Discussion about Display Motion Physics (BFI, Color Mixing, Video Issues, Etc)

[Chief Blur Buster]

I have done one more video test, although you say that the video tests are not valid.

You misunderstand. Video tests are valid for a lot of things except for black frame insertion.
You have to understand the academic & scientific tools, and the proverb saying "The Right Tool For The Right Job".

Many video tests are valid and will properly demonstrate a lot of display concepts. For example, RTINGS uses this video to explain motion blur.

phpBB [video]

This is the correct use of video for the correct type of educational demonstration. This specific TestUFO test is not microsecond-sensitive like software black frame. You have to understand which tests are valid with video, and which tests are invalid with video. It's important to scientifically understand why sometimes this is a yes or no.

This test have impressed my . I have done a 60fps video with the ufo static and only changing the color from red to green form one frame to the next for one minute, to see if the frame color mix is the same at an static image as in a motion one. This is the edited video, is not a camera record is don by an editos.

I have tested it at a CRT at 60hz and at a LCD at 60Hz. And the ufo color is static, is a dark red as a mix of bright red an the dark green.
And now the impressive effect I see is that, at the CRT, if you move your head like saying no, horizontally. I see the ufo two times at the red and the green color . I don't happens at the LCD.

I don't know how to explain this. Could it be maybe a retina-brain effect?, as I it couldn't be that my video or the player have an error. Why it happens at the CRT and not at the LCD?. If this doesn't seem relevant to you I don't know what could do it. Is like is as I said that two frames are mixed to get one. And I have been thinking that, strangely, this could explain all the effects you explain here with your test.

It's a fascinating effect when you see it for the first time, but It's very ordinary to us. Has research papers. Not a surprise.

It's the well known Rainbow Effect, similar to the DLP Rainbow Effect. Temporal color-separation effects are nothing new, and is amplified by shorter pulse widths (large flicker duty cycles, where pixels are OFF longer than ON, amplifies the rainbow effect). Google "Rainbow effect on displays".

You do not see it easily on LCDs because the test you did does not amplify the weaker rainbow effect caused by 1:1 duty cycles with no black periods between color changes.

(A) Longer black periods amplify rainbow effect.
It's easier to see if it's "RED-OFF-GREEN-OFF-RED-OFF-GREEN" rather than "RED-GREEN-RED-GREEN".

(B) Faster pixel response amplify rainbow effect.
It's easier to see if it's "RED-(sudden)OFF-(sudden)GREEN-(sudden)OFF-(sudden)RED" than if it is "RED-(fade)OFF-(fade)GREEN-(fade)OFF-(fade)RED". LCD pixel response is slow, GtG is when pixel colors fade slowly from one color to the next. See 1000fps high speed video of LCD refresh cycles. Software-based rainbow effects on LCDs will have the fade problem because of slower pixel response.

So you can't easily see the rainbow effect on LCD because you have two simultaneous problems of (A) and (B). Rainbow effect is still there, but it's below your human-eye detection ability because of (A) and (B). You need to amplify it more in order to be visible to your eyes on LCD. CRTs always permanently has black periods between refresh cycles, satisfying (A). CRTs also have extremely fast pixel response, satisfying (B). As a result, it's much easier to see on CRT than LCD.

Here's a 1000fps high speed video of www.testufo.com/scanout running on a laptop LCD. Observe how slow LCD pixels are, preventing the rainbow effect from being easy to see:

phpBB [video]

Notice how the pixels are slow? GtG = Grey to Grey = LCD pixels fade from one grey to another grey. (LCD subpixels are always monochrome, they're just color-filtered as necessary to R,G,B).

However, a software-based rainbow effect is still visible on LCDs if you try fast motion at www.testufo.com/rainboweffect and roll your eyes around -- try speeding up the scrolling to 1920 pixels per second and rolling your eyes around on different images (map, stars, Paris Eiffel Tower), you will see similar rainbow effect. It's EASIER to see on a CRT, but it's STILL visible on LCD if the test pattern is blatantly exaggerating the rainbow effect.

All of these effects already have pre-existing research. It's simply because different colors are being stamped along your retinas, and it's easier to see the color-separation effects with an impulsed-display. However, sample-and-hold still can generate rainbow effect (just fainter).

Now do you understand?

[mynm]

Thank you very much for the explanations and videos. Sorry, I will try to understand what you are saying and your info.

I will tey tu understand why a black frame insertion like video is not valid with a video with your info.

For what you are saying, as Rtings's video is using one of your test at the video, and it seems to be recorded with software fron what the gpus sends to the monitor, not a video recorded with a camera. I understand that my software recorded videos form the gpu of your web test are valid, if they don't have black or white insertion. Is like that?.

