Hi everyone

Here's my latest Lego render. I've used focal blur in this scene but I 
think I need to use motion blur. Is motion blur available in latest 
version of POV-Ray or do I have to use MegaPov (or another patched 
version or POV-Ray)?

Bye for now

Reuben

=====================================
Download 3D Lego models and other resources from:
http://www.pearse.co.uk/lego
=====================================

Post a reply to this message

Attachments:
Download 'superman.png' (380 KB)

Preview of image 'superman.png'

Reuben Pearse wrote:

> Here's my latest Lego render. I've used focal blur in this scene but I 
> think I need to use motion blur. Is motion blur available in latest 
> version of POV-Ray

no, but you can simulate it by rendering an animation moving the
object in "sub-exposure" steps and then averaging the frames. This
also gives you the artistic freedom to give more weight to the final
frame if you want a more comic-like asymmetric blur for this scene.

Post a reply to this message

On 04/11/2012 14:30, Christian Froeschlin wrote:
> Reuben Pearse wrote:
>
>> Here's my latest Lego render. I've used focal blur in this scene but I
>> think I need to use motion blur. Is motion blur available in latest
>> version of POV-Ray
>
> no, but you can simulate it by rendering an animation moving the
> object in "sub-exposure" steps and then averaging the frames. This
> also gives you the artistic freedom to give more weight to the final
> frame if you want a more comic-like asymmetric blur for this scene.

Thanks for the info. Can you point me to any websites or tutorials that 
explain this technique in more detail? Some example POV code would be 
useful.

Reuben

Post a reply to this message

Reuben Pearse wrote:

> Thanks for the info. Can you point me to any websites or tutorials that 
> explain this technique in more detail? Some example POV code would be 
> useful.

Mostly I just remembered the thread "motion blur test" in these groups:

http://news.povray.org/povray.binaries.animations/thread/%3Cweb.4cdc16a1920b620ecaa39c860%40news.povray.org%3E/?mtop=352976

Post a reply to this message

On 4-11-2012 15:30, Christian Froeschlin wrote:
> Reuben Pearse wrote:
>
>> Here's my latest Lego render. I've used focal blur in this scene but I
>> think I need to use motion blur. Is motion blur available in latest
>> version of POV-Ray
>
> no, but you can simulate it by rendering an animation moving the
> object in "sub-exposure" steps and then averaging the frames. This
> also gives you the artistic freedom to give more weight to the final
> frame if you want a more comic-like asymmetric blur for this scene.

Ah! A little trick I might need. The animation ng is not regularly open 
on my system, but I should give it a look once in a while :-)

Thomas

Post a reply to this message

Reuben Pearse <reu### [at] pearsecouk> wrote:
> On 04/11/2012 14:30, Christian Froeschlin wrote:
> > Reuben Pearse wrote:
> >
> >> Here's my latest Lego render.

cool!

> >> I've used focal blur in this scene but I
> >> think I need to use motion blur. Is motion blur available in latest
> >> version of POV-Ray
> >
> > no, but you can simulate it by rendering an animation moving the
> > object in "sub-exposure" steps and then averaging the frames. This
> > also gives you the artistic freedom to give more weight to the final
> > frame if you want a more comic-like asymmetric blur for this scene.
>
> Thanks for the info. Can you point me to any websites or tutorials that
> explain this technique in more detail? Some example POV code would be
> useful.

Here's the scene I use when averaging a number of pre-rendered frames:

#version 3.7;
global_settings{assumed_gamma 1.0}
#default{finish{ambient 1}}

camera{
 orthographic
 right x up y
 location -z
}

#declare n_frames = 80;
#declare image_name_ = "superman"
#macro image_type_() png #end

plane{z,0
 pigment{
  average
  pigment_map{
   #for(V, 1, n_frames)
    [1
     #declare image_name = concat(image_name_, str(V, -(log(n_frames)/log(10) +
1), 0))
     image_map{image_type_() image_name}
     translate -(x+y)/2
    ]
   #end
  }
 }
}


Memory requirements can become pretty high when using large numbers of images,
especially if they are in the HDR format (but they work really well, especially
when rendering star fields).

Sam

Post a reply to this message

Hm, I'm not quite sure, if this alone solves the issue of having motion blur
with POV 3.7. Should this wonderful superman not having decreasing transparency
during the flight? If you only average pictures of the flight, you will probably
get more blur than expected, since POV doesn't know the sequence of the
pictures. May be, proper weights in the average map will help. But you are still
limited to 256 entries. An other approach is to have copies of the blurred
object with decreasing transparency in the scene. But to find the correct
positions and transparencies is the hard job. Christian Froeschlin gave an
example with his entry "glocken" at the IRTC april 2006, which shows the idea
but the difficulties likewise (rendered at a higher resolution then the original
entry).

