 |
|
|
|
|
|
| |
| |
|
|
From: Nicolas Alvarez
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 27 Jan 2008 22:05:32
Message: <479d467c$1@news.povray.org>
|
|
|
| |
| |
|
|
Nicolas Alvarez escribió:
> I'm now rendering a 160x120 image to see the raytrace statistics at the
> end, and see if there were a lot of misses or not. Stay tuned...
Ray->Shape Intersection Tests Succeeded Percentage
Plane 3224478 124068 3.85
Torus 12394622 976314 7.88
Torus Bound 12394622 1072053 8.65
Cone/Cylinder 103967072 15537453 14.94
CSG Intersection 95235655 24004815 25.21
Sphere 48158288 13067745 27.13
CSG Union 28433556 12742408 44.81
Box 261303211 123172730 47.14
Light Buffer 418693292 219308071 52.38
Blob 96908476 51630734 53.28
Blob Bound 883968592 471672540 53.36
Bounding Box 166046777 100557601 60.56
Vista Buffer 2803512 1912181 68.21
Blob Component 317778794 254071826 79.95
CSG Merge 25841750 25837528 99.98
(table sorted by percentage)
Intersection wasn't as bad as I expected. Torus has a particularly low
value. Plane calculation is quite fast, so that one isn't important.
It's still possible that some manual bounding could improve things.
Post a reply to this message
|
|
| |
| |
|
|
From: Jan Dvorak
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 27 Jan 2008 22:20:05
Message: <479d49e5@news.povray.org>
|
|
|
| |
| |
|
|
Nicolas Alvarez napsal(a):
> Nicolas Alvarez escribió:
>> I'm now rendering a 160x120 image to see the raytrace statistics at
>> the end, and see if there were a lot of misses or not. Stay tuned...
>
> Ray->Shape Intersection Tests Succeeded Percentage
>
> Plane 3224478 124068 3.85
> Torus 12394622 976314 7.88
> Torus Bound 12394622 1072053 8.65
> Cone/Cylinder 103967072 15537453 14.94
> CSG Intersection 95235655 24004815 25.21
> Sphere 48158288 13067745 27.13
> CSG Union 28433556 12742408 44.81
> Box 261303211 123172730 47.14
> Light Buffer 418693292 219308071 52.38
> Blob 96908476 51630734 53.28
> Blob Bound 883968592 471672540 53.36
> Bounding Box 166046777 100557601 60.56
> Vista Buffer 2803512 1912181 68.21
> Blob Component 317778794 254071826 79.95
> CSG Merge 25841750 25837528 99.98
>
> (table sorted by percentage)
>
> Intersection wasn't as bad as I expected. Torus has a particularly low
> value. Plane calculation is quite fast, so that one isn't important.
>
> It's still possible that some manual bounding could improve things.
torus is tested only if its bound fails. It is bounded by a tight
difference of two cyliders.
Post a reply to this message
|
|
| |
| |
|
|
From: Nicolas Alvarez
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 27 Jan 2008 22:27:39
Message: <479d4bab$1@news.povray.org>
|
|
|
| |
| |
|
|
Jan Dvorak escribió:
> torus is tested only if its bound fails. It is bounded by a tight
> difference of two cyliders.
Interesting fact. I knew it, now that you mention it; but I didn't
remember (been a while since I read it).
Any idea why both torus and its bounding shape have such low hit percentage?
Post a reply to this message
|
|
| |
| |
|
|
From: Nicolas Alvarez
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 27 Jan 2008 22:31:31
Message: <479d4c93$1@news.povray.org>
|
|
|
| |
| |
|
|
> I tried with photons, and
> there were very long periods where the "photons shot" counter didn't
> change, which could mean POV-Ray is shooting lots of photons but
> "missing the object".
Photon count went up until a minute of shooting passed, then it got
stuck at 41294 photons shot. Been at that number for *20 minutes*.
There is clearly some bounding issue!
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Nicolas Alvarez napsal(a):
> Jan Dvorak escribió:
>> torus is tested only if its bound fails. It is bounded by a tight
>> difference of two cyliders.
>
> Interesting fact. I knew it, now that you mention it; but I didn't
> remember (been a while since I read it).
>
> Any idea why both torus and its bounding shape have such low hit
> percentage?
It seems that the number of torus bound tests is used to calculate the
success percentage. See the table, rays shot. The actual success rate
for torus in that table is 976314/1072053=91%
Post a reply to this message
|
|
| |
| |
|
|
From: Sven Littkowski
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 28 Jan 2008 01:27:13
Message: <479d75c1@news.povray.org>
|
|
|
| |
| |
|
|
Thanks!
Sven
"Jan Dvorak" <jan### [at] centrum cz> schrieb im Newsbeitrag
news:479d4019@news.povray.org...
> global_settings{
> photons{
> count image_width*image_height
> ...
> }
> ...
