John VanSickle <evi### [at] hotmailcom> wrote:
> The only drawback I see is that the motion of nearby clumps of grass
> should vary less than it does.  The amount of motion at any point looks
> fairly good, but the distribution of the randomness is a bit tight.
>
> So the small-scale turbulence should be lower, and the large-scale
> turbulence should be higher.

Interesting, and something I didn't see. (I ran a bunch of tests, varying one
thing and another, and wasn't really happy with *any* of them, except the final
one I posted.) I'll take a more critical look at those aspects. There are
actually *three* things that could be altered for each of the two bumps 'wind
turbulences' I used, all of which interact (visually): the intensity,
the size of the bumps pattern, and its speed (relative to the other pattern, as
well as the absolute speeds of both.) From your comments, I take it that the
*intensities* of the two wind patterns could use some relative adjustment. I'll
give that a try.

My own basic understanding (opinion??) of such windy grass is that the higher
the wind speed, the more 'chaotic' it all becomes.  But that's just a guess; it
could be the opposite!

Ken

Post a reply to this message

Looks pretty realistic to me! The camera movement's very good is it 
procedural or following a spline? I could swear it's rocking with the camera 
man's footsteps

-- 
Tek
http://evilsuperbrain.com

"Kenneth" <kdw### [at] earthlinknet> wrote in message 
news:web.4c89cc1ba3a1e657196b08580@news.povray.org...
> See
>
http://news.povray.org/povray.binaries.images/thread/%3Cweb.4c89c03947870ce5196b08580%40news.povray.org%3E/
>
> for notes on this scene.
>
> 640X480, with some motion blur and camera shake. Compressed with the Xvid 
> MPEG
> compressor. Sorry for the somewhat small image size; my original animation 
> was
> actually done at 1024 X 768 (without AA), but that produced a *large* MPEG 
> file
> (MPEG doesn't seem to like all those constantly-moving grass blades.)
>
> I let POV-Ray do the downsizing (as a post-processed copying step), using 
> AA at
> that stage, plus image_map interpolation. (The whole subject of using AA 
> while
> downsizing images is an interesting one--which I'll leave 'til a later 
> date!)
> The 640X480 result was OK, but a bit 'smooth' (or a bit blurry, take your 
> pick.)
> Xvid has some nice animation 'filters' built-in, so I used 'sharpen' to 
> bring
> back some of the image detail.  Xvid has its own downsizing filter, too, 
> but I
> decided to give the chore to POV, in order to experiment.
>
> Ken W.
>

Post a reply to this message

On 9/17/2010 5:58 AM, Kenneth wrote:
> John VanSickle<evi### [at] hotmailcom>  wrote:
>> The only drawback I see is that the motion of nearby clumps of grass
>> should vary less than it does.  The amount of motion at any point looks
>> fairly good, but the distribution of the randomness is a bit tight.
>>
>> So the small-scale turbulence should be lower, and the large-scale
>> turbulence should be higher.
>
> Interesting, and something I didn't see. (I ran a bunch of tests, varying one
> thing and another, and wasn't really happy with *any* of them, except the final
> one I posted.) I'll take a more critical look at those aspects. There are
> actually *three* things that could be altered for each of the two bumps 'wind
> turbulences' I used, all of which interact (visually): the intensity,
> the size of the bumps pattern, and its speed (relative to the other pattern, as
> well as the absolute speeds of both.) From your comments, I take it that the
> *intensities* of the two wind patterns could use some relative adjustment. I'll
> give that a try.

It may be that you need turbulence at a spatial frequency below even the 
lowest one you've put into the sim.

Regards,
John

Post a reply to this message

"Tek" <tek### [at] evilsuperbraincom> wrote:
> Looks pretty realistic to me!

Thanks.  I've discovered, over the course of animating various scenes, that a
'bumpy' camera really adds a GREAT deal of realism to even the most mundane
animation. And it has other benefits: Since I usually incorporate some motion
blur (i.e., just 5-to-10 averaged-together frames to get a final one), AND I
usually run my animation without AA (to save MUCH time), the bumpy camera helps
hide the inherent pixel aliasing. For *most* composite frames, anyway. The
result isn't as sharp and clean as it could be, but I can usually live with
that.

> The camera movement's very good is it procedural or following a spline? I
> could swear it's rocking with the camera man's footsteps.

Nope, not a spline in this case, just some animated translation and rotation of
the camera. HOWEVER, the camera's bumpiness *is* done with a spline set-up--a
1-dimensional spline macro . I'm essentially using the spline as an array, just
a holder of a certain number of random values. The animation then picks out
interpolated points from the spline, which are then applied as an additional
small rotation amount of the camera, or as jitter of its position or look_at
point. It's a handy little macro, which I use for lots of different things, not
just camera jitter. I've finally gotten it into a form that's simple,
problem-free and logical in its use, so I'll be posting it soon.

