"Maetes" <nomail@nomail> wrote:
> Few ideas:
> let the robots sometimes think a while.
> or let them stop before grabbing, and choose another move.
> and move the camera, maybe slowly aroud the scene, but always a tiny bit.
>
> ma

I like that idea,  moving the camera will be simple.  Changing its mind will be
a bit more difficult.

Post a reply to this message

"Maetes" <nomail@nomail> wrote:
> "Clarence1898" <dle### [at] comcastnet> wrote:
> > Going through an old archive I found this animation I wrote 20 years ago.
>
>
> Looks really really cool :)
> Cant wait to see Version 2.0
>
> Ma

This is the reason I made the change, so I could run multiple games at the same
time.   I think it is a bit amusing watching the robots bob up and down.  I have
a question about how everyone else renders animations.  I wrote the original
scene in 2003, probably on a pentium cpu using povray 3.5.  It was terribly
slow.  When I got another pc, I wrote a program to render frames on both pc's.
When I upgraded to pc's with multiple cpus I changed the program to use multiple
instances of povray on each pc.  With 3.7 I changed the program to use the work
thread option,  but I found that using multiple instances of povray still gave
me better throughput.  Four years ago after recovering from covid, I splurged
and  purchased an intel I9 10900k pc with 10 cores and 20 logical processors.  I
know it sounds ridiculous but running 20 instances of povray was 4 times faster
than running 1 instance with 20 threads.  Individual elapsed times were slower
but total throughput was 4 times as many frames per minute.  I know that this is
low resolution (720x400) and a very simple scene so I tested several
resolutions: 1920x1080 was twice as fast, 3840x2160 was still 60% faster.  I'm
sure more complicated scenes would decrease the difference.  Still I would like
to know how everyone does their rendering.

Post a reply to this message

Attachments:
Download 'robot4.mp4.dat' (4660 KB)

"Clarence1898" <dle### [at] comcastnet> wrote:
> Going through an old archive I found this animation I wrote 20 years ago.  I
> wanted to make a few additions but couldn't believe the code I wrote.  Decided I
> could do a better job 2nd time around.  The rewrite isn't finished yet so still
> a work in progress.

I'm like you going though my old archives. Have you found any code that you say
to yourself 'Did I really write that!?'.

I like the animation after I got it to play. I down loaded it and tried window
media player, it came up with screen 'needs a new codex 99 cents'
Where did the game come from? A while back I was testing ways to position pieces
on the chess board and taken pieces placed around it. I got online and found a
lots of tournament games (which I didn't bookmark their locations)

Have Fun!

Post a reply to this message

"Leroy" <whe### [at] gmailcom> wrote:
> "Clarence1898" <dle### [at] comcastnet> wrote:
> > Going through an old archive I found this animation I wrote 20 years ago.  I
> > wanted to make a few additions but couldn't believe the code I wrote.  Decided I
> > could do a better job 2nd time around.  The rewrite isn't finished yet so still
> > a work in progress.
>
> I'm like you going though my old archives. Have you found any code that you say
> to yourself 'Did I really write that!?'.
>
> I like the animation after I got it to play. I down loaded it and tried window
> media player, it came up with screen 'needs a new codex 99 cents'
> Where did the game come from? A while back I was testing ways to position pieces
> on the chess board and taken pieces placed around it. I got online and found a
> lots of tournament games (which I didn't bookmark their locations)
>
> Have Fun!

I have looked at my old code so many times and thought what was I on when I
wrote that.   Not only my personal code but code I got paid for.   The games I
used came from chessgames.com 1000 Best Short Games here:

https://www.chessgames.com/perl/chesscollection?cid=1034766

I used the short games to keep the video size below 5000K.

I always have fun.

Post a reply to this message

"Clarence1898" <dle### [at] comcastnet> wrote:
>  Still I would like
> to know how everyone does their rendering.

I use Raspberry Pis for it.
One is my standard computer for all days work, and another I turn on when I have
long movies to render.
Connection via SSH or virtual desktop.
That they need 1 or 2 days for a 2000 frames HD Moon- or Mars-Video doesnt
matter to me.

Pis are cheap, small, silent, energy-efficient.
And when there is nothing to render, you can use it as Media-Server or anything
else.

ma

Post a reply to this message

"Clarence1898" <dle### [at] comcastnet> wrote:
> "Maetes" <nomail@nomail> wrote:
>
> Still I would like to know how everyone does their rendering.

[running Windows 10]

Hi!

I have been rendering POV-ray animations for many years, all the way back to
Windows 98 days, when machines had only single cores. Now I have a Corei7
machine with 8 cores/16 threads. (Quite an improvement, ha.)

To be honest, I was not even aware that multiple instances of the app could be
run *successfully*...that is, with an overall speed improvement for a scene or
an animation...since it has always been my understanding that even a single
instance of POV-ray uses most of the resources of the entire machine for
rendering. At least, that's the way it used to be on my various single-core
machines through the years. (And AFAIU, initially *parsing* a scene uses only
one thread, regardless of how many threads are available.)

But the faster results of your multi-instance use are a real surprise! I will
have to give that a try, to see what the speed improvement might be for my own
animations.

There remains a question, though: When running multiple instances of the app on
multi core/multi thread machines, does each instance's scene parsing use it's
*own* single thread to do so? I.e., with 3 instances running (for example), are
there now 3 threads dedicated to parsing the (3) scenes? Or just one thread as
is typical-- split up between the 3 instances?

