Skin test

Post by **OnoSendai** » Sun Feb 11, 2007 4:02 pm

Very nice tests Ryjo.
One thing to watch out for, is that in the paper, when it says C_m = 0.5% (melanin fraction of 0.5%), then the actual fraction as required by Indigo is 0.005 (e.g. half a percent)

Kram1032 · Post by **Kram1032** » Sun Feb 11, 2007 9:35 pm

ryjo wrote:Do you mean the patterns visible on the forehead? Those are more visible the longer the render goes, this has rendered about 20 mins longer. It is a low res model, ~55000 vertices.

Yes, those and also those crossing the cheeks^^
So SSS is TOO accurate atm xD
Can SSS be normalsmoothed, yet?

ryjo · Post by **ryjo** » Mon Feb 12, 2007 12:01 am

OnoSendai wrote:Very nice tests Ryjo.
One thing to watch out for, is that in the paper, when it says C_m = 0.5% (melanin fraction of 0.5%), then the actual fraction as required by Indigo is 0.005 (e.g. half a percent)

Yeah, I did a fair share of mistakes during development of the scene, but I think I got it right at last. At least I hope it's correct now

One thing I noticed, some parts are not as "translucent" as I expected, e.g the ears look a little too solid. But it may be that the mesh is not a perfect model in this case.

Post by **OnoSendai** » Mon Feb 12, 2007 12:05 am

You may also want to do some tests with a specular material as opposed to glossy transparent.
I've added in bump for glossy transparent for next release

ryjo · Post by **ryjo** » Mon Feb 12, 2007 12:07 am

Kram1032 wrote:
ryjo wrote:Do you mean the patterns visible on the forehead? Those are more visible the longer the render goes, this has rendered about 20 mins longer. It is a low res model, ~55000 vertices.
Yes, those and also those crossing the cheeks^^
So SSS is TOO accurate atm xD
Can SSS be normalsmoothed, yet?

The streaks crossing the cheeks are probably temporary and appear in the beginning of the render. They are more visible on light skin tone than on darker. They go away after some time, but then the mesh resolution become visible instead.

I used normal smooting in the scene.

neepneep · Post by **neepneep** » Mon Feb 12, 2007 2:23 am

OnoSendai wrote:I've added in bump for glossy transparent for next release

I see your

.....

.... and I raise it by

ryjo · Post by **ryjo** » Mon Feb 12, 2007 7:53 pm

OnoSendai wrote:You may also want to do some tests with a specular material as opposed to glossy transparent.
I've added in bump for glossy transparent for next release

Sounds great

I did try specular transparent, but it made the model look like a jar of honey had been poured over the head.

Here is the final update. Rendered for ~11 hours. The skin parameters are the same as in the previous render.

melanin_fraction: 0.15
melanin_type_blend: 0.0
hemoglobin_fraction: 0.001

Post by **OnoSendai** » Mon Feb 12, 2007 8:53 pm

Cool...
Glad to see that a SSS render can clear up well

The skin doesn't look especially convincing to me, but that may be the lack of fine bump detail throwing off my perception.

Kram1032 · Post by **Kram1032** » Tue Feb 13, 2007 3:09 am

Whoa!
Cool!
...
Woah!
(Have I said "woah" already?)

Camox · Post by **Camox** » Tue Feb 13, 2007 3:32 am

wow

awesome !

manitwo · Post by **manitwo** » Tue Feb 13, 2007 3:45 am

Why is this so dark? In reallife i think this wouldn't look like that. Could it be that this is a energy loss related problem?

Kram1032 · Post by **Kram1032** » Tue Feb 13, 2007 3:57 am

Maybe, it's 'cause of the epidermis is uniform 25mm
It should be thinner @ three sites:
1) Ears
2) Nose (exactly the part of your pic^^)
3) Eyelids

ryjo · Post by **ryjo** » Tue Feb 13, 2007 5:10 am

manitwo wrote:Why is this so dark? In reallife i think this wouldn't look like that. Could it be that this is a energy loss related problem?

I think that the skin doesn't transmit sufficient light through the nose, so it looks darker than it should. I have used "glossy transparent" material surface, which may be wrong for skin.

ryjo · Post by **ryjo** » Tue Feb 13, 2007 5:16 am

Kram1032 wrote:Maybe, it's 'cause of the epidermis is uniform 25mm
It should be thinner @ three sites:
1) Ears
2) Nose (exactly the part of your pic^^)
3) Eyelids

The paper says it uses 0.25 mm epidermis across all skin, and the examples in that paper look ok.

See section 3.5 and examples at page 10:

http://graphics.ucsd.edu/papers/egsr200 ... 06skin.pdf

ryjo · Post by **ryjo** » Tue Feb 13, 2007 5:19 am

OnoSendai wrote:Cool...
Glad to see that a SSS render can clear up well
The skin doesn't look especially convincing to me, but that may be the lack of fine bump detail throwing off my perception.

Thanks,

I kind of convincing until you look at it too closely, then it looks like a "doll". I think that a bump to the skin would help, but it needs more "translucancy", especially at the ears and nose regions (as manitwo and Kram1032 pointed out).

Skin test

Who is online