So is a video that is showing for example two fotograms like this ones of the image:



that shows the distance between these two different frames, considerated a white frame insertion, as is not showing the same full image between frames?. This is the video.

I don't know how this image results in one full image but with a small white line in betten the two parts, and its color blinking with white. And I don't know when the distance between the frames will separate them.

Edited: OK, I understand that the distance between frames is not what is show, and that it will not not affect it untill the frame rate is low or the distance is high. And that my video is white frame insertion, as is not one object are two objets.

But things like that and the black frame insertion, show that even you can see a white or black full frame between two frames at 60 fps and that at more fps it will be the less noticiable. And as even the same object with a two color cycle difference is show with a mixed colors but, so tha at 60 fps and hz is like seen 30fps, because the eye apparently can't see this type the differences.
Even I have testet a four colors cycle and at the 60 fps line I see one flickering mixed color.

But it seems that you need less frames to see an object in a perfect motion at 960 pps in a CRT than flickering, so even 30 fps at 30hz at an CRT could be enough for motion, if the speed is not high and flickering didn't happen, and you will see no difference. But this not happens at LCDs because of the lack of respones.

I was thinking that changes in colors could be equivalen as an object position change, but is not at CRTs. Could it be equivalent until you start seen a two color change between frames at CRTs?. Sorry for all the questions.

Edited 2: All this misunderstood for my part, is only that I didn't have in count that what at 30 fps and 60 hz is doing the separation in motion between the two repated image is the retina and not the distance between the two objects.

The probleme is that I didn't have seen the retina explanation at forums or even video game reviwers. At forums 30 fps at 60 hz is "slideshow" as like the timing or distance is what is doing the effect. An even at video game reviwiers is called "cinematic effect", not explainig this retina effect, saying like that it have a good implemented motion blur that masks the distance between frames, but there is not distance between frames to mask, and the motion blur effect is not covering this effect in any game ad I think that maybe is impossible. Also more mathematical formulas and less words will make me and people understand that this better, but I don't see any of this anywhere.

Thank you very much for your explanations, I think now that all my videos can be explained with that double image retina effect and the capacity of the eyes to see changes.

Edited 3: Or am I misunderstanding all again and the retine double image effect is only happening at CRTs and not at sample and hold displais and is what I see the lack of the panel response as there is no flash per static frame per static frame?.

I guess that's how it is, as the first rigth of the two ufos at the LCD and is caused because the lack of response so is changing to black. And that is why the 30 fps at 60 hz is lees noticiable at LCDs than CRT. I was misundestanding the causes, because for not knowing the retina effect. Sorry.

[thatoneguy]

There are over 1000 different refresh rates at 60 Hz -- one GPU is outputting at 60.001366Hz and another GPU is outputting 59.99983Hz, due to clock manufacturing differences.

What is advertised at 60.0000000 Hz never happens

Thank God for Variable Refresh Rate technology

[mynm]

I have done one more test, the changes I did to the original gpu software recorded video is thar colors are inverted and I have numerated the frames at the left of the ufo at the 60 fps line. This is the video. The number is chaning from one to two (it is little bit thicker), from one frame to the next frame, is not a black frame like video insection, I don't have added any white patch, like you can see at the video editor:



As the two is little bit thicker, at the video I can't see the one, the two is displayed every frame at the CRT and at the LCD. Do anybody can see the one?.

What is causing this effect?, is the video compresion or its format?, is the video player?, are the monitor Hzs changin in ms?. At the CRT I don't see any ghosting, only a phosphor trail when the black ground is black, but do you think it is?. Is it a monitor problem?, I think I don't need more fps to see the one as is a frame that is at the video, are my eyes and it is the retina image persistence?. I don't know but it seems that at 60 fps at 60 hz I see 30 mixed fps. I have to suppose that this is like motion works at the eyes. Do anybody can explain this?.

[Chief Blur Buster]

Your interesting post requires a longer response, which I will try to write by December.
Shorter questions will be answered quicker.

Edited 3: Or am I misunderstanding all again and the retine double image effect is only happening at CRTs and not at sample and hold displais and is what I see the lack of the panel response as there is no flash per static frame per static frame?.

I guess that's how it is, as the first rigth of the two ufos at the LCD and is caused because the lack of response so is changing to black. And that is why the 30 fps at 60 hz is lees noticiable at LCDs than CRT. I was misundestanding the causes, because for not knowing the retina effect. Sorry.

There’s no halved UFO effect at all.

It’s a whole UFO vibrating sideways.
Imagine a slow vibrating string (guitar string) that vibrates noticeably
Or a fast vibrating string (guitar string) that blends to blur (double string with a blur in between).

