www.si-community.com

milanvasek wrote: i did this a long time ago with version 2 i guess. a lot of displacement (simple grid and subdivided cubes) and hair and dof...

Yeah I remember this image from XSI base, but didn't knew it was yours
Still, real DOF gives something... outstanding, compared to 3d images on the internet.
Thanks for posting !

milanvasek wrote:bison is cool. and those SSS tests as well.

i did this a long time ago with version 2 i guess. a lot of displacement (simple grid and subdivided cubes) and hair and dof...



and some characters i did with 3delight:

http://www.milanvasek.com/newimgs/3d/ellieweb.jpg
http://www.milanvasek.com/wp-content/up ... rd_web.jpg

hopefully i'll put here more recent stuff soon

Hi. Will you be kind to share that scene (hair nad displacement), without maps. I like that hair, but I can't achieve that render quality. Thanks in advance.

doca wrote: Hi. Will you be kind to share that scene (hair nad displacement), without maps. I like that hair, but I can't achieve that render quality. Thanks in advance.

hi. I can try to find the scene, but i'm afraid it's lost somewhere on my old harddrive. anyway it was nothing special. few spot lights with volumic shadow maps, shading rate very low (i guess it was around 0.2, mostly because of all that displacement) and high oversampling. and some transparency on hair i think...
i can make similar scene again and share it (during weekend probably)

milanvasek wrote:

doca wrote: Hi. Will you be kind to share that scene (hair nad displacement), without maps. I like that hair, but I can't achieve that render quality. Thanks in advance.

hi. I can try to find the scene, but i'm afraid it's lost somewhere on my old harddrive. anyway it was nothing special. few spot lights with volumic shadow maps, shading rate very low (i guess it was around 0.2, mostly because of all that displacement) and high oversampling. and some transparency on hair i think...
i can make similar scene again and share it (during weekend probably)

Thank you very much. I asked because I have problem to get good looking hair-I like soft shadows (AO look) on your render, and again, there is nice and recognizable volume of hair. Can you, remember settings for shadow maps?
Thanks again.

and some characters i did with 3delight:

http://www.milanvasek.com/newimgs/3d/ellieweb.jpg
http://www.milanvasek.com/wp-content/up ... rd_web.jpg

I like these! They're great!

doca wrote: Thank you very much. I asked because I have problem to get good looking hair-I like soft shadows (AO look) on your render, and again, there is nice and recognizable volume of hair. Can you, remember settings for shadow maps?
Thanks again.

cant really remember. but i used some kind of light dome (around 10-12 lights) to make that soft look and volumic shadow maps with big softness

Hi all,

this time, a bit more about 'hair translucency' effect, especially noticeable on bright hairs. One simple method to get this effect is using of colored shadows.
That's how my trial looks like:



For 1280x1280, 2-core 3delight took about 8 minutes for render. There is 80k hairs, two spot lights with 1024 deep shadow maps. I've used only standard XSI shaders.
Take look at full res.

So how to get this in 3delight:
1: In global render options, under 'Shadow Maps > Shadow Map Shaders', enable 'Surface'.
2: Apply 3delight's 'Hit Mode' property, to Point Cloud with ICE strands.
3: In 'Hit Mode' property, enable 'Run Material' for Shadow Rays.
4: Enable 'Detailed Shadow Map' for your lights.

Step 3 is needed, because XSI Hair Shader doesn't allow colored shadows, even it can be used as a shadow shader. That's how it looks like when XSI Hair Shader is used for shadows too (on left side ).



So, we will supply another shader for shadows. Some recipe is a bit darker and more saturated color for Constant surface, some gray variance for transparency. Basic render tree looks like this one:



And that's all.

Tips:

Somewhere in installation folder, probably \Softimage\.....\Data\DSPresets\Materials\Hair, there is a lot of color presets for XSI Hair Shader. I've found them *very* usable.
Here I've used a 'Hair Golden Blonde'.

When rendering ICE strand segments, 3delight automatically does the curve interpolation. Generally, you should be able to use a much lower count of strand segments.