Post a reply to this message

"Maetes" <nomail@nomail> wrote:
> "Clarence1898" <dle### [at] comcastnet> wrote:
> >  Still I would like
> > to know how everyone does their rendering.
>
> I use Raspberry Pis for it.
> One is my standard computer for all days work, and another I turn on when I have
> long movies to render.
> Connection via SSH or virtual desktop.
> That they need 1 or 2 days for a 2000 frames HD Moon- or Mars-Video doesnt
> matter to me.
>
> Pis are cheap, small, silent, energy-efficient.
> And when there is nothing to render, you can use it as Media-Server or anything
> else.
>
> ma

Yes the Pi is a very nice machine.  I run Hercules, an IBM mainframe emulator on
a pi 5.  More ram, aux storage, and faster than the 3081 I worked on in the
1980's.

Post a reply to this message

"Kenneth" <kdw### [at] gmailcom> wrote:
> "Clarence1898" <dle### [at] comcastnet> wrote:
> > "Maetes" <nomail@nomail> wrote:
> >
> > Still I would like to know how everyone does their rendering.
>
> [running Windows 10]
>
> Hi!
>
> I have been rendering POV-ray animations for many years, all the way back to
> Windows 98 days, when machines had only single cores. Now I have a Corei7
> machine with 8 cores/16 threads. (Quite an improvement, ha.)
>
> To be honest, I was not even aware that multiple instances of the app could be
> run *successfully*...that is, with an overall speed improvement for a scene or
> an animation...since it has always been my understanding that even a single
> instance of POV-ray uses most of the resources of the entire machine for
> rendering. At least, that's the way it used to be on my various single-core
> machines through the years. (And AFAIU, initially *parsing* a scene uses only
> one thread, regardless of how many threads are available.)
>
> But the faster results of your multi-instance use are a real surprise! I will
> have to give that a try, to see what the speed improvement might be for my own
> animations.
>
> There remains a question, though: When running multiple instances of the app on
> multi core/multi thread machines, does each instance's scene parsing use it's
> *own* single thread to do so? I.e., with 3 instances running (for example), are
> there now 3 threads dedicated to parsing the (3) scenes? Or just one thread as
> is typical-- split up between the 3 instances?

When I started using multiprocessors I noticed the performance advantage of
multiple instances, but it was on a much smaller scale.  It was only with more
and faster processors that it became noticeable.  With my first quad-core system
the difference was about 15% improvement.  With a 20 cpu i9 the difference is up
to 60%.  For more complex scenes the smaller the difference, but still
noticeable.

As I understand it, each instance is completely independent from the others.
Each instance will have its own threads.

Post a reply to this message

hi,

"Clarence1898" <dle### [at] comcastnet> wrote:
> ...  Still I would like to know how everyone does their rendering.

I have an "elderly" i5 dedicated to renders.  I have not tried running multiple
instances[*] but like your time-savings observations.

[*] played briefly, some years ago, with multiple version 3.6, even tried to get
3.5 to build/run :-).

and thanks for including the ffmpeg command-line too when you posted robot1
(3.4M is small).


regards, jr.

Post a reply to this message

Le 2025-10-05 à 20:04, Clarence1898 a écrit :
> "Kenneth" <kdw### [at] gmailcom> wrote:
>> "Clarence1898" <dle### [at] comcastnet> wrote:
>>> "Maetes" <nomail@nomail> wrote:
>>>
>>> Still I would like to know how everyone does their rendering.
>>
>> [running Windows 10]
>>
>> Hi!
>>
>> I have been rendering POV-ray animations for many years, all the way back to
>> Windows 98 days, when machines had only single cores. Now I have a Corei7
>> machine with 8 cores/16 threads. (Quite an improvement, ha.)
>>
>> To be honest, I was not even aware that multiple instances of the app could be
>> run *successfully*...that is, with an overall speed improvement for a scene or
>> an animation...since it has always been my understanding that even a single
>> instance of POV-ray uses most of the resources of the entire machine for
>> rendering. At least, that's the way it used to be on my various single-core
>> machines through the years. (And AFAIU, initially *parsing* a scene uses only
>> one thread, regardless of how many threads are available.)
>>
>> But the faster results of your multi-instance use are a real surprise! I will
>> have to give that a try, to see what the speed improvement might be for my own
>> animations.
>>
>> There remains a question, though: When running multiple instances of the app on
>> multi core/multi thread machines, does each instance's scene parsing use it's
>> *own* single thread to do so? I.e., with 3 instances running (for example), are
>> there now 3 threads dedicated to parsing the (3) scenes? Or just one thread as
>> is typical-- split up between the 3 instances?
> 
> When I started using multiprocessors I noticed the performance advantage of
> multiple instances, but it was on a much smaller scale.  It was only with more
> and faster processors that it became noticeable.  With my first quad-core system
> the difference was about 15% improvement.  With a 20 cpu i9 the difference is up
> to 60%.  For more complex scenes the smaller the difference, but still
> noticeable.
> 
> As I understand it, each instance is completely independent from the others.
> Each instance will have its own threads.
> 
> 

When using POV-Ray version 3.6 and earlier, you couldn't use more than a 
single core/thread. So, back then, using several instances was the only 
way to use all of your cores on a multi-core system.
The main feature of version 3.7, and the reason it was developed, is the 
ability to use all the cores of modern CPUs.

Post a reply to this message