POV-Ray : Newsgroups : povray.binaries.animations : motion blur test Server Time
1 May 2024 20:57:50 EDT (-0400)
  motion blur test (Message 1 to 10 of 21)  
Goto Latest 10 Messages Next 10 Messages >>>
From: Tek
Subject: motion blur test
Date: 11 Nov 2010 11:20:02
Message: <web.4cdc16a1920b620ecaa39c860@news.povray.org>
Since I'm unlikely to ever go back to Megapov 1.21, and Pov 3.7 beta doesn't
have motion blur, I decided to write a scene file to blend together multiple
frames.

I also wanted to use low quality 2-sample focal blur, so that when the frames
blend together I'll effectively get 20-samples, but I ran into some problems
because the samples are in the same place on every frame, as discussed here:
http://news.povray.org/povray.beta-test/thread/%3Cweb.4cdaa135f1635598caa39c860%40news.povray.org%3E/

Anyway this is a little test animation I knocked together, it looks kinda cool
and it's very easy to do motion-blur like this.

--
Tek
http://evilsuperbrain.com


Post a reply to this message


Attachments:
Download 'anim_test.avi.dat' (1079 KB)

From: Christian Froeschlin
Subject: Re: motion blur test
Date: 11 Nov 2010 15:35:07
Message: <4cdc537b$1@news.povray.org>
Tek wrote:

> Anyway this is a little test animation I knocked together, it looks kinda cool
> and it's very easy to do motion-blur like this.

It looks good. Do you render extra frames for that or do you
just smooth over regular frames so it is basically a general
video postprocessing step?


Post a reply to this message

From: Le Forgeron
Subject: Re: motion blur test
Date: 11 Nov 2010 16:07:22
Message: <4cdc5b0a@news.povray.org>
Le 11/11/2010 17:15, Tek nous fit lire :
> Since I'm unlikely to ever go back to Megapov 1.21, and Pov 3.7 beta doesn't
> have motion blur

Motion blur... is there a real benefit to allow to include light(s) in
motion blur ? really ?
(I just had look at the megapov code for motion blur, and it might be
simpler without lights... just thinking about it currently as a kind of
CSG (so I'm pervert, it would be a very special CSG))
(lights, not such high ambient textured objects that might appears as
luminous)


Post a reply to this message

From: Kenneth
Subject: Re: motion blur test
Date: 11 Nov 2010 17:45:01
Message: <web.4cdc70ea7450cbb2196b08580@news.povray.org>
That's very nice! The focal-blur trick seems to work quite well; I can't 'see'
any of the 'static-position' blur noise you mentioned. I assume you *averaged* a
bunch of frames together for each motion-blurred one. (That's the way I do it.)
How many individual frames per final frame?

I was wondering what kind of code (clock-driven?) you used to get the gray
mirrored cube to rotate at different speeds. Very effective.

The only drawback I've noticed when averaging frames is that any scene elements
that are supposed to be *really* bright (like looking directly at a
motion-blurred lightbulb) will end up being smeared out to a dimmer, translucent
streak (when the same thing captured on *film* would be a very bright streak.)
Because of the inherent averaging of the 'moving' lightbulb with its dimmer
background. Someone on the newsgroups pointed this out to me awhile ago; but
most of the time, averaging-of-frames works quite well.

From what I understand, MegaPOV's motion blur (of objects) is done a different
way--so that bright lights produce a bright streak. I don't know for sure, as
I've never used it.

Ken


Post a reply to this message

From: Edouard
Subject: Re: motion blur test
Date: 11 Nov 2010 19:15:00
Message: <web.4cdc86867450cbb2ed3aba6c0@news.povray.org>
"Kenneth" <kdw### [at] earthlinknet> wrote:

> The only drawback I've noticed when averaging frames is that any scene elements
> that are supposed to be *really* bright (like looking directly at a
> motion-blurred lightbulb) will end up being smeared out to a dimmer, translucent
> streak (when the same thing captured on *film* would be a very bright streak.)
> Because of the inherent averaging of the 'moving' lightbulb with its dimmer
> background. Someone on the newsgroups pointed this out to me awhile ago; but
> most of the time, averaging-of-frames works quite well.

Not if you set the output file format to HDR (radience) or OpenEXR - the
highlights are preserved correctly with those. That's how I do my all frame
averaging.

> Ken

Cheers,
Edouard.


Post a reply to this message

From: Kenneth
Subject: Re: motion blur test
Date: 11 Nov 2010 22:35:01
Message: <web.4cdcb51f7450cbb2196b08580@news.povray.org>
"Edouard" <pov### [at] edouardinfo> wrote:
> "Kenneth" <kdw### [at] earthlinknet> wrote:
>
> > The only drawback I've noticed when averaging frames is that any scene
> > elements that are supposed to be *really* bright (like looking directly at
> > a motion-blurred lightbulb) will end up being smeared out to a dimmer,
> > translucent streak...
>
> Not if you set the output file format to HDR (radience) or OpenEXR - the
> highlights are preserved correctly with those. That's how I do my all frame
> averaging.

Interesting, and good to know.  (Alas, I'm still muddling along in v3.6.1; no
HDR or OpenEXR output.) I should have mentioned that my own frame averaging is
done within POV-Ray (by simply applying the static images to a box for
averaging/re-rendering, using ambient 1.0)

For HDR/OpenEXR averaging or blending, do you use an external image editor like
Photoshop? Or can the averaging be done in POV-Ray like I do (using the
appropriate beta version, of course) and still retain those highlights and
dynamic range? I.e., has the *average* pattern in POV-Ray been changed as well,
to retain the higher dynamic range of such averaged images?

Ken


Post a reply to this message

From: Tek
Subject: Re: motion blur test
Date: 12 Nov 2010 05:15:01
Message: <web.4cdd12d37450cbb2caa39c860@news.povray.org>
"Edouard" <pov### [at] edouardinfo> wrote:
> "Kenneth" <kdw### [at] earthlinknet> wrote:
>
> > The only drawback I've noticed when averaging frames is that any scene elements
> > that are supposed to be *really* bright (like looking directly at a
> > motion-blurred lightbulb) will end up being smeared out to a dimmer, translucent
> > streak (when the same thing captured on *film* would be a very bright streak.)
> > Because of the inherent averaging of the 'moving' lightbulb with its dimmer
> > background. Someone on the newsgroups pointed this out to me awhile ago; but
> > most of the time, averaging-of-frames works quite well.
>
> Not if you set the output file format to HDR (radience) or OpenEXR - the
> highlights are preserved correctly with those. That's how I do my all frame
> averaging.

Good point, I'd forgotten about that! Yeah HDR is kind of important. It's not
obvious in this scene because only the highlight is brighter than white.


Post a reply to this message

From: Tek
Subject: Re: motion blur test
Date: 12 Nov 2010 05:20:01
Message: <web.4cdd14857450cbb2caa39c860@news.povray.org>
"Kenneth" <kdw### [at] earthlinknet> wrote:
> That's very nice! The focal-blur trick seems to work quite well; I can't 'see'
> any of the 'static-position' blur noise you mentioned. I assume you *averaged* a
> bunch of frames together for each motion-blurred one. (That's the way I do it.)
> How many individual frames per final frame?

There's 10 frames blended together into each frame.

The static blur isn't obvious with the camera moving around so much, but if you
pause it you'll see that the motion blur looks quite smooth compared to the very
noisy focal blur. This is despite the fact that I used 2 samples for focal blur,
so it should look twice as smooth. The problem is that the 2 samples are nearly
always in the same place (despite me messing with bokeh to try to seperate
them), so the finished shot only looks as good as about 4 blur samples, rather
than the 20 I wanted.

I'm honestly tempted to render the scene bigger and jiggle the camera around
using a non-orthogonal matrix, to effectively create my desired image at a
different 2D position every frame, which I can then crop to remove the jiggle,
thereby randomising the focal blur samples. But it's kind of an ugly work round
to change what is, after all, just a random seed.


Post a reply to this message

From: Tek
Subject: Re: motion blur test
Date: 12 Nov 2010 05:25:00
Message: <web.4cdd15c77450cbb2caa39c860@news.povray.org>
Le_Forgeron <jgr### [at] freefr> wrote:
> Le 11/11/2010 17:15, Tek nous fit lire :
> > Since I'm unlikely to ever go back to Megapov 1.21, and Pov 3.7 beta doesn't
> > have motion blur
>
> Motion blur... is there a real benefit to allow to include light(s) in
> motion blur ? really ?
> (I just had look at the megapov code for motion blur, and it might be
> simpler without lights... just thinking about it currently as a kind of
> CSG (so I'm pervert, it would be a very special CSG))
> (lights, not such high ambient textured objects that might appears as
> luminous)

Lights would be good, for vehicles. Though it's also worth bearing in mind that
I want to use radiosity in animations, which I'm pretty sure doesn't get motion
blurred in megapov. TBH I'd always find a situation where I want to blend frames
myself unless pov could literally have multiple versions of the entire scene and
randomly jitter rays through time. Just doing a subset of features will never be
enough.

I guess what I'm saying is I don't think the official version of pov should have
a feature that's fussy about what other features it works with, it should be one
homogeneous whole. (Talking of which, can we get radiosity illuminating media?)


Post a reply to this message

From: Tek
Subject: Re: motion blur test
Date: 12 Nov 2010 05:30:01
Message: <web.4cdd16767450cbb2caa39c860@news.povray.org>
Christian Froeschlin <chr### [at] chrfrde> wrote:
> Tek wrote:
>
> > Anyway this is a little test animation I knocked together, it looks kinda cool
> > and it's very easy to do motion-blur like this.
>
> It looks good. Do you render extra frames for that or do you
> just smooth over regular frames so it is basically a general
> video postprocessing step?

There's 10 frames per final frame, blended together. The original animation was
1000 frames, the final one is 100. There's not really any way to get this effect
by blending without extra frames (though you can get close with some good motion
interpolation software, which I don't have).


Post a reply to this message

Goto Latest 10 Messages Next 10 Messages >>>

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.