> }
>
> untested
> Sven Littkowski napsal(a):
>> Shooting the image with those keywords... Hmm, do you mean that I need to
>> put these two keywords inside the camera directive? To be honest, due the
>> lack of experience and knowledge with these two keywords, i would just
>> like to ask you to give an example where and how to use these two
>> keywords in a scene. I think, that should help me.
>>
>> Thanks, Jan.
>>
>> Sven
>>
>>
>>
>> "Jan Dvorak" <jan### [at] centrumcz> schrieb im Newsbeitrag
>> news:479cff7f@news.povray.org...
>>> You might decide to shoot image_width*image_height*some_constant
>>> photons. Some_constant==1 might work well but feel free to increase for
>>> high quality or decrease for faster render. Use the count keyword and
>>> watch POV-ray do its best.
>>> Sven Littkowski napsal(a):
>>>> Hmm... THAT sounds interesting and promising, Jan. I looked into the
>>>> POV help file, and see the keywords "image_width" and "image_height".
>>>> And also some examples, how to use them. I really think, this could
>>>> shorten the calculation time, if I can find out, how to use them
>>>> properly in connection with my photon map creation.
>>>>
>>>> Jan, can you give me some more advice in this? If you like, I could
>>>> post the scene file to you. Please let me know. Thanks,
>>>>
>>>> Sven
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> "Jan Dvorak" <jan### [at] centrumcz> schrieb im Newsbeitrag
>>>> news:479c31b4$1@news.povray.org...
>>>>> try using image_width to choose the photon count/spacing
>>>>> Sven Littkowski napsal(a):
>>>>>> Hi Ben,
>>>>>>
>>>>>> but even so, based on what I heard already, meanwhile, the photon map
>>>>>> would still take 6 days, right?
>>>>>>
>>>>>> Sven
>>>>>>
>>>>>>
>>>>>>
>>>>>> "Chambers" <ben### [at] pacificwebguycom> schrieb im Newsbeitrag
>>>>>> news:479bdf30$1@news.povray.org...
>>>>>>> Sven Littkowski wrote:
>>>>>>>> Hi.
>>>>>>>>
>>>>>>>> I have a scene, which already takes around 6 days to create the
>>>>>>>> photon map for, if it is done in the same resolution like the
>>>>>>>> remaining scene.
>>>>>>>>
>>>>>>>> I would like to know, if it is possible to use a lower resolution
>>>>>>>> (let's say: 400x300) for the photon map (with the "save_file"
>>>>>>>> option enabled), and then to stop and continue the scene in a high
>>>>>>>> resolution (now with the "load_file" option enabled).
>>>>>>> Sure, I do it myself. However, since the photon map is independent
>>>>>>> from the scene resolution, you can set the image size absurdly low
>>>>>>> (like +w1 +h1) and still get a full-quality photon map.
>>>>>>>
>>>>>>> --
>>>>>>> ...Ben Chambers
>>>>>>> www.pacificwebguy.com
>>
Post a reply to this message
|
|
| |
| |
|
|
From: Sven Littkowski
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 28 Jan 2008 01:32:21
Message: <479d76f5$1@news.povray.org>
|
|
|
| |
| |
|
|
Definatively, I will stay tunes! I am eager for the result, as well as I am
in such a big need for that photon map file!
Be assured, you will like what you see, as soon as you can see the image
itself.
The object(s) even allow you to set different day or night times (which
influences the light inside the scene), as well as you have influence on the
object self (such as rotor rotation and anges, open doors, etc.).
Best greetings,
Sven
"Nicolas Alvarez" <nic### [at] gmailisthebestcom> schrieb im
Newsbeitrag news:479d42ea$1@news.povray.org...
> [Posting to newsgroup since other people may be interested]
>
> Now that you sent me the files... The scene is quite slow to render
> (55pps) even without radiosity or photons. I tried with photons, and there
> were very long periods where the "photons shot" counter didn't change,
> which could mean POV-Ray is shooting lots of photons but "missing the
> object".
>
> What I'm suspecting is that POV-Ray is creating sub-optimal bounding
> boxes. I see a lot of difference objects on your scene. POV-Ray sometimes
> does a bad job bounding difference/intersection CSG.
>
> I'm now rendering a 160x120 image to see the raytrace statistics at the
> end, and see if there were a lot of misses or not. Stay tuned...
Post a reply to this message
|
|
| |
| |
|
|
From: Florian Brucker
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 28 Jan 2008 06:32:49
Message: <479dbd61$1@news.povray.org>
|
|
|
| |
| |
|
|
Hi Sven!
I haven't seen your scene or know anything about it, but it could be
that two of Mike's excellent tutorials could help you with your bounding
issues:
Mike's isosurface tutorial:
http://www.econym.demon.co.uk/isotut/
Mike's holes tutorial:
http://www.econym.demon.co.uk/holetut/index.htm
HTH,
Florian
Post a reply to this message
|
|
| |
| |
|
|
From: Sven Littkowski
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 28 Jan 2008 08:41:16
Message: <479ddb7c$1@news.povray.org>
|
|
|
| |
| |
|
|
Hi Jan,
I added that line to my photon global statement, and so far, I recognize
that POV-Ray behaves very different to how it was doing the map before.
