[BUG] Smoothing Errors - Glossy T. + Epidermis

[Zom-B]

While Testing the Epidermis Dermis SSS stuff for fun the last days,
I found some strange normal smoothing issues using Epidermis
on a Glossy Transparent Material.



I used an 0,75cm radius ring (~an average male finger), scene File + C4D source is attached ( + the HDRI from Paul D.)
**EDIT** it seems per exident I rendered some of the example scenes not in 0.75cm Radius, but 100 times bigger... but finally the normal smoothing is the problem here, so this doesn't matter...

The attached Pictures show this effect very well, I also provided some reference shoots of other materials for comparison and proof that its only an Epidermis+Glossy_transparent Bug.

at the end I did some much finer subdivision to the mesh, to show, that the artifacts are still there, even on very detailed/subdivided meshes

I noticed this back in 0.7t3 once, but skipped testing because of the super slow render Times of this Skin simulation

[Kram1032]

I think, he disabled normalsmoothing for some reason...
If you search in the forums, you'll find, why...
Not sure, but I think, it was in CCJ's BIIIIG test thread, where he also tested Epi-/dermis

[OnoSendai]

Thanks for your tests ZomB.
I can't remember what the exact problem is but I've come across this before.
As far as the skin stuff goes, I'll probably rework it at some point to only use a single layered mesh, as per the original paper of Jensen.

[Zom-B]

As far as the skin stuff goes, I'll probably rework it at some point to only use a single layered mesh, as per the original paper of Jensen.

Wohooo this would be fantastic Ono.... the reason we have seen so few "Skin renderings" isn't mainly the low speed, but this 2 Mesh thing, that is some Pain in the a** for some High Poly models on some 3D apps...

BTW: I tried to do a noise free Skin rendering Test... and failed! I was rendering for over 5 Days, and the result is still not convincing...

Rendered on 2x2GHZ using Indigo 1.0.4.
I used MLT + BiDir on 2 instanced 0.75cm rings with epidermis of 0.25mm.

large_mutation_prob = 0.5
max_change = 0.005
max_num_consec_rejections = 1000
max_depth = 10000
super_sample_factor = 1

I did an Animated GIF showing all the resume stadiums of the Picture.



The bitch in Speed is the Dermis for sure, rendering an Dermis only sphere for example will last for ever, while Epidermis cleans quite fast like every other SSS Mat...

[OnoSendai]

Yup, plus the skin system just sooo slow right now.
That does looks quite like a real 'skin-donut' tho