In fact I'm not 'separating' the layers at render time. I do create several versions of the same asset, but with different look for each.

For the 'Albedo' pass, I've created a version of the asset with all the materials setted to constant ( full white diffuse, self illumination 100% ) and with each Diffuse material channel setted to the color textures. This way you get the full color information of the asset without any lighting.

For the 'Ambient' and 'Lighting' passes, I've created another version of the asset with materials setted to their respective surface properties ( diffuse/specular color, no self-illumination of course, specular/glossiness ). The ambient color is irrelevant as no ambient color is used ( even for the 'Ambient' pass wich uses IBL in my case ). Materials have no diffuse textures, only 'Bump'/'Specular'.

For the 'Depth' pass, I've created another version of the asset with all material setted to constant black ( diffuse black, self-illumination 100% ) and no textures. The depth rendering from black ( near camera ) to white ( far camera ) is then created using white Fog.

For the 'Light' pass, I've created another version of the asset with anything other than lights setted as for the depth pass, constant black, and the lights recieve their diffuse and glow texture and strength, if any, and self-illumination at 100%.

For the 'Ambient Occlusion' pass, I've created another version of the asset, similar to the 'Albedo' but rather than using the color textures the baked ambient occlusion textures are used. This is also here where I have to use the asset with a different set of UVs ( unique UVs ), and this is the only reason why a second set of UVs has to be created ( in XSI, Blender, Max, whatever.. ).

So in all I have 6 versions of the asset that I switch in my project to render each pass. The project itself contains the fog settings ( for the depth pass ), the IBL settings ( for the ambient pass ) and the lights settings ( for the lighting pass ). I can save those setting as an 'Atmosphere', ableit if I use more than 4 shadowed lights I have to save several 'Atmosphere'/'Project' ( especialy if they are animated ). I can use several IBLs too, for the ambient.

So yes, it's a bit complex, albeit once you get used to it it's not that complex, but it's certainly a lot more work than straight rendering as you have to prepare your sets, props, characters, etc.. to create thoses versions prior to render. In the case of characters, I wont create several versions, just the materials presets. I have yet to solve the ambient occlusion problem on character, for wich, SSAO would be welcome ;)

If anything is unclear, don't hesitate to ask for more details.

Cheers,
Guy.

--
guy rabiller | GOETZIWOOD STUDIOS
"N.O.E." (Nations Of Earth) Sci-Fi TV Show, Showrunner.

Hey Cris,

Yes I understand it's a bit complex for the average hobbyist, but my concern here is how to overcome some iClone limitations ( regarding rendering here ) by using known production technics, to see if, ultimately, iClone can be used for what I intend to do.

If I hit a wall, a hard limitation, and then there is no workaround or upgrade from Reallusion, then I'll find another software/solution, simple as that.

I want to see if, technicaly - and I insist on the technical side here, I'm not speaking about artistical results - a certain level of quality can be achieved for the result to be considered by the main stream audience ( tv, movie, web-serie, etc.. ) as 'acceptable'.

The Machinima community is very tolerant regarding the technical quality of the result ( rendering, animation, .. ), the main stream audience is not, believe me.

So, if this goal can be achieved, then a all new world of ( indy ) entertainment can be opened, with commercial results ( albeit the goal here is not to create machinima to make money, but to make money to be able to create machinima movies, series, etc.. on a higher level ).

I believe there are some hobbyists here who ultimately share the idea of a professional development of their work, and dream to, or want to become professional directors/writers/animators/