• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

My super scientific poll on 60fps/30fps, frame blending, motion blur

H

hesido

Member
Update:
As of 23rd of May 21:00 UTC, 69 people have responded to the test at: (the test received over 2800 views)
https://www.shadertoy.com/view/4lSGDy

68 of them mentioned their preferred effect, with or without the mention of effect they disliked the most, 1 only mentioned their most disliked effect.
Since the responses are not in the same format, I picked the most obvious information and made a chart:
8AOvutV.png

  • An overwhelming amount of people (72%) preferred 60fps with motion blur.
  • 19% chose the good old 60fps plain.
  • 60 fps with blending, although 60fps, received no love from the participants as their favorite effect
  • 4 people preferred 30fps with motion blur over others.

Information not shown in graph:
  • Some people (will update with numbers after I rescan thread) preferred 30fps with motion blur over 60fps plain, while having the 60fps motion blur as their favourite.
  • 30fps with blending is the most hated / least liked effect. 28 people out of 29 either mentioned it as being horrible or listed it as their least liked effect. 1 person thought 30fps moblur was the worst.
  • Some people mentioned 60fps blending to be as bad as the 30fps blending and grouped it alongside their least liked effect.
  • Some respondants only preferred 60fps moblur and plain, and threw everything else in the bin.

Obligatory pie chart, Gaffers will know:
0N7C8nz.png


Detailed information is in the thread. Feel free to make further deductions from the info at hand, and share with us.

I will update the numbers if there's enough responses past this.
----
Hi fellow Gaffers.

The visual test:
https://www.shadertoy.com/view/4lSGDy

I modified an existing shader on Shadertoy.com by HLorenzi to display 6 circles at different frame rates and different effects. The first 3 updates at 60fps, the last 3 updates at 30fps. Here are the circles in order:
  1. 60fps, no effect.
  2. 60fps, frame blended (exhibiting ghosting)
  3. 60fps, motion blurred
  4. 30fps, no effect.
  5. 30fps, frame blended (exhibiting ghosting)
  6. 30fps, motion blurred
This test is just for motion clarity, and not much else. The test won't work if the frame rate displayed is not ~60fps.

Your opinion will have
no
effect on future game development.

In your honest opinion, how do you rank circles with respect to motion clarity? Which one is easier on the eyes? Please answer with your display type. Those with different types of displays could test for different displays, because the display type seems to effect the results.

The test isn't meant to be a serious attempt in laying out a clear picture, but meant as a conversation starter and add bits to our knowledge. The test clearly doesn't represent a proper game scene with complex shapes (which may reduce the perceived negative effects of ghosting, for example).

Here's my take on the perceived smoothness of motion, on a laptop LCD:
60fps with motion blur >>> 60fps plain >> 30fps motion blur > 60fps blending >>> 30fps plain >>>>> 30 fps blending

Added take away notes based on my own perception:
60fps is not enough to simulate motion on a continuous line, there's strobing because object jumps between positions due to limited number of temporal samples displayed. 60fps with motion blur fills those gaps and eliminates strobing.
30 fps with motion blur looks really smooth!
30 fps without motion blur is bad.
30 fps with blending is absolute shit.
60 fps with blending imbues the strobing in the image, but it still looks much better than 30fps, however, it'd better not miss a single frame becaues 30fps blending is horror story material

edit: Now when you click inside the window, the refresh rate is halved, so you can see 30fps / 15fps instead of 60fps/30fps. You can also change the MAXFPS if you have a higher refresh rate monitor, but I'm not sure if it would work.

edit2: The best would be to not tell what the circles are, but ideally the circles would have to be displayed in random order so users do not "group" circles sub-consciously, since that wasn't possible I just felt like telling the users.
 
O

ohyouagain

Member
I would buy a panel that would allow me to play at a constant perceived 60 FPS with no tearing, no stutters and no artifacts.

The rest doesn't matter.
 
F

Fisty

Member
I thought #3 looked best overall, but im watching on mobile so i cant really give a definitive opinion. Pretty neat stuff you got there, though.
 
