Not there yet? Post your work in progress here to receive feedback from the users.
User avatar
By eric nixon
#391182
Testing the new sss performance in mw3.2

I grabbed a bunch of photo-scanned food models from this site:http://www.blankrepository.com/

The quality there is pretty good, esp the maps, but the models are raw highpoly scans, I just rendered them as is, I would retopo them if I knew how to.



Image This Avocado has a second geometry just underneath with a pure black material, which makes the sss more solid looking, like a shell/skin.

Image Image

Image Image

Image Image

I think the mango still needs some work..

Image Image Image

Image Image Image

Image Image


Really pleased with the new sss, it is much more tolerant of weightmapping and high contrast mapped trans, and its faster ofcourse. (but these renders can still be pretty slow).

I'm not that happy with the dry bread materials, its really hard to keep the overall contrast, when mixing in a broad soft forward scattering bsdf. I hope a skin shader will help (one day) for dry breads.

There is some light-pollution in the maps so its a bit of hit and miss as to which models can be rendered well, the mxm's are fairly simple; a baselayer with 1 diffuse bsdf and 1 sss bsdf (a second broad sss-bsdf needed for the breads + the mushroom render). An additive spec layer is on top. Eyeballing the results is painful and requires patience < warning.

I'll post some more tests soon and some other renders that I cant find right now.
User avatar
By T0M0
#391183
Very nice tests.

Could you please do one of the scenes with noSSS material for comparison ?
Keeping spec. additive layer and diffuse bsdf with r0,r90 textured, Rgh at 97.
I would like to see the difference, how much realism is possible to achieve with good looking subsurface scattering. Thanks.
User avatar
By eric nixon
#391192
Yes, I think what your getting at, is that the way things look highly depends on their form esp transparent things.

rough 97, with original asset texture map;

Image Image

I also thought it would be interesting to break down the material I used;
Image

Another point worth mentioning is that I am only using bump-mapping for the specular layer. The scanned geo cant hope to represent the true shape (of a baked muffin for example), so I dont want to emphasize the actual polygons with a bump/normal map, I want to hide them, I want the surface to be less explicit. A real muffins surface area with all its nooks and crannies is kind of infinite, its impossible to represent that with a bump map, vector disp might work in theory.
By numerobis
#391194
Very cool! Thanks for sharing your results. I saw them announced at ronenbekerman.com. Interesting to see them rendered in maxwell.
User avatar
By eric nixon
#391195
Thanks, regarding that blog, I hope BBB sees these renders by chance, he likes the challenge of food renders too, he is also pretty handy with maxwell..

Obviously food rendering is very subjective and technically challenging with any engine. Vray with the skin shader can cover more scenarios, but mw3.2 is in a different league visually. Fingers crossed for a skin/depth mapping sss shader.
User avatar
By Hervé
#391210
Very Nice Renders Eric ... 8)


kkk lemme hijack a little :)
Croissant-4-emits-hdri.png
You do not have the required permissions to view the files attached to this post.
User avatar
By eric nixon
#391211
Ah, thats the fixed croissant scan. Its looking really good (prob as good as it can get, considering the fudge factor, but I'm still tempted to try and to tweak it, can you send me a copy?) I havent found a good material/technique for 'air bubble' pastry, I tried to get a fried donut look for hours/maybe days :oops: , ended up modelling the bubbles in the batter and the render was rlly slow...

Some image fragments which got saved, was using nested priorities for icing and sprinkles, all is modelled, not scan;

Image Image

Second one has better icing but donut has morphed into a baked potato :oops:

I also did a sugary version with modelled sugar grains and that technique was quite useful. not too slow.

Image

I wish we had a 'crystalline' shader, there are many materials which sparkle due to refracted light, even the powdery icing sugar on cakes has a distinct look which is impossible to render with a bsdf. Another pipe dream I have is for 3D-procedural textures to map the volume of sss like swirling ice cream etc.

I was hoping you would find this thread. What do you think about doing the retopo of any of these assets, Is it hard work ( I'm still a retopo virgin, I do have 3d coat but haven't used it yet :? )
User avatar
By Hervé
#391212
hey Eric this is a nice try.. 8) I found somehow a way to get fluffy pastries like the croissant. It renders fast and clean, and is good with emitters, sky or hdri.. but you have to think outside of the Maxwell logic.. (aka hervé's logic)
I can send you this double deck croissant with default bsdf's. I am curious how you will get the same fluffy effect.

well Yes a 3D procedural for inside sss.. yum yum. :D
User avatar
By eric nixon
#391215
Double deck? So you have 2 layers of geo? In one mesh? how that will render quickly, hmmm curious.

Can I have look with the material you used.
By paulo
#391218
j'ai pris 10 livres juste a regarder ce beau croissant :)

That's a awesome render

Hervé wrote:Very Nice Renders Eric ... 8)


kkk lemme hijack a little :)
Croissant-4-emits-hdri.png
User avatar
By Hervé
#391222
eric nixon wrote:Double deck? So you have 2 layers of geo? In one mesh? how that will render quickly, hmmm curious.

Can I have look with the material you used.
hehe I was just kidding you, only one geo layer... :mrgreen:
Texture/finish lost in render.

Nothing?

I answered this question myself in Rhino 6 documen[…]

New developments for Maxwell 5.1

- First one the random color don't work with hdr[…]

Material preview

Yes, I have set Maxwell for Rhino as Current Rende[…]