The funny thing about this animated scene is that the cameraman's shadow is
missing! I never even considered that little detail, but it's obvious to me now.

Ken

Post a reply to this message

"Kenneth" <kdw### [at] earthlinknet> wrote in message 
news:web.4c94b67b40e56284196b08580@news.povray.org...
> HOWEVER, the camera's bumpiness *is* done with a spline set-up--a
> 1-dimensional spline macro . I'm essentially using the spline as an array, 
> just
> a holder of a certain number of random values.

Nice! Are the values actually random? Have you considered using perlin noise 
instead? (i.e. f_spotted(x,0,0))
Much of my time with POV has been spent finding interesting uses for perlin 
noise!

> The funny thing about this animated scene is that the cameraman's shadow 
> is
> missing! I never even considered that little detail, but it's obvious to 
> me now.


I had thought that, but the sun angle means it would have been out of shot 
for most of the animation.

-- 
Tek
http://evilsuperbrain.com

Post a reply to this message

"Kenneth" <kdw### [at] earthlinknet> wrote:
> See
>
http://news.povray.org/povray.binaries.images/thread/%3Cweb.4c89c03947870ce5196b08580%40news.povray.org%3E/
>

Very nice and realistic feel. I like the dust blowing over the scene. I do feel
towards the end the camera rocking should become less as the walking comes to an
end.

Joost

Post a reply to this message

"Joost" <nomail@nomail> wrote:

>
> Very nice and realistic feel. I like the dust blowing over the scene. I do feel
> towards the end the camera rocking should become less as the walking comes to an
> end.

Probably true--but the wind is so strong that it's knocking the cameraman about!
 ;-)

Post a reply to this message

"Tek" <tek### [at] evilsuperbraincom> wrote:
> "Kenneth" <kdw### [at] earthlinknet> wrote in message
> > ...I'm essentially using the spline as an array, just
> > a holder of a certain number of random values.
>
> Nice! Are the values actually random? Have you considered using perlin noise
> instead? (i.e. f_spotted(x,0,0))
> Much of my time with POV has been spent finding interesting uses for perlin
> noise!

That's an interesting idea; hadn't considered it.  (The macro just takes a
user-chosen seed value and creates its many random spline values from that.) I
don't exactly understand how I would pull a string of 'different' values out of
f_spotted(x,0,0) to create the spline with; would something like this work?
(Just a general example)...

f_spotted(200*counter,0,0) // counter is just some advancing integer

I.e., translating the pattern a fair distance each count, to get a new 'random'
value each time?

Ken

Post a reply to this message

Well I wouldn't bother using a spline. The values from spotted give a nice 
smooth randomly varying result, so I'd just use f_spotted(time*speed,0,0). 
Or if you want a 3D value something like 
<f_spotted(T,0,0),f_spotted(T,seed1,0),f_spotted(T,seed2)> where seed1 and 
seed2 randomly offset the pattern.

Alternatively you can use vturbulence(2,.5,1, T*x) which gives a vector 
result and is also perlin noise based. You can increase the number of 
octaves (1 in that example) to make it bumpier.

-- 
Tek
http://evilsuperbrain.com

"Kenneth" <kdw### [at] earthlinknet> wrote in message 
news:web.4c9744cd40e56284196b08580@news.povray.org...
> "Tek" <tek### [at] evilsuperbraincom> wrote:
>> "Kenneth" <kdw### [at] earthlinknet> wrote in message
>> > ...I'm essentially using the spline as an array, just
>> > a holder of a certain number of random values.
>>
>> Nice! Are the values actually random? Have you considered using perlin 
>> noise
>> instead? (i.e. f_spotted(x,0,0))
>> Much of my time with POV has been spent finding interesting uses for 
>> perlin
>> noise!
>
> That's an interesting idea; hadn't considered it.  (The macro just takes a
> user-chosen seed value and creates its many random spline values from 
> that.) I
> don't exactly understand how I would pull a string of 'different' values 
> out of
> f_spotted(x,0,0) to create the spline with; would something like this 
> work?
> (Just a general example)...
>
> f_spotted(200*counter,0,0) // counter is just some advancing integer
>
> I.e., translating the pattern a fair distance each count, to get a new 
> 'random'
> value each time?
>
> Ken
>
>
>
>

Post a reply to this message

John VanSickle wrote:
> The only drawback I see is that the motion of nearby clumps of grass 
> should vary less than it does. 

Yes. It looked more like water than air making the grass wave.


-- 
Darren New, San Diego CA, USA (PST)
    Quoth the raven:
        Need S'Mores!

Post a reply to this message