About mip-mapping for Deep Shadow Maps, it seems 3delight adjusts this automatically. Corresponding 'Sub-Samples' parameter seems it has no effect, even it's mentioned in the docs. On the other side, by watching render log, I've noticed reports about various mip-map levels. In short, I believe in 'common' rendering strategy - start with low-res DSM, but don't hesitate to increase it if needed.

Cheers

Wow, that's looking good! Thanks for all the info!

I used the same setup, only with a color savestate instead of the constant - no difference really.
It's a really nice setup, especially with a light rig generated from a HDR. I would say 30 lights or more make sense to get the most realistic lighting. Also had to increase shadow map resolution to 2048 to get rid of some shadow artefacts.
No problem to render this with 3delight's speeds even on my oldish laptop.

Just to add some confusion for those unaccustomed with them terms ...
Shadow map = 2 color-only shadow map, no transparency
Deep shadow map = Pixar's original "with tranparency" shadow map format (pdf)
Detail shadow map = mental image's version of deep shadow maps
Volume shadow map = Softimage's GUI name for MR detail shadow map

Moderator edit: I've reintegrated the slightly off topic discussion about the license problems ActionArt experienced (initiated by myself )
into the other 3Delight thread, where the posts seem to belong...
;)

rray wrote:Also had to increase shadow map resolution to 2048 to get rid of some shadow artefacts.

Yeah, it seems that DSM stuff needs some time to figure out. I'm still in process.... Some basic nodes behaves different, for example, color_math_exponent has a clamped input-output... On the other side, I've been able to get a 'distance fog' effect, using vector_state nodes.

Anyway, here's with-without motion blur variance. I got motion blur from 3delight, there s no ICE velocity attribute. About 7 - 5 minutes on 1920x1080.



So... thank You all
I'm planning my next update for next weekend...

DOF and rendertime blobs. 3delight does the blobs in render time, using it's own algorithm. This making them much faster to pre-compute than in 'standard' way, by Softimage. Not the all 3delight blobby operators are supported, it's just SI 'add'. At least, 3delight 'subtract' would be nice to have in SI, but here it is what it is.
Here I've intentionally tried a heavy render, using Deep shadow maps, ray-traced displacement, DOF, ray-traced reflection and glossy ray-traced refraction, point based AO and color bleeding.
Depending of what is used, render time on 2 cores is from 10 min to 2 hours for 1600 x 800. Here is a 'full' configuration. I've adjusted defaults a bit - smaller tile size for better memory handling, smaller ray-trace bias for better intersections.

Take look at 1600x800.

And yet another render-time blobs. Here's about 500K of them, together with point-based AO.
Render time is about 5-7 min.

If anyone is interested, here's a small ICE compound used to generate these 'blob-strokes'. I had something like this before, but I really wasn't so excited by the speed of generation. Now with 3delight, it's definitely feasible, updates are performed immediately. The compound creates particles on curve list, evenly distributed, count is relative to particular sub-curve length (still, you'll need to take care of curve re-parametrization). It's also able to perform 'write on' on one-after-another subcurve (by sub-curve, or normalized by sub-curve length). In short, you can 'draw' 3d strokes with it, by animating the slider value.
There are a few additional parameters, scaling along curve, scaling by angle to some direction ( it's called 'side scale'). I'd believe it's not hard to figure it out, just by playing with sliders.
It's designed for use in non- simulated ICE tree.

Cheers

thanks Mathaeus! this is really nice
i've opened the compound and got depressed..i'll never learn ICE lol..thats too complicated:/ there are like 90 nodes..

great job, much appreciated:)

Max, if you need some variation of it, no big deal to do that. Btw it is a bit more complex, because it needs to find the length of each sub-curve, and few another 'unusual' values.
Here's an update, down-loadable on the same link a few posts above. As it's probably noticeable, this one has a better distribution, it will keep particles together, even on some wild parametrization (actually it will scale them to 'fit'). Probably you can use it for your jewelry...
('pre samples' parameter is for measuring sub-curve length, it needs to be higher on some very complex curve).