills and decision-making processes [3]. The performance of CFF in humans and predators alike is dependent on these factors. Umeton et al. also describe preys’ features like a pattern or even the way they move as relevant in perceiving the flicker fusion effect [2].
None of these factors are adequately handled by the current standard, as they are not handled at all. It's a constant number of 2ms that is not modified by any of these parameters.
Consider, furthermore, that some of these factors may be influenced in a movement scenario in a way different from a flickering light test. A moving object will have its light reach different parts of the eye's retina over time. The amount of light that a stationary object would all project on one part of the eye, is instead distributed on a greater area making the effective average illumination intensity less, as the same amount of light is spread more.
Given such considerations, I am confident to say that the source in itself would likely not agree with this usage of the value.
I will furthermore point out the the 2ms value is only produced in the study, only produced such high results with a specific setup: "The high spatial frequency image is first “bright on the left half of the frame and black on the right”, and then inverted. We observed the effect described in this paper whenever we displayed an image containing an edge and its inverse in rapid succession."
I.e. the specifically inverted the colours and specifically in an uneven way. Equalizing this to real life conditions, which may have edges but no inversion of said edges, seems faulty.
In general, I feel like the translation of FFT to blitzing is not scientifically validated. Human vision is highly complex and, as one has to consider, the brain does a lot of post-processing to every image. It for example pieces together several separate quick looks into a proper picture, one of which takes about 200ms alone. Speed based illusions are for instance also not fully explained by current state of science. It is a common idea that the brain's visual system simulates events before perception which can lead to illusions in movement that last longer than the mentioned 2ms. It is known that at least under certain circumstances the brain will decide to suppress blur for a better image, although that isn't the specific case.
The point is, scientifically hand waving the equalization between flickering light and movement is highly questionable due to the complexity of the subject.
Another problem comes from a translation of the FFT to a movement scenario.
Until they are out of the field of view? Until they reappear in the vision? I don't think so. Reasonably speaking, as long as you do not occupy the same place for longer than a flicker you should be below the threshold, as no part of the eye sees you longer than one flicker. It would be the equivalent of the left the of the image flickering below the threshold and right after the right side below the threshold. Considering that most humans are about half a meter wide, that severely would reduce the necessary speed. (half a meter in 2ms is subsonic+)
Of course, that is my argument as I see it as reasonable, not proper science. No scientific source describes how exactly one would apply flickering to invisibility via speed exactly, so that much is a given. But such uncertainties are exactly the problem.
Logistical problems aside, I believe the current state also is in conflict with real-life observations. I will say that while I will provide some videos in the following, since cameras with their shutter speed and videos with their FPS obviously screw with the observation, they are more for illustration purposes than that they prove anything.
First, let's talk about fans.
It is documented that when fans move fast enough, they turn practically invisible to, at minimum, some people.
Why is that a problem? Well, fan blades stay constantly in the field of vision. And they pass through a spot focussed on many times a second and for fans with thick blades they also occupy each point for a significant total percentage of the time. Furthermore, they are just subsonic. A standing fan is about 1300 to 2100 round per minute, which equates to 22 to 35 rounds a second. If we assume a blade takes up 1/6th of the total circle, then the time in which it passes through one spot in 1/132s to 1/210s, which is more than twice the time we have from FFT.
This appears not in line with the standard. The section assumes that a character can not stay invisible longer than said 1/500th of a second (otherwise, it would not make sense to conclude that a character that seemingly teleports must do so in that timeframe), but this example demonstrates that one can in principle stay invisible indefinitely.
Next, let's talk about guns.
A glock has a muzzle velocity of 375 m/s. That means that in 2ms the bullet passes a distance of 75 cm. In other words, by the current standard, we would assume that if someone shoots you with a Glock in the face from an arm's length away you can see the bullet emerge from the gun.
And... that is wrong. Even from a greater distance, you don't simply see bullets. (given bullets slow down with distance and if you stand far away they appear to move slower, so I don't think it's per se impossible to see a bullet fly under the right conditions)
Now, bullets are smaller than humans, but at less than a meter away one can still clearly see them. Now, I have no doubt that size does play a role in whether or not you can see a fast object, even in the realms where you can see the object when it stands still. Consider at that, that humans are not a regular size and the apparent size of a person lowers with distance. Meaning distance is also a factor.
However, notice that the standards do not account for that in any way. In fact, the flicker test does not account for size. (And there might be a difference between the relevance of the size of a light source and size of a passively illuminated object)
And, when translating from the flicker test results to actual movement... well, there it would need to be considered as well.
In total what we see by this example is that either 1/500s is not enough or the value at least needs a component to correct for size.
Conclusion
I believe I have sufficiently demonstrated that 2ms is not a plausible general-purpose timeframe. Furthermore, I believe to have shown that a general purpose timeframe is impossible. A number of factors do appear relevant, such as size/distance, illumination, contrast etc.Furthermore, I believe it to be plausible that such a method needs to account for the possibility of a character being invisible by speed, without leaving the field of view and for extended periods of time. A theory that does not account for that appears to be incomplete to me.
As such I propose to revert things to the prior standards on the matter.
Staff Votes:
- Agree: DMUA, Flashlight237, Therefir, Damage3245, DontTalkDT
- Split: DarkDragonMedeus
- Agreed but wanted to wait until more arguments are made: Dalesean027
- Disagree: DarkGrath
Last edited: