32 or 16 bit textures for shades in Corona ?

[Fibonacci]

Hi all,

Somebody could you answer it ?

So, if I use 32 or 16 bit textures for the shades, tha Corona is it couting down to 8 bits ?

Or should I use 16bit textures for get all of benefits ?

I think if I want to get the real and correct LWF in 32 bit workflow, then my textures should be same at least 16 bit...

Thank you and Peace for all!

[Ondra]

I think if I want to get the real and correct LWF in 32 bit workflow, then my textures should be same at least 16 bit...

How are those things connected?

[borisquezadaa]

Is there any noticeable loss in render using diffuse images with less deep color information?.
If i use 16 bit image textures for a scene, do we see any noticeable change in the rendered image if we use 8 bit images for the same (or 32bit)?....
Maybe a silly question, but i wonder the same.
Are there any source of 16 bit textures to try even?.

[Ondra]

texture input is completely independent of the output. I can imagine for some subtle gradients 8bits being too small resolution, but generally, I havent seen anyone having any problems in the whole 3D community

[Fibonacci]

Well...I think if I use the 8bit texture in the 32bit output ( that I'd supposed to LWF )...something "Full Linear Workflow" - FLWF - the new challenge ;)
So, the point is...better if I use at least 16 bit textures for the diffuse. At the gradiens and to the spec, bump, any grayscale map has  the real benefit of 16bit...But in the diffuse ?

I think if I use 16bit diffuse map for shading that's should be give back "much" color and luma information for my output...
So...in my mind is very simple...If I want to get the correct render output in 32bit, than I have (???) to forgot the 8bit textures...? :)

Or this is not matter ?

 

[Fibonacci]

Well Boris,

I think we have to try what's goin' "behind the scene"...! :)

[Ondra]

you are mixing 2 unrelated topics: color depth of your textures and proper use of color spaces (or gamma). Linear workflow covers just the second issue. If using correct LWF( max 2014+ properly set up to use gamma 2.2), you are fine. Only problem you could get from using only 8bit textures is banding. No "incorrect result", "worse image", "less information", "less color", "lost luma", or anything like that

[Juraj]

"Linear" refers to linear function, not color (bit) depth, so your input textures won't matter as they get converted anyway.
16bit can in certain cases where embedded information matters provide better quality (particularly notable example of displacement maps), but it has nothing to do with LFW.

32bit textures do exists btw (outside of HDRi spherical images), mostly for flexibility of their use, I will have access to some as I am betatester of Megascans.

Edit: Ups, Keymaster was much faster....

[Fibonacci]

Thank you guys ! Now it's clear me...a bit.. ;)

[borisquezadaa]

Ok. i  understand the mixed concepts here Linear workflow v/s bit depth.
Mi question was pointed more to the bit depth v/s diffuse effect.
So for displacement (to avoid issues with banding), 16 bits is better.
But for diffuse... is there any noticeable change?... is there an advantage of using 16 or 32 bits images for diffuse maps over common 8 bits? (Specifically for complex diffusse surfaces, not just plain color ones).

Sorry if my question is not specific.
So let me put an example of what is on my mind...

Megascans have a great selection of high bit depth textures scanned reallife like so as example is there any difference in final image whatsoever if i replace  lets say a interior scene made ENTIRELY of only 32 bit textures (diff, spec,bump, etc) from megascans but deliveratedly replacing just diffuse with 8 bit depth textures?.

[Juraj]

No.

The purpose is flexibility to tweak and adjust that texture (in case of Megascans it's due to their intention to be intermediate product).

[borisquezadaa]

Thanks Juraj!. (Near post 1000!.. have a beer!).