Hi,

I am finishing a set of scenes that make intensive use of radiosity as the 
light comes from outside of the space that I am designing. To allow more 
light to enter through the windows, I am applying radiosity in 2 steps:

1. Render without windows storing the radiosity info (save_file)
2. Render with windows and reading the radiosity info from the file 
(load_file).

I have some doubts about how storing radiosity info works. I would 
appreciate some lighting (in my mind):

1. What radiosity parameters have influence on the info that is stored in 
the file (all of them?)
2. What radiosity parameters have influence on the image if I am using 
load_file?
3. Does the info in the file overwrite the radiosity parameters in the SDL 
code?
4. Can I make the first step at a lower size than the second step?
5. Does it matter where I put the parameters load_file and save_file in the 
code (after the actual radiosity parameters, after them, ...?
6. If I get some artifacts in the first step, may I remove them in the 
second one?

Thanks for your help,
Zoly

Post a reply to this message

Zoly nous apporta ses lumieres en ce 2007/08/31 12:49:
> Hi,
> 
> I am finishing a set of scenes that make intensive use of radiosity as the 
> light comes from outside of the space that I am designing. To allow more 
> light to enter through the windows, I am applying radiosity in 2 steps:
> 
> 1. Render without windows storing the radiosity info (save_file)
Windows without the "glass" pane.
> 2. Render with windows and reading the radiosity info from the file 
> (load_file).
> 
> I have some doubts about how storing radiosity info works. I would 
> appreciate some lighting (in my mind):
> 
> 1. What radiosity parameters have influence on the info that is stored in 
> the file (all of them?)
All of them.
> 2. What radiosity parameters have influence on the image if I am using 
> load_file?
Any can have an effect, when there is a need to gather some extra samples.
> 3. Does the info in the file overwrite the radiosity parameters in the SDL 
> code?
NO.
> 4. Can I make the first step at a lower size than the second step?
Absolutely. You can also remove any AA.
> 5. Does it matter where I put the parameters load_file and save_file in the 
> code (after the actual radiosity parameters, after them, ...?
load_file
radiosity parameters
save_file
or
radiosity parameters
load_file
save_file

If you save_file before load_file, you may reload the data you just saved.
> 6. If I get some artifacts in the first step, may I remove them in the 
> second one?
The second step will gather additional samples if needed. They may, or not, 
remove or reduce those artefacts.
Use high quality parameters in the first pass, and more relaxed ones on the 
second pass.
In the first pass, remove any flat transparent, colorless, objects. Reflective 
objects should have the texture replaced by a plain white one IF they are far 
from bright surfaces: high ambient ones.
Keep shaped with ior or colored transparent objects as they actively change the 
radiosity samples. Also, keep non-flat reflective
> 
> Thanks for your help,
> Zoly 
> 
> 
If you also use photons, you must keep any object that is a target for the photons.

-- 
Alain
-------------------------------------------------
You know you've been raytracing too long when you have ever brought your 
computer to its knees by mistakenly launching 64 simultaneous frames to be 
traced, while trying to maximizing the benefits of parallelizing them.
Carsten Whimster

Post a reply to this message

Thank you for tips.
One clarification: may I remove AA from the first step as it will not affect 
the values stored in the radiosity file?

I also just realized with more tests that if I use the same radiosity 
parameters at both steps, the second one runs much faster. If I change 
'count', 'error_bound', 'recursion_limit' even to more relaxed values, it 
takes more than double of the time to render.

Rgds,
Zoly

"Alain" <ele### [at] netscapenet> wrote in message 
news:46d86fb9@news.povray.org...
> Zoly nous apporta ses lumieres en ce 2007/08/31 12:49:
>> Hi,
>>
>> I am finishing a set of scenes that make intensive use of radiosity as 
>> the light comes from outside of the space that I am designing. To allow 
>> more light to enter through the windows, I am applying radiosity in 2 
>> steps:
>>
>> 1. Render without windows storing the radiosity info (save_file)
> Windows without the "glass" pane.
>> 2. Render with windows and reading the radiosity info from the file 
>> (load_file).
>>
>> I have some doubts about how storing radiosity info works. I would 
>> appreciate some lighting (in my mind):
>>
>> 1. What radiosity parameters have influence on the info that is stored in 
>> the file (all of them?)
> All of them.
>> 2. What radiosity parameters have influence on the image if I am using 
>> load_file?
> Any can have an effect, when there is a need to gather some extra samples.
>> 3. Does the info in the file overwrite the radiosity parameters in the 
>> SDL code?
> NO.
>> 4. Can I make the first step at a lower size than the second step?
> Absolutely. You can also remove any AA.
>> 5. Does it matter where I put the parameters load_file and save_file in 
>> the code (after the actual radiosity parameters, after them, ...?
> load_file
> radiosity parameters
> save_file
> or
> radiosity parameters
> load_file
> save_file
>
> If you save_file before load_file, you may reload the data you just saved.
>> 6. If I get some artifacts in the first step, may I remove them in the 
>> second one?
> The second step will gather additional samples if needed. They may, or 
> not, remove or reduce those artefacts.
> Use high quality parameters in the first pass, and more relaxed ones on 
> the second pass.
> In the first pass, remove any flat transparent, colorless, objects. 
> Reflective objects should have the texture replaced by a plain white one 
> IF they are far from bright surfaces: high ambient ones.
> Keep shaped with ior or colored transparent objects as they actively 
> change the radiosity samples. Also, keep non-flat reflective
>>
>> Thanks for your help,
>> Zoly
> If you also use photons, you must keep any object that is a target for the 
> photons.
>
> -- 
> Alain
> -------------------------------------------------
> You know you've been raytracing too long when you have ever brought your 
> computer to its knees by mistakenly launching 64 simultaneous frames to be 
> traced, while trying to maximizing the benefits of parallelizing them.
> Carsten Whimster