Best regards,
Michael

Post a reply to this message

> Ah! A little trick I might need. The animation ng is not regularly open
> on my system, but I should give it a look once in a while :-)
>
> Thomas

Ah! That is why I get no comment from you to my first (very technically) entry
there. And I value your comments very much ;-)

Best regards,
Michael

Post a reply to this message

Le 04/11/12 1:44 PM, MichaelJF a écrit :
> Hm, I'm not quite sure, if this alone solves the issue of having motion blur
> with POV 3.7. Should this wonderful superman not having decreasing transparency
> during the flight? If you only average pictures of the flight, you will probably
> get more blur than expected, since POV doesn't know the sequence of the
> pictures. May be, proper weights in the average map will help. But you are still
> limited to 256 entries. An other approach is to have copies of the blurred
> object with decreasing transparency in the scene. But to find the correct
> positions and transparencies is the hard job. Christian Froeschlin gave an
> example with his entry "glocken" at the IRTC april 2006, which shows the idea
> but the difficulties likewise (rendered at a higher resolution then the original
> entry).
>
> Best regards,
> Michael
>
>

For the comicks like decreased transparency.
In this sample scene, all frames have the same weight, but it can be 
changed:
    #for(V, 1, n_frames)
     [V // the weight goes from 1 to n_frames.
      #declare image_name = concat(image_name_, str(V, 
-(log(n_frames)/log(10) +
1), 0))
      image_map{image_type_() image_name}
      translate -(x+y)/2
     ]
    #end

By replacing the [1 by [V, the first image have a weight of 1, the 
second a weight of 2,...
You can also use V/n_frames if you want to keep the weight parameter no 
larger than 1.
You can increase the last image further in a few ways:
Add some number to the last image's weight.
Use pow(V,2) or pow(V/n_frames,2) for the weight. You can use a power of 
3, 4 or some intermediate value if you want.
This assume that each frame is created in the direction of the movement 
and that you want the last frame to dominate.

The limit of 256 entries is not a problem unless you have a very long 
motion blur trail. If you use decreasing transparency, it's even less of 
a problem, even with rather long trails.



Alain

Post a reply to this message

Alain <kua### [at] videotronca> wrote:
> Le 04/11/12 1:44 PM, MichaelJF a écrit :
> > Hm, I'm not quite sure, if this alone solves the issue of having motion blur
> > with POV 3.7. Should this wonderful superman not having decreasing transparency
> > during the flight? If you only average pictures of the flight, you will probably
> > get more blur than expected, since POV doesn't know the sequence of the
> > pictures. May be, proper weights in the average map will help. But you are still
> > limited to 256 entries. An other approach is to have copies of the blurred
> > object with decreasing transparency in the scene. But to find the correct
> > positions and transparencies is the hard job. Christian Froeschlin gave an
> > example with his entry "glocken" at the IRTC april 2006, which shows the idea
> > but the difficulties likewise (rendered at a higher resolution then the original
> > entry).
> >
> > Best regards,
> > Michael
> >
> >
>
> For the comicks like decreased transparency.
> In this sample scene, all frames have the same weight, but it can be
> changed:
>     #for(V, 1, n_frames)
>      [V // the weight goes from 1 to n_frames.
>       #declare image_name = concat(image_name_, str(V,
> -(log(n_frames)/log(10) +
> 1), 0))
>       image_map{image_type_() image_name}
>       translate -(x+y)/2
>      ]
>     #end
>
> By replacing the [1 by [V, the first image have a weight of 1, the
> second a weight of 2,...
> You can also use V/n_frames if you want to keep the weight parameter no
> larger than 1.
> You can increase the last image further in a few ways:
> Add some number to the last image's weight.
> Use pow(V,2) or pow(V/n_frames,2) for the weight. You can use a power of
> 3, 4 or some intermediate value if you want.
> This assume that each frame is created in the direction of the movement
> and that you want the last frame to dominate.
>
> The limit of 256 entries is not a problem unless you have a very long
> motion blur trail. If you use decreasing transparency, it's even less of
> a problem, even with rather long trails.
>
>
>
> Alain

Yes, your proposal is much more detailed than mine. In fact I only mentioned the
general idea that simply averaging pictures of a flight will not yield a motion
blurred picture and proper weights are needed. But I think you have to play
around a lot with the weights and the number of images to find a proper image.