Trying to cut the UFO in half is the wrong way to explain this effect. Please study/learn in a different way. Do not use a math textbook to learn biology, and do not use a biology textbook to learn math. Right education for the right job. Cutting UFO in half is wrong. It’s a whole UFO vibrating sideways. It’s easier to see the edges than the overlap though, but it’s still a whole UFO vibrating sideways.

Also, there are two different pixel response benchmarks, GtG and MPRT. Please read www.blurbusters.com/gtg-vs-mprt

Yes, adding black period makes it easier to see. So you need to add BFI like
Red UFO frame
then
Black frame
Then
Green UFO frame
Then
Black frame

So instead of Red-Green-Red-Green, you have Red-Black-Green-Black-Red-Black-Green. That’s what you need to do on LCDs to more easily see the rainbow effect a bit easier in these cases. The black compensates for the slow GtG, by reducing MPRT for a given frame rate. A 50%:50% software black frame can halve MPRT (down to one refresh cycle granularity) despite intervals between beginnings of visible frametime much longer than a refresh cycle.

Anyway, due to time constraints (big move), longer replies to long questions may not be very frequent. If you can reduce the size of your question to one, and use shorter paragraphs, I may be able to explain things in a more incremental manner.

I implore you to re-read all 20+ articles at www.blurbusters.com/area51 including the older articles on the next page (click Next at bottom), to reduce the size of your questions as replying to forum threads is unpaid time, and you should narrow the scope of your questions over a longer time period, as a courtesy (for now). Thanks!

I also implore you to borrow a 120Hz+ display to retest some UFO effects, because you need higher frame rates for the stutter-to-blur continuum. Blur Busters education is 10x faster to explain when my students have a 120Hz+ Or 240Hz+ display. It’s like buying a mandatory university textbook, owning a 120Hz+ display or borrowing a 120Hz+ display, is a defacto textbook requirement in my Blur Busters training classes in person.

Until then, please slow down/throttle your question count and/or repost only one question in a shorter paragraph, asked in a correct manner. You do not ask a biology teacher a math question. So please forget the left half / right half questions, and try to understand why that’s the wrong way to ask the question. It’s simply a whole UFO vibrating sideways.

Just like www.testufo.com …. Slower frame rates vibrate more, higher frame rates blends to blur. And black periods simply hides the blur between the vibration extremes. Stutter is the same thing as motion blur during eye-tracking. (low frequency visibly stutters, high frequency blends to blur), as ween at www.testufo.com/eyetracking#speed=-1 because understanding that animation is a pre-requisite to understanding some of the questions you’re asking me.

Also, some effects can only be seen during real time animation, rather than by video. When the frame rate is modified (e.g. from a video playback problem or a video file format limitation), some blurs becomes stutters. Some stutters becomes blurs. Some blurs becomes double images. Some stutters become double images. Or flicker appears/disappears. When you do things at the wrong frame rate or change the frame rate, or single-step.

In this case, you want to read more before you ask questions. Imagine a long line to a very important professor in a 300-person class, you narrow down your questions to only the most important question that is not already covered in an existing resource. So again, please read all 20+ articles at www.blurbusters.com/area51 including the “Page 1”, “Page 2” and “Page 3” button at bottom to see additional older articles that already answers a portion of these questions. This is your mandatory homework before your next question. This will help you construct your next question carefully, at one question per turn. Thank you!

Cheers,

[mynm]

Double post sorry.

[mynm]

Thanks for the response. I will try to l undesrtand what you say and read your links.

I think I have been short at my exposure, less at this post, for get it:

Forget all my post if are not relevant, but please, what is happening what I see here:

Is not like I am saying that more than 30 fps are not visibles, is only like frames are mixed to get one, I have to test but with more fps and Hz I think I will see the same. Is what my eyes are shown, I can't change that. I only I want to know why this happens, cause I didn't have read anything like that.

[mynm]

Slower frame rates vibrate more, higher frame rates blends to blur.

If two frames are mixed, with hight framerates them will get to one, and if you see a vibration, is maybe because more that one will be mixed, so maybe you only see the borderders for the multiple frames of the objest. So Maybe is not like you see half of the frames of the Hz, maybe it is exponecial. Is a guess, do you think is wong?.

[Chief Blur Buster]

For an improved understanding, check 1 minute into this RTINGS video:

phpBB [video]

The vibration is because of the digital stepping effect of a finite frame rate, versus an analog moving eyes.

It is also demonstrated by multiple tests such as TestUFO Eye Tracking, "Variable" speed setting.

It is shown on TestUFO cover page:
- 5 UFOs, Default Speed
- 5 UFOs, Fast Speed

Lower frame rates look like they're vibrating around your analog eye tracking gazepoint as you're tracking.
Higher frame rates look like they're blurring around your analog eye tracking gazepoint as you're tracking.

