Motion blur thread

[salmonmoose]

which frame are these renders from?

if it's frame 2 then lightwave has got it right, but it appears that at least the max based ones are from frame 3. I've never seen motion blur predict the motion of an object, are you sure you havn't set the animation up on frams 0,1,2 in max?

[Morbid Angel]

im sorry for not clarifying it, in all cases i rendered frame 2.

[ulukai33]

It's a bit off-topic, but finalRender stage-1 has great 3d motion blur. :buttrock:

[salmonmoose]

Do you want to check your results again then?

Remember, max starts it's framecount at 0, lightwave starts it at 1. which would explain why it looks broken.

Try max at frame 1, lightwave at frame 2 :)

[MatthiasBuehlmann]

not lightwave but mentalray and VRay Interval Center at 0.5 show the true effect. You're rendering the 2nd Frame. In real world this means, that the shutter of the camera opens at this moment and stays open for some ms (somewher between frame 2 and 3) and the camera captures all the action during this time. So you can see in frame 2 the action that happens between frame 2 and 3 . . .

[CapnPanic]

[QUOTE=Morbid Angel]in the real world where camera doesnt know where the object will go at frame 3 it will only capture blur from frame 1-2, so in this conclusion the lightwave's blur is correct and everything else isnt.[QUOTE]

Okay, so i just wrote a big long explaination of how this is wrong and how motion blur actually works, but the damned server was 'too busy' and destroyed my reply. sigh.

Basically it just amounts to cameras working differently then CG and if you set the scanline multipass duration to .5 you will get the correct result and all the images above are incorrect.

If anyone is interested in the complicated explaination, say so, and when I am less pissed off at CGTalk i can come back and elaborate.

[Morbid Angel]

Do you want to check your results again then?

Remember, max starts it's framecount at 0, lightwave starts it at 1. which would explain why it looks broken.

Try max at frame 1, lightwave at frame 2 :)
Yeah but in all cases the sphere is in the center of the screen in 3 frames so it makes no difference, and besides lighwave 8 starts at frame 0...so i did everything right

[Morbid Angel]

Basically it just amounts to cameras working differently then CG and if you set the scanline multipass duration to .5 you will get the correct result and all the images above are incorrect.

If anyone is interested in the complicated explaination, say so, and when I am less pissed off at CGTalk i can come back and elaborate.
Please do and before you submit do copy to clipboard so you dont lose ur super explanation :D

[CapnPanic]

ok so a real film camera (35mm motion picture) generally has a half-disc shutter (180 degree shutter angle). This disc spins and exposes film for half of every rotation. Thinking of every rotation as the duration of a 'frame' this would amount to a blur duration of .5. The other .5 of the frame is used to scoot the film along so that the next frame is not overlapping the one just shot.

In a camera running at 24fps (standard film speed) this amounts to 1/48th of a second of exposure and then 1/48th of a second of non-exposure. Because of the length of the non-exposure, the end of a blur trail of an object on frame 1 should not meet up with the blur trail of the same object on frame 2 (if they were overlayed).

the scanline multipass does this correctly if it's duration is set to .5 instead of the default 1.0. This has the effect of temporally centering the blur on the frame being rendered. The effects rendered above, look like various ways of biasing the blur towards one frame or another, but none really match 'real-world' motion blur as it applies to film and most video shot at normal speeds.

But at the same time, it's all CG and therefore all fake, so it's really more a matter of what suits your needs or the needs of whoever you are doing the work for :)

(hopefully this was all that i covered in my original post)

[Bercon]

http://www.vantaa-vaskivuori.fi/%7Ejerrylil/images/MotionBlur.jpg


I dont have Lightwave nor Brazil so I can't test those, but which one of these spheres would be the most 'realistic'?

[Morbid Angel]

acording to capn panic the multipass with duration of 0.5 is right or the closest one...which means that vray is way off...strange...as well as lightwave because there is no settings there to specify interval center, its only shutter speed. The reason im posting this is because we had problems of cg elements sliding against real life footage. And also cg elements sliging against different cg elements rendered with different motion blurs. I think if I understand correctly then the moblur is captures on current frame aswell as the 0.25 of previos and next frame. But if thats the case there is a ghosting effect which is added to the moving object before the actual object....
Im confused still haha...

[CapnPanic]

the scanline object blur and fR 50% ones look right to me. The multipass looks to be offset just a bit, which is odd as i have never encountered that before. I might need to look into that a little more to see what it is doing

[Morbid Angel]

the scanline object blur and fR 50% ones look right to me. The multipass looks to be offset just a bit, which is odd as i have never encountered that before. I might need to look into that a little more to see what it is doing

Duration of 0.5 or 1.0 is just specifies the amount of moblur so I dont think it makes differents...as much as like vray interval center.

[Chinwagon]

My belief is that where the sphere is located in the rendered frame is dependant on whether you believe that the sphere displayed in your viewport is either at the start of the camera's shutter opening, the middle of it's exposure or at the point of the shutter closing. It's not about the camera predicting where the sphere goes but where it moves through the frame in the amount of time you set for the shutter the be opened for.
I believe that the viewport should show the middle point of an object's movement through frame, which means that Vray with a centre of 0.0 and finalRender with 50% phase are correct.
:¬)

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.