Page 1 of 1

IBL workflow

Posted: Thu May 20, 2010 3:26 pm
by fuso
Afternoon all,

Looking into IBL after having finally upgraded most of the visualisation software in my office. I also discovered a new source
of texures at CGTextures.com. As most of the skies provided there, are only sky domes (half spheres) I was wondering what
the best way is to convert them into decent HDRI's or MXI's to be used with Maxwell V2.

- is it better to extend the canvas size to full spherical maps (eg. from 6000x1500 to 6000x3000) in order to avoid having
'two sun's' when tiling U: 1.0, V: 2.0? If so, should the colour in the extended area be a neutral medium grey or else?

- does it still optimise the lighting when using a much smaller version of the HDR/MXI for illumination than the ones used in
all other channels like background, reflection and refraction (eg. 600x300 or even smaller?)

- is it still the best result when loading the original jpg's into Maxwell, adjusting the shutter speed and ISO to match your
camera settings in the scene and then saving them out as HDR/MXI? How do you set the intensity as HDR's still come in
a lot darker than they appeared in Maxell?

It's much appreciated if someone could shed some light into this for me and remind me briefly what the best way is to come
to a decent result and at the same time optimising it as much as possible.

Thank you so much in advance...

Re: IBL workflow

Posted: Thu May 20, 2010 4:09 pm
by tom
fuso wrote:- is it better to extend the canvas size to full spherical maps (eg. from 6000x1500 to 6000x3000) in order to avoid having
'two sun's' when tiling U: 1.0, V: 2.0? If so, should the colour in the extended area be a neutral medium grey or else?
Depends. Do you really have maps with 4:1 aspect? (e.g 6000 x 1500) Can you post a small preview? Ideally, the maps should be in 2:1 aspect and require no tiling.
fuso wrote:- does it still optimise the lighting when using a much smaller version of the HDR/MXI for illumination than the ones used in
all other channels like background, reflection and refraction (eg. 600x300 or even smaller?)
This has never been correct for Maxwell. You do not need additional reduced illumination maps as it can't affect the render speed.
fuso wrote:- is it still the best result when loading the original jpg's into Maxwell, adjusting the shutter speed and ISO to match your
camera settings in the scene and then saving them out as HDR/MXI? How do you set the intensity as HDR's still come in
a lot darker than they appeared in Maxwell?
This is still relative as we don't store all the camera information in the images. You need to adjust it with intensity.

Re: IBL workflow

Posted: Thu May 20, 2010 5:12 pm
by fuso
Thanks for your quick reply Tom and good to 'see you' again, too! =)
tom wrote:Depends. Do you really have maps with 4:1 aspect? (e.g 6000 x 1500) Can you post a small preview? Ideally, the maps should be in 2:1 aspect and require no tiling.
This is the website http://www.cgtextures.com/ and here is a sample as well as my modified version:

Image

Image
tom wrote:This has never been correct for Maxwell. You do not need additional reduced illumination maps as it can't affect the render speed.

I thought it might affect the lighting quality but not neccessarily the render speed.

Re: IBL workflow

Posted: Thu May 20, 2010 6:27 pm
by tom
fuso wrote:here is a sample as well as my modified version.
That's correct. The original one is not a spherical map, it's hemispheric and your conversion is correct.
tom wrote:I thought it might affect the lighting quality but not neccessarily the render speed.
This is also true.

Re: IBL workflow

Posted: Thu May 20, 2010 6:37 pm
by fuso
tom wrote:
fuso wrote:I thought it might affect the lighting quality but not neccessarily the render speed.
This is also true.
True as in the quality gets worse when decreasing the HDR image size?

Re: IBL workflow

Posted: Thu May 20, 2010 7:25 pm
by tom
Theoretically. But in practice it may not be significant.

Re: IBL workflow

Posted: Thu May 20, 2010 7:50 pm
by Bubbaloo
Also worth mentioning, if those sky images are 8 bit jpeg's, they will not produce good scene lighting. IBL works optimally with true high dynamic range 32 bit images.

Re: IBL workflow

Posted: Thu May 20, 2010 9:06 pm
by fuso
Bubbaloo wrote:Also worth mentioning, if those sky images are 8 bit jpeg's, they will not produce good scene lighting. IBL works optimally with true high dynamic range 32 bit images.
You're absolutely right, I just realised that as I struggle to get any decent hard shadows out of those converted ones.
They are 8 bit, any workaround to improve it??

Re: IBL workflow

Posted: Thu May 20, 2010 9:15 pm
by Bubbaloo
Nope... you need to invest in some high quality HDRI's. The good ones come with high res back plates for the background channel. I think HyperFocal is a good source.

Re: IBL workflow

Posted: Fri May 21, 2010 11:50 am
by fuso
I know Brian, I just thought it would be a cheap alternative and additional source. We have recently purchased 'Radiant
Skies' and 'Skies V2' from Doschdesign which are brilliant. I can still use them as backplates and combine it with physical
sky. As we're architects and I have to do a lot of photomontages this applies in most cases anyway.

However, thanks for all the replies so far and enjoy a sunny weekend!

Cheers