Post a reply to this message

Zoly wrote:
> One clarification: may I remove AA from the first step as it will not affect 
> the values stored in the radiosity file?

  Of course! ...and don't forget to do the same with focal blur and area lights!

> I also just realized with more tests that if I use the same radiosity 
> parameters at both steps, the second one runs much faster. If I change 
> 'count', 'error_bound', 'recursion_limit' even to more relaxed values, it 
> takes more than double of the time to render.

   Yes, that's my observation also... I suppose it's because if the params 
are the same, it just has to load the rad data file (provided you have 
always_sample off).


--
Jaime

Post a reply to this message

Zoly nous apporta ses lumieres en ce 2007/08/31 16:05:
> Thank you for tips.
> One clarification: may I remove AA from the first step as it will not affect 
> the values stored in the radiosity file?
AA and focal blur can be removed as they don't affect the radiosity data.
area_light can also be replaced by normal lights, even if they alter radiosity 
samples from the penumbrae regions.
> 
> I also just realized with more tests that if I use the same radiosity 
> parameters at both steps, the second one runs much faster. If I change 
> 'count', 'error_bound', 'recursion_limit' even to more relaxed values, it 
> takes more than double of the time to render.
> 
> Rgds,
> Zoly
If you change the settings, exept for pretrace_start/pretrace_end and 
always_sample off, you force new sampling with the new parameters. Those samples 
will be added to the loaded ones.
> 
> "Alain" <ele### [at] netscapenet> wrote in message 
> news:46d86fb9@news.povray.org...
>> Zoly nous apporta ses lumieres en ce 2007/08/31 12:49:
>>> Hi,
>>>
>>> I am finishing a set of scenes that make intensive use of radiosity as 
>>> the light comes from outside of the space that I am designing. To allow 
>>> more light to enter through the windows, I am applying radiosity in 2 
>>> steps:
>>>
>>> 1. Render without windows storing the radiosity info (save_file)
>> Windows without the "glass" pane.
>>> 2. Render with windows and reading the radiosity info from the file 
>>> (load_file).
>>>
>>> I have some doubts about how storing radiosity info works. I would 
>>> appreciate some lighting (in my mind):
>>>
>>> 1. What radiosity parameters have influence on the info that is stored in 
>>> the file (all of them?)
>> All of them.
>>> 2. What radiosity parameters have influence on the image if I am using 
>>> load_file?
>> Any can have an effect, when there is a need to gather some extra samples.
>>> 3. Does the info in the file overwrite the radiosity parameters in the 
>>> SDL code?
>> NO.
>>> 4. Can I make the first step at a lower size than the second step?
>> Absolutely. You can also remove any AA.
>>> 5. Does it matter where I put the parameters load_file and save_file in 
>>> the code (after the actual radiosity parameters, after them, ...?
>> load_file
>> radiosity parameters
>> save_file
>> or
>> radiosity parameters
>> load_file
>> save_file
>>
>> If you save_file before load_file, you may reload the data you just saved.
>>> 6. If I get some artifacts in the first step, may I remove them in the 
>>> second one?
>> The second step will gather additional samples if needed. They may, or 
>> not, remove or reduce those artefacts.
>> Use high quality parameters in the first pass, and more relaxed ones on 
>> the second pass.
>> In the first pass, remove any flat transparent, colorless, objects. 
>> Reflective objects should have the texture replaced by a plain white one 
>> IF they are far from bright surfaces: high ambient ones.
>> Keep shaped with ior or colored transparent objects as they actively 
>> change the radiosity samples. Also, keep non-flat reflective
>>> Thanks for your help,
>>> Zoly
>> If you also use photons, you must keep any object that is a target for the 
>> photons.
>>
>> -- 
>> Alain
>> -------------------------------------------------
>> You know you've been raytracing too long when you have ever brought your 
>> computer to its knees by mistakenly launching 64 simultaneous frames to be 
>> traced, while trying to maximizing the benefits of parallelizing them.
>> Carsten Whimster 
> 
> 


-- 
Alain
-------------------------------------------------
WARNING: The consumption of alcohol may make you think you can logically 
converse with members of the opposite sex without spitting.

Post a reply to this message