Yes, there is a clear speed bonus, I recognized that when setting POV to a
resolution of 160x120. Now I have started at 3200x2400, and wait until some
result or crash comes. The blocks of photos are now much smaller (before:
ca. 40,000 before each break, now only 320). I will keep you up to date.
Sven
"Sven Littkowski" <sven [] jamaica-focus [] com> schrieb im Newsbeitrag
news:479d75c1@news.povray.org...
> Thanks!
>
> Sven
>
>
> "Jan Dvorak" <jan### [at] centrumcz> schrieb im Newsbeitrag
> news:479d4019@news.povray.org...
>> global_settings{
>> photons{
>> count image_width*image_height
>> ...
>> }
>> ...
>> }
>>
>> untested
>> Sven Littkowski napsal(a):
>>> Shooting the image with those keywords... Hmm, do you mean that I need
>>> to put these two keywords inside the camera directive? To be honest, due
>>> the lack of experience and knowledge with these two keywords, i would
>>> just like to ask you to give an example where and how to use these two
>>> keywords in a scene. I think, that should help me.
>>>
>>> Thanks, Jan.
>>>
>>> Sven
>>>
>>>
>>>
>>> "Jan Dvorak" <jan### [at] centrumcz> schrieb im Newsbeitrag
>>> news:479cff7f@news.povray.org...
>>>> You might decide to shoot image_width*image_height*some_constant
>>>> photons. Some_constant==1 might work well but feel free to increase for
>>>> high quality or decrease for faster render. Use the count keyword and
>>>> watch POV-ray do its best.
>>>> Sven Littkowski napsal(a):
>>>>> Hmm... THAT sounds interesting and promising, Jan. I looked into the
>>>>> POV help file, and see the keywords "image_width" and "image_height".
>>>>> And also some examples, how to use them. I really think, this could
>>>>> shorten the calculation time, if I can find out, how to use them
>>>>> properly in connection with my photon map creation.
>>>>>
>>>>> Jan, can you give me some more advice in this? If you like, I could
>>>>> post the scene file to you. Please let me know. Thanks,
>>>>>
>>>>> Sven
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> "Jan Dvorak" <jan### [at] centrumcz> schrieb im Newsbeitrag
>>>>> news:479c31b4$1@news.povray.org...
>>>>>> try using image_width to choose the photon count/spacing
>>>>>> Sven Littkowski napsal(a):
>>>>>>> Hi Ben,
>>>>>>>
>>>>>>> but even so, based on what I heard already, meanwhile, the photon
>>>>>>> map would still take 6 days, right?
>>>>>>>
>>>>>>> Sven
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> "Chambers" <ben### [at] pacificwebguycom> schrieb im Newsbeitrag
>>>>>>> news:479bdf30$1@news.povray.org...
>>>>>>>> Sven Littkowski wrote:
>>>>>>>>> Hi.
>>>>>>>>>
>>>>>>>>> I have a scene, which already takes around 6 days to create the
>>>>>>>>> photon map for, if it is done in the same resolution like the
>>>>>>>>> remaining scene.
>>>>>>>>>
>>>>>>>>> I would like to know, if it is possible to use a lower resolution
>>>>>>>>> (let's say: 400x300) for the photon map (with the "save_file"
>>>>>>>>> option enabled), and then to stop and continue the scene in a high
>>>>>>>>> resolution (now with the "load_file" option enabled).
>>>>>>>> Sure, I do it myself. However, since the photon map is independent
>>>>>>>> from the scene resolution, you can set the image size absurdly low
>>>>>>>> (like +w1 +h1) and still get a full-quality photon map.
>>>>>>>>
>>>>>>>> --
>>>>>>>> ...Ben Chambers
>>>>>>>> www.pacificwebguy.com
>>>
>
Post a reply to this message
|
|
| |
| |
|
|
From: Sven Littkowski
Subject: Re: Photon Map: Created at Lower Res than Scene?
Date: 28 Jan 2008 08:51:54
Message: <479dddfa$1@news.povray.org>
|
|
|
| |
| |
|
|
Thanks, Florian. I am gonna check 'em out.
Sven
"Florian Brucker" <tor### [at] torfboldcom> schrieb im Newsbeitrag
news:479dbd61$1@news.povray.org...
> Hi Sven!
>
> I haven't seen your scene or know anything about it, but it could be
> that two of Mike's excellent tutorials could help you with your bounding
> issues:
>
> Mike's isosurface tutorial:
> http://www.econym.demon.co.uk/isotut/
>
> Mike's holes tutorial:
> http://www.econym.demon.co.uk/holetut/index.htm
>
>
> HTH,
> Florian
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