B

bennywhatever

Member
To me:

30fps Plain > 60 fps Plain > 60fps Motion Blur > 30fps Motion Blur >> 60fps Frame thing >>>>>>> 30fps Frame thing

I don't know why, but 30fps Plain just looked the most pleasing to my eyes.
 
H

hesido

Member
SigSig said:
I agree with your ranking, Asus LCD Screen.
Can you modify it to display 120/60/30fps?
Click to expand...

Hmm, does your frame rate counter report higher than 60 on this? I may put that in a variable somewhere if that's the case. (Or maybe activate that when user clicks on the page)

bennywhatever said:
To me:

30fps Plain > 60 fps Plain > 60fps Motion Blur > 30fps Motion Blur >> 60fps Frame thing >>>>>>> 30fps Frame thing

I don't know why, but 30fps Plain just looked the most pleasing to my eyes.
Click to expand...

I have no explanations for this. What's your display type? And does the frame rate counter report ~60fps? If it doesn't, the test may look wonky..
 
McHuj

McHuj

Member
Interesting. This is pretty cool.


My preference on 24" HP IPS monitor:

1. 60fps, motion blurred
2. 30fps, motion blurred
3. 60fps, no effect.
4. 30fps, no effect.

5/6. 60fps, frame blended (exhibiting ghosting)
30fps, frame blended (exhibiting ghosting)

Last two are equally bad imo.
 
L

LukasTaves

Member
Surprisingly for me 30fps with motion blur was the most visually pleasing. 60fps+MB was the smoothest, but somehow felt worse.
 
N

nictron

Member
The 60fps and 30fps Motion Blur seem the smoothest. My eyes prefer the 60 fps and 30fps plain, with 60fps being my favorite.
 
E

Enter the Dragon Punch

Banned
While this is a good example, the issues are that a lot of games seem to handle motion blur completely differently. Crysis 2, for example, has awful, awful motion blur that makes the game feel worse even at 30FPS, IMO. Meanwhile, something like Alan Wake handles motion blur pretty beautifully and adds to the experience, whether it's 30FPS, 60FPS, or higher.
 
SerTapTap

SerTapTap

Member
1)Lots of ghosting but smooth
2) Too much ghosting
3) Very good
4) AAAAAAAAAAAAAAAAAAAAAAAAAA
5) I WANT TO GET OFF THE RIDE
6) Smoother than 3 but not an extreme difference. It seems "faster" when going up and down for some reason

Woah 6 is 30 FPS not 3? Weird, aside from that feel of 6 moving "faster" I honestly can't tell much difference
 
D

Durante

Member
3 and 1 are the only acceptable ones, in that order. 6 is borderline acceptable.
(I decided this before looking at what they are)
 
A

AJLma

Member
That's some cool looking motion blur.

float frametime = (120. / (floor(view / 3.) + 1.));
Click to expand...

I set this to 120, does that mean the left half is 120FPS and the right half is at 60 now?
 
S

Skyzard

Banned
I like 60fps motion blur
then 30fps motion blur
then 60fps
then frame blended 60
then 30fps
then frame blended 30

Might be to do with the motion blur looking great though (30fps + motion blur being high up).
 
Bonfires Down

Bonfires Down

Member
1. 60 fps motion blur
2. 30 fps motion blur
3. 60 fps plain
4. 60 fps frame blended
5. 30 fps plain
6. 30 fps frame blended

On Macbook Pro Retina

Not sure how well this test translates to actual games though.
 
H

hesido

Member
AJLma said:
That's some cool looking motion blur.



I set this to 120, does that mean the left half is 120FPS and the right half is at 60 now?
Click to expand...

Yes, indeed.. Does your frame rate counter display 120?

If so, I'll add it as a pre-defined variable at the top so the 120hz activates when user clicks on the window.
 
