PDA

View Full Version : What's wrong with my motion blur?



IMI
04-03-2009, 08:33 PM
See the attached H 264 QT H 264 movie, please. It's only 52 KB.

I just did this as a quickie test for an animation I'm working on. I never messed with motion blur before in an animation and figured I ought to test it out on something simple before trying it on a real scene.

The problem is, as the blue ball moves forward, it seems the motion blur is in FRONT of its motion, not behind it as I would expect. I don't know how else to describe that. It appears most solid in the middle of the MB, where I would think it would be the leading edge that should appear most "solid".

I realize the motion isn't quite right - the inertia of the blue ball seems off, but as I said it was a test only for MB, not an attempt at realistic motion.

MB settings:
Photoreal
Use Global is unchecked
Blur Length: 300%
MB passes: 4
Shutter efficiency: 100%

I have to admit, I really don't understand what the settings do other than Blur Length and Photoreal. I assume MB passes is a quality setting.
LW 9.6 x64

Thanks if anyone has any suggestions. :)


EDIT:
It does it on the white ball too, but it's not as noticeable as it's moving much faster. Could that be the problem - that the blue ball should be moving faster considering the settings I used?

Jockomo
04-03-2009, 09:56 PM
motion blur passes determines the number of times it will redraw the object, if you look closely it looks like there are 4 instances of the ball in the blur. The faster your object moves, the more blur passes it will need to hide that kind of artifact.

The blur length determines how far it smears the object. You probably have it set too high, the default is 50.

IMI
04-04-2009, 04:06 AM
Thanks for the info, Jockomo, that helps alot. :)

I notice now the Blur Length setting seems to actually be the Exposure Time. At 50% it's .0167 sec. At 500% it's .1667, or 10x slower, and so on up and down the scale.

Honestly, if they just called it Shutter Speed or Exposure Time instead of Blur Length I'd have probably figured it out and not had to ask. Brain wasn't working fast enough last night. Or is the term Blur Length something real photographers use more commonly? ;)

EDIT:
But I don't think this answers why some of the MB seems to be ahead of the object's motion. Regardless of settings, shouldn't the MB create a trail in the wake of the object's motion?

dandeentremont
04-04-2009, 10:48 AM
Nah, motion blur is actually thickest at the center of the movement. I used to think it trailed, too. I can get some pics up as an example once I finish some errands today.

Tobian
04-04-2009, 10:51 AM
It used to be the other way round but they made shutter speed 'forward' for some sort of compatibility reasons.

Anyway, the solution is simple set blur length to -300 - the trail will then be behind it...

Oh and PS: Why did you set it to 300%? That basically means 3 frames worth of movement, which is going to look a little off, unless you want to exaggerate it, it's generally going to look 'unrealistic' unless you want akira light trails :) Also the shutter efficiency is basically for how much of the frame's movement you want to sample (virtually: the cameras iris is open for X amount of the time in the frame), so again a value lower than 100% is realistic for in-motion video, though stills photography could obviously have a huge exposure...

toby
04-04-2009, 03:38 PM
The mb was reversed to play nice with other big apps like Maya and Houdini - while at dd commercials I always had to enter -50% blur to match other renderers.
300% will calculate the current frame and look into the future by 2 frames, only good for special effects, not realism.

IMI
04-04-2009, 04:19 PM
Thanks for all the info, guys, I do appreciate it! :thumbsup:



Oh and PS: Why did you set it to 300%?


To be brutally honest, ignorance is why I set it to 300%. :D
Just had no idea what I was doing.

THREEL
04-04-2009, 04:51 PM
To be brutally honest, ignorance is why I set it to 300%. :D
Just had no idea what I was doing.

I'm sure most, if not all, of us have been in that boat. :D

Anyway, shouldn't there be at least a little forward motion blur at the very beginning of the blue ball's movement, because of the transfer of energy from the white ball to the blue ball, the motion blur would then go to none very briefly, and then, behind?

Just a hypothetical question.

toby
04-05-2009, 04:36 AM
I'm sure most, if not all, of us have been in that boat. :D
Yep, every single one of us. I spend so much time in that boat you'd think it was my commuter :p


Anyway, shouldn't there be at least a little forward motion blur at the very beginning of the blue ball's movement, because of the transfer of energy from the white ball to the blue ball, the motion blur would then go to none very briefly, and then, behind?

Just a hypothetical question.
No it's generally centered.
http://www.cs.mtu.edu/%7Eshene/DigiCam/User-Guide/filter/nd-4-8.jpg
http://i.ehow.com/images/GlobalPhoto/Articles/2036673/motionblur03_Full.jpg

The shutter is open for a certain length of time, an object in camera moving at a constant speed will blur evenly from beginning to end of that time. It may look like it's biased towards the center, but that's only because there's more overlap at the center, it doesn't have as much background showing through.

IMI
04-05-2009, 05:00 AM
No it's generally centered.
http://www.cs.mtu.edu/%7Eshene/DigiCam/User-Guide/filter/nd-4-8.jpg
http://i.ehow.com/images/GlobalPhoto/Articles/2036673/motionblur03_Full.jpg

The shutter is open for a certain length of time, an object in camera moving at a constant speed will blur evenly from beginning to end of that time. It may look like it's biased towards the center, but that's only because there's more overlap at the center, it doesn't have as much background showing through.


Interesting, toby, thanks. :thumbsup:
I suppose seeing is believing. Certainly can't deny the evidence, but it seems strange. Is this one of those quantum things? The duality of the nature of light as a particle and a wave, I wonder?
It just seems really odd that a car moving forward would not appear most solid at its front. I don't know how many times though I've seen such pictures and simply never thought about it.

THREEL
04-05-2009, 06:02 AM
Interesting, toby, thanks. :thumbsup:
I suppose seeing is believing. Certainly can't deny the evidence, but it seems strange. Is this one of those quantum things? The duality of the nature of light as a particle and a wave, I wonder?
It just seems really odd that a car moving forward would not appear most solid at its front. I don't know how many times though I've seen such pictures and simply never thought about it.

Sort of like when a car, or truck wheel is rotating forward quickly, the motion blur will actually appear to be in reverse at times. It's an optical illusion.

toby
04-07-2009, 07:27 PM
Think of it in terms of classic aa passes; low aa will sample the object 5 times giving you 5 steps of motion blur. If the object doesn't change speed, all the passes will be evenly spaced, no bias towards either end.

Maybe it's because we only see motion blur explicitly when it's done artistically for a still, that we always picture it as trailing? You know, like silver surfer. Or maybe because everything else trails fast moving things, like flags or exhaust