Is pyro shader res-dependent?

[~nature~]

Hi,guys

I wonder whether the pyro shader is res-dependent.The problem is the rendering the result is varying between my low-res test(320*240) and high-res final(1024*720).with the final one,I just regenerate the scattering PCs and leave all the other parameters the same as the low-res one.

The result is below

Edited July 16, 2010 by ~nature~

[Macha]

It appears to be. I made some little tests and in my case the fire intensity parameter is related to the luminosity. I made some assumptions about it and tried a remedy that appears to work in my case:

Find the difference in resolution between the two setups and then take the cube-root of that difference and use it as a factor to scale the parameter in question (fire intensity in my test).

If the problem is scattering try if the Max search radius responds to this method and/or play with the max contributor settings.

Does that help?

Edited July 17, 2010 by Macha

[~nature~]

It appears to be. I made some little tests and in my case the fire intensity parameter is related to the luminosity. I made some assumptions about it and tried a remedy that appears to work in my case:

Find the difference in resolution between the two setups and then take the cube-root of that difference and use it as a factor to scale the parameter in question (fire intensity in my test).

If the problem is scattering try if the Max search radius responds to this method and/or play with the max contributor settings.

Does that help?

Hi,Macha,it does help thanks very much. I changed the scattering parameters and got the similar result.

I'm still curious to know why the pyro-shader is res-dependent under the hood,maybe Mario Marengo could explain it

[Macha]

Hi,Macha,it does help thanks very much. I changed the scattering parameters and got the similar result.

I'm still curious to know why the pyro-shader is res-dependent under the hood,maybe Mario Marengo could explain it

My uninformed guess was that the result of shading depends to some degree on the density and amount of samples. Since those increase by the cube as you increase resolution I thought that perhaps you could counteract the effect by calculating a factor as I described above. But I also think the problem is more complex than this and rather than badly fix it with an obscure formula hidden in the innards they left it as it is.