M

MaxwellParrish

Member
I think that there is something off in your implementation of motion blur 'cause it seems a little exaggerated. I would rank first 60 fr with MB and close second 30fr with MB. 60fr without MB follows 'cause when the movement get fast you can easily see the various position of the ball as independent from the movement and, obviously, things get even worse with the 30 fr. I really don't know how to positionate the two with frame blending because while somehow improve the sensation of movement it's very unnatural and hard to look at.
 
H

hesido

Member
MaxwellParrish said:
I think that there is something off in your implementation of motion blur 'cause it seems a little exaggerated. I would rank first 60 fr with MB and close second 30fr with MB. 60fr without MB follows 'cause when the movement get fast you can easily see the various position of the ball as independent from the movement and, obviously, things get even worse with the 30 fr. I really don't know how to positionate the two with frame blending because while somehow improve the sensation of movement it's very unnatural and hard to look at.
Click to expand...

You can reduce the predefined variable:
#define EXPOSURE .5
to your liking. If set too small, it wouldn't be noticable of course. You can also simulate less precise motion blur implementations by reducing the number of samples:
#define MOTIONSAMPLES 24

Obviously this is a very "high quality" motion blur, it also brute forces its way, renders the circle as many times as the number of samples to get the effect. It's of course not a direct representation of game quality motion blur but games are getting there.
 
H

hesido

Member
For people with higher refresh rate monitors: I added a pre-defined variable:
#define MAXFPS 60 //maximum simulated fps regardless of display refresh rate

You can change it to 120fps, for example.

Don't know if that would work at all, for it to work the reported frame rate has to be ~120fps.

I just added a new feature, that halves the simulated refresh rate when you click inside the window, so you can see what 15fps looks like if max fps is set to 60. 15fps with motion blur holds up surprisingly well :)
 
dr_rus

dr_rus

Member
3>1>6 on a Dell U3014. Can't imaging it being different on a 144Hz display though.

Frame blended circles look like something glitched in my opinion.

Have to add that motion blur on these objects is not the same as the full screen motion blur which is used in most of games to cover the lack of fps. This is a per object motion blur which is looking good in case of high velocities. I tend to turn off any full screen motion blur in games as this creates more problems than it solves in my opinion.
 
L

LeleSocho

Banned
60fps plain
60fps frame blended
60fps motion blur
30fps frame blended
30fps plain
30fps motion blur


Frame blending create a little of smoothness but ghosting is bad
Motion blur make things reeeeallly smooth but destroys the image in the meantime

I put frame blending over plain on 30fps because the frames just aren't enough so even a little of smoothness helps a ton but as i said earlier motion blur really damages the quality to make the smoothness worthwhile.


Late 2011 MBP 15 inches screen.
 
M

MaxwellParrish

Member
hesido said:
You can reduce the predefined variable:
#define EXPOSURE .5
to your liking. .
Click to expand...

Sorry to sound completly clueless (indeed, I am), but do you need to do something for applying the changes? I tried to change that line to .0 and 1.5 but nothing happened...
By the way, I think that the blended frames ones would look less hideous if the two frames weren't blended 50-50 but the new rendered frame would overweight the old one, like 60 to 40 or something
 
J

J-Fr

Member
3 and 6 looks good on my 60Hz monitor.
Compared to that on the 144Hz monitor only the 4 and 5 have a noticeable amount of ghosting.

If I put both screens side to side, all the dots on the 60Hz monitor are looking horrible and vice versa..
 
A

Aureon

Please do not let me serve on a jury. I am actually a crazy person.
Without actually reading what they are, so you'll get some blind data:

3 > 6 > 1=4 > 2 >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> 5
Seriously, the hell is up with 5?
 
C

CarbonFire

Member
On a 60HZ Dell Ultrasharp IPS

3 = 1 > 6 > 2 > 4 > 5

I don't mind motion blur when done well, So 1 and 3 are neck and neck. Motion blur 30 is ok in this example, but would be very smeary and hard to look at if it's used on an entire scene in a game. 2 Looks worse on my secondary TN display due to lower ghosting, but the slightly higher ghosting on my IPs display seems to smooth it out. Regular 30 is kinda hard to look at, and the frame blended 30 is just laughably bad.
 
H

hesido

Member
MaxwellParrish said:
Sorry to sound completly clueless (indeed, I am), but do you need to do something for applying the changes? I tried to change that line to .0 and 1.5 but nothing happened...
By the way, I think that the blended frames ones would look less hideous if the two frames weren't blended 50-50 but the new rendered frame would overweight the old one, like 60 to 40 or something
Click to expand...

After making changes, press the compite button (shaped like a play button) at the bottom of the code editor.

You can adjust the blending to your liking by modifying the line, see if if 60/40 makes tangible difference:
vec4 ghstCol = (mainCol + scene(uv, time - (1. / frametime))) * .5;

You need to multiply mainCol with .6 and the result from scene with .4 and add those, instead of adding those and multiplying by .5 . I'll leave that as an exercise for you :)
 
V

Version 3.0

Member
3, 6, 1, 4, 2, 5 in order of preference. 2 and 5 are atrocious. I'm on a Samsung SyncMaster E2420, don't know its specs (work PC).
 
I

IcyEyes

Member
I will try this visual test on some other monitors as well.

Anyway, on a Dell ultraHD the number 3 is my first choice and I'm not sure if the second place belongs to the number one or six.

Two and five are pretty bad!
 
T

thepotatoman

Unconfirmed Member
When it's going super fast, it does look choppy without motion blur, but when it's going slow, it looks fine without the motion blur, while the motion blur makes the slow moving ball look terrible. The only time you are focusing at something moving that fast is if you're turning around with a high mouse/controller sensitivity.

I'd also say the red on white color scheme makes the no motion blur choppiness extra noticeable, while a normal video game situation has a lot more colors to hid the effect of the extra choppiness.
 
E

ElTorro

I wanted to dominate the living room. Then I took an ESRAM in the knee.
On my display (MacBook Pro 15' Retina) the 30fps circles look surprisingly decent. The test doesn't seem to really reflect the impact of framerates in actual games, at least not on my display.

/edit: Nah, I take that back. After looking at it for some more time, only the motion-blured 30fps circle is tolerable.
 
M

Mulgrok

Member
On a 60hz IPS
They all are distorted, but 3 seemed the most acceptable.
Wish I had a 120hz+ display with a similar test to see how much better it looks. Might need even more fps to complete eliminate the flickering.

Motion blur is just a band-aid for low framerate... keeps my eyes from bleeding, but doesn't make it good. :p
 
F

FLAguy954

Junior Member
122 fps plain
122 fps motion blur
61 fps plain
61 fps motion blur

All seen on my Seiki SE39UY04-1 at 1080p/122Hz.
 
L

LCGeek

formerly sane
ohyouagain said:
I would buy a panel that would allow me to play at a constant perceived 60 FPS with no tearing, no stutters and no artifacts.

The rest doesn't matter.
Click to expand...

The panel isn't the cause of those issues your OS,mobo, and GPU are more at fault for those issues than anything else.
 
N

Nickle

Cool Facts: Game of War has been a hit since July 2013
I think the animations are incorrect, I prefer the 30fps and that isn't possible.
 
You must log in or register to reply here.

Similar threads

I am so tired of the whole 30 fps this 60fps that discussion. You all lost the sight on what really matters and it's ruining console gaming
Social Drama Hardware 6 7 8 9 10
452
21K
The Stig replied
Why 30FPS was once acceptable, BIT ISN'T ANYMORE!... mostly...
2 3 4
183
12K
rofif replied
Introducing G-SYNC Pulsar tech
2
51
4K
TrebleShot replied
[VG Tech] Forspoken PS5 Frame Rate Test (Demo)
Review 2 3 4
164
8K
In Cold Blood replied
[DF] Resident Evil 4 Remake on iPhone 15 Pro: Can Mobile Truly Handle AAA Gaming?
44
3K
StereoVsn replied
Top Bottom