Page 3 of 4
Re: Spectral rendering
Posted: Wed Jan 22, 2020 3:05 pm
by kintuX
lacilaci wrote: Wed Jan 22, 2020 2:54 pm
Yes but how do you compare octane vs spectral rendering octane? Is there a switch to turn of spectral rendering in octane?
This new build of cycles allows for pretty much direct head to head comparison to see if it's worth it in some produciton (dispersion effects aside)
One way of comparative study is you give both samples to X number of people asking them which one they favor more, then do the stats. The other is unbiased, scientific way of using specialized image analysis tools and color science to describe the methods, techniques, do the work, get results with which you are then able to make a conclusion.
Re: Spectral rendering
Posted: Wed Jan 22, 2020 3:30 pm
by Sharlybg
One way of comparative study is you give both samples to X number of people asking them which one they favor more, then do the stats. The other is unbiased, scientific way of using specialized image analysis tools and color science to describe the methods, techniques, do the work, get results with which you are then able to make a conclusion.
Top
I think laci mean it could not be a way to switch off spectrale on octane so you can do the test with only one render engine.
Spectrale
ON vs Spectral
Off
Re: Spectral rendering
Posted: Wed Jan 22, 2020 3:50 pm
by lacilaci
Sharlybg wrote: Wed Jan 22, 2020 3:30 pm
One way of comparative study is you give both samples to X number of people asking them which one they favor more, then do the stats. The other is unbiased, scientific way of using specialized image analysis tools and color science to describe the methods, techniques, do the work, get results with which you are then able to make a conclusion.
Top
I think laci mean it could not be a way to switch off spectrale on octane so you can do the test with only one render engine.
Spectrale
ON vs Spectral
Off
↑ this!
Re: Spectral rendering
Posted: Wed Jan 22, 2020 4:18 pm
by Sharlybg
Anyway i have octane ready to use on my machine now. just playing arround to understand it.
Re: Spectral rendering
Posted: Fri Jan 24, 2020 6:04 am
by lighting_freak
Hi,
From my understanding the real benefit of spectral rendering comes with spectral input data.
The demand for spectral defined lights is as necessary as spectral defined surfaces and volumes. If anything regarding spectral rendering is planned for luxcore something like an additional database for those input data would be appreciated.
BR
Re: Spectral rendering
Posted: Fri Jan 24, 2020 9:57 am
by Sharlybg
spectral input data.
This is a nice input guy. what kind of spectral data are available or needed :
__ spectral material ?
__ Spectral light input ?
__ Spectral Volume ?
Re: Spectral rendering
Posted: Fri Jan 24, 2020 8:17 pm
by CodeHD
In terms of data, I think light sources are not a big problem. The thermal specturm of an incandescent light bulb is easily modelled, and typical spectra of LEDs or fluorescent lamps are readily available. Laser are also trivial, just make them one resolution element of the spectrum

Manufacturers of filters for theatrical/architectural applications , for example LEE, provide transmission spectra of filters, so those could be modelled accurately.
The bigger problem will come in at the materials, to replicate their appearance under different light sources. This would need to be researched a bit. There is a lot of data on spectra of atoms and molecules, but mostly concerning emission/absorption lines - at much higher resolution than I guess LuxCore would compute - and also a lot in the infrared. Many effects are also related to fluorescence, which would be a bit more complicated to implement. (I think it was discussed before

)
Re: Spectral rendering
Posted: Fri Jan 24, 2020 8:33 pm
by CodeHD
I should add, that does not mean its too difficult to achieve a benefit. I tend to say things too negatively

Re: Spectral rendering
Posted: Fri Jan 24, 2020 9:43 pm
by lighting_freak
Hi,
Just like to agree to the facts mentioned before.
For light sources most spectral information is available.
A database may makes selection easier.
A little harder becomes the light source when it comes to display, hdri and skies. I know that light simulation software tend to use individual RGB spectra (mostly gaussian) to turn the RGB values of the input images into something physics based.
For materials that are used to transmit light in lots of cases the spectral transmission or reflection or absorption is available. A database would be helpful as well.
Other materials may become the problem. Maybe some nice node in blender can generate spectral curves, together with the preview (eg defined as observed in daylight D65) it could be powerful and essay to use.
BR
Re: Spectral rendering
Posted: Fri Jan 24, 2020 9:52 pm
by Sharlybg
The simple way can also be just to do what engine like Maxwell indigo or opensource Mitsuba does if possible.