At 60fps,120fps,240fps it is mostly blur.
The stutterwidth / blurwidth is the same, and will be halved at double frame rate.

Remember software black frame insertion can be variable blur width, based on number of visible frames and number of blank frames.
- 5 UFO with black frame insertion at fifth of refresh rate
- 4 UFO with black frame insertion at quarter of refresh rate
- 3 UFO with black frame insertion at third of refresh rate
- 2 UFO with black frame insertion at half of refresh rate
- Observe how the blurwidth (stutterwidth) of the bottom-most two UFOs is identical in all the above animations! (within GtG error margin)

To avoid reductions in frame rate or refresh rate, it needs hardware black frame insertion via impulse driving (e.g. flickering CRT, flickering backlight, etc). However, the math and science of display motion blur is the same for software black frame insertion:



For equivalent motion blur without black frame insertion:



Do you have access to a display of at least 120Hz in order to understand this display education better? Display education is far easier to teach when you have access to a 120Hz+ or 240Hz+ display, since the vibrations will be related to your flicker fusion threshold. Perfect regular stutter of sample-and-hold becomes blur when the the stutter is above your flicker fusion threshold. That's why 15fps and 30fps looks shaky on high persistence (high MPRT displays) but smooth at 60fps, 120fps, and 240fps.

- The width of blur(or stutter amplitude) of 240fps is half the width of blur(stutter amplitude) of 120fps at the same motion speed.
- The width of blur(or stutter amplitude) of 120ps is half the width of blur(stutter amplitude) of 60fps at the same motion speed.
- The width of blur(or stutter amplitude) of 60ps is half the width of blur(stutter amplitude) of 30fps at the same motion speed.
- The width of blur(or stutter amplitude) of 30ps is half the width of blur(stutter amplitude) of 15fps at the same motion speed.
(and so on)

The threshold where stutter is blur or blur is stutter, depends on various variables such as your flicker-detection threshold. The thickness of stutter (how far an object vibrates) = the thickness of motion blur (how thick the motion blur is ahead/behind the object). It's the same thing, whether it is stutter or it is blur (because it stutter-vibrates too fast). It's like a vibrating string, where slow vibration is visible, and fast vibration is blurry. That's why we can still see display motion blur even at high refresh rates, and why 240fps is half as motion-blurry as 120fps (assuming GtG=0 or near 0).

You can already see that at www.testufo.com that twice the frame rate = half the amplitude of stutter. This continues when going beyond 60fps. This is a property of sample-and-hold that does not happen to CRTs or hardware-strobed displays. This display motion blur problem still exists even at 0ms GtG (see Why Does Some OLEDs Have Motion Blur?).

Ideally, a student of display science needs to 100% fully understand the basics of sample and hold technology, before asking too many questions about black frame insertion and why the double-image effect is created.

Now, double image effect is from interrupting the motion blur between the outer amplitudes of the vibration. So the copies of the UFO is "stamped" onto your analog-moving retinas, as there are repeat-flickers.

So you're having an interrupted motion-blur that turns into a duplicate-image-blurring instead, like this:



If you didn't flash and you did sample-and-hold instead for these frame rates, you would have 1x blurwidth, 2x blurwidth, 3x blurwidth, and 4x blurwidth (quarter frame rate = 4x the motion blur width). So the amplitudes are the same, except you have a continuous blur of the same distance, instead of just a stroboscopic duplicate-image effect.

Understanding sample-and-hold AND also understanding black frames / impulse driving is essentially interrupting the sample-and-hold effect (eye tracking based motion blur) can turn blur into a phantom array effect if you repeat-strobe along the persistence-based motion blur. This is the CRT 30fps at 60Hz effect in a nutshell. To avoid the duplicate images, requires flashing occur only once, e.g. flash rate = frame rate (flicker frequency = frame rate = refresh rate). Whether it's software BFI or hardware impulsing, that needs to become a constant.

Do you understand the math of sample-and-hold? If not, take time to study it some more (lots of articles on Blur Busters) and come back before asking next question.

If my post has created more questions than answers for you, please study research more before coming back to ask your next question.

[mynm]

Thanks I will take a look when I could.

About why the one is not see and the two is seen. I understand that is becuse it is not repeated the enought number of times. If it will be repeated two or more times, and then the two two or more times, the one and the two will be seen. I think is ovum. So more repetitions are needed to see the one and the two. And with more fps you can see more thinks as them can be repeated more times in the same space, not because you can see one specific frame. Is not like that?. If always have been like that, sorry, but everybody is explainig it wrong.