Best regards,
Michael

Post a reply to this message

Am 04.11.2012 20:46, schrieb MichaelJF:

> Yes, your proposal is much more detailed than mine. In fact I only mentioned the
> general idea that simply averaging pictures of a flight will not yield a motion
> blurred picture and proper weights are needed.

As a matter of fact, for /realistic/ motion blur simple averaging of the 
images /is/ the proper method. Whike the "opaque object with transparent 
trail" look is a common staple - probably invented to give the 
impression of motion while at the same time preserving object detail - 
it is outright wrong.

Post a reply to this message

On 4-11-2012 20:02, MichaelJF wrote:
>
>> Ah! A little trick I might need. The animation ng is not regularly open
>> on my system, but I should give it a look once in a while :-)
>>
>> Thomas
>
> Ah! That is why I get no comment from you to my first (very technically) entry
> there. And I value your comments very much ;-)

<blush> I shall have to experiment with this then... ;-)

Thomas

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> Am 04.11.2012 20:46, schrieb MichaelJF:
>
> > Yes, your proposal is much more detailed than mine. In fact I only mentioned the
> > general idea that simply averaging pictures of a flight will not yield a motion
> > blurred picture and proper weights are needed.
>
> As a matter of fact, for /realistic/ motion blur simple averaging of the
> images /is/ the proper method. Whike the "opaque object with transparent
> trail" look is a common staple - probably invented to give the
> impression of motion while at the same time preserving object detail -
> it is outright wrong.

Thanks, clipka... I knew I wasn't going crazy :) Maybe MJF is confusing motion
blur with tracers?

Post a reply to this message

clipka wrote:
> Am 04.11.2012 20:46, schrieb MichaelJF:
> 
>> Yes, your proposal is much more detailed than mine. In fact I only
>> mentioned the
>> general idea that simply averaging pictures of a flight will not yield

>> a motion
>> blurred picture and proper weights are needed.
> 
> As a matter of fact, for /realistic/ motion blur simple averaging of th
e
> images /is/ the proper method. Whike the "opaque object with transparen
t
> trail" look is a common staple - probably invented to give the
> impression of motion while at the same time preserving object detail -
> it is outright wrong.
> 
	The "opaque object with transparent trail" is also what you get
with a flash light synced on the rear curtain and a long(ish)
exposure time (if you don't sync on the rear curtain, the "trail"
will be in front of the object, or sometimes on both sides).

	For example:
http://www.apnphotographyschool.com/wp-content/uploads/2010/12/358x500xni
kon-speedlight-rear-curtain-sync.jpg.pagespeed.ic.qeHOKzR258.jpg

		Jerome
-- 
mailto:jeb### [at] freefr
http://jeberger.free.fr
Jabber: jeb### [at] jabberfr

Post a reply to this message

Attachments:
Download 'us-ascii' (1 KB)

"MichaelJF" <mi-### [at] t-onlinede> wrote:
> But you are still limited to 256 entries.

Actually, you're not. While it's true that any given _map is limited to 256
entries, you can nest maps to exceed that limitation. You're pretty much
averaging averaged pigments at that point :)

Post a reply to this message

"Samuel Benge" <stb### [at] hotmailcom> wrote:

> Thanks, clipka... I knew I wasn't going crazy :) Maybe MJF is confusing motion
> blur with tracers?

Sorry, I don't know what I mixed up, may be impressions from ads, I took for
photographies or something like Jerome presented. Tracers certainly not, because
I first had to look up the wikipedia what this could be... But is superman very
far away from them?

Yes, as ever clipka is right, as is clear, if one thinks a bit more before
posting...

Best regards,
Michael

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:

>
> <blush> I shall have to experiment with this then... ;-)
>
> Thomas

No need to blush;-) I only thought that my cheap trick could interest you
regarding your former profession.

Best regards,
Michael

Post a reply to this message

On 5-11-2012 20:13, MichaelJF wrote:
> No need to blush;-) I only thought that my cheap trick could interest you
> regarding your former profession.

Oh, absolutely. In fact I have been thinking about the contour line 
possibilities for a long time without getting very far. So, I see great 
possibilities in your technique.

But with regard of the motion blur proper, I have been thinking about 
how to use the averaging method (to replace the Megapov application) 
without loosing to much (render) time. One thing your comment above made 
me think of was (1) to isolate the moving object from the scene proper, 
(2) animate it to get motion blur on a transparent background and 
average the frames, and finally (3) average the blurred object image 
with the scene image.

While more work than with Megapov, I suppose that it might work, and it 
probably is the technique used originally by Megapov, but then embedded 
into the code.

Thomas

Post a reply to this message

Thomas de Groot <tho### [at] degrootorg> wrote:
> On 5-11-2012 20:13, MichaelJF wrote:
> > No need to blush;-) I only thought that my cheap trick could interest you
> > regarding your former profession.
>
> Oh, absolutely. In fact I have been thinking about the contour line
> possibilities for a long time without getting very far. So, I see great
> possibilities in your technique.

Thanks, I just investigate if exr-pictures can achieve an improvement after
having reached a certain approximation of the landscape with png.

> But with regard of the motion blur proper, I have been thinking about
> how to use the averaging method (to replace the Megapov application)
> without loosing to much (render) time. One thing your comment above made
> me think of was (1) to isolate the moving object from the scene proper,
> (2) animate it to get motion blur on a transparent background and
> average the frames, and finally (3) average the blurred object image
> with the scene image.
>
> While more work than with Megapov, I suppose that it might work, and it
> probably is the technique used originally by Megapov, but then embedded
> into the code.
>
> Thomas

Sounds great, I think, this is the way to go.

Best regards,
Michael

Post a reply to this message

> Thomas de Groot <tho### [at] degrootorg> wrote:
> > (3) average the blurred object image
> > with the scene image.

Second thought: average the images with the transparent background and layer the
result over the original scene only. Otherwise the original scene (background)
has a chance to come into the object in motion. Ok, may be only to a small,
hardly visible amount...

Best regards,
Michael

Post a reply to this message

Am 06.11.2012 20:42, schrieb MichaelJF:
> Thomas de Groot <tho### [at] degrootorg> wrote:
>> On 5-11-2012 20:13, MichaelJF wrote:
>>> No need to blush;-) I only thought that my cheap trick could interest you
>>> regarding your former profession.
>>
>> Oh, absolutely. In fact I have been thinking about the contour line
>> possibilities for a long time without getting very far. So, I see great
>> possibilities in your technique.
>
> Thanks, I just investigate if exr-pictures can achieve an improvement after
> having reached a certain approximation of the landscape with png.

While OpenEXR will give you an improvement, it would still be a 
comparatively poor choice, as its absolute precision varies considerably 
between dark and bright regions of an image, so you'd have different 
quality on high plateaus than in low plains.

For your purposes, any classic file format with a 16-bit color depth 
would be better suited, such as 16-bit png.

Post a reply to this message

Am 06.11.2012 21:17, schrieb MichaelJF:
>> Thomas de Groot <tho### [at] degrootorg> wrote:
>>> (3) average the blurred object image
>>> with the scene image.
>
> Second thought: average the images with the transparent background and layer the
> result over the original scene only. Otherwise the original scene (background)
> has a chance to come into the object in motion. Ok, may be only to a small,
> hardly visible amount...

No, you're perfectly right there: averaging is the wrong operation for 
this job; you do want an overlay operation.

Post a reply to this message

clipka <ano### [at] anonymousorg> wrote:
> While OpenEXR will give you an improvement, it would still be a
> comparatively poor choice, as its absolute precision varies considerably
> between dark and bright regions of an image, so you'd have different
> quality on high plateaus than in low plains.
>
> For your purposes, any classic file format with a 16-bit color depth
> would be better suited, such as 16-bit png.

Many thanks, it's just an experiment. And I see I have still much to learn,
again. But this is one reason I try to make pictures with POV, I learn. By
observation or even by being corrected for errors or misjudgements by this
wonderful community.

Best regards,
Michael

Post a reply to this message

On 6-11-2012 21:43, clipka wrote:
> Am 06.11.2012 21:17, schrieb MichaelJF:
>>> Thomas de Groot <tho### [at] degrootorg> wrote:
>>>> (3) average the blurred object image
>>>> with the scene image.
>>
>> Second thought: average the images with the transparent background and
>> layer the
>> result over the original scene only. Otherwise the original scene
>> (background)
>> has a chance to come into the object in motion. Ok, may be only to a
>> small,
>> hardly visible amount...
>
> No, you're perfectly right there: averaging is the wrong operation for
> this job; you do want an overlay operation.
>

Hmm yes. That is true. Not as straightforward as I thought. And if some 
element of the scene partly obscures the moving object both techniques 
won't work correctly. But otherwise an overlay would do.

We still need a megapov-type piece of code in the software.

Thomas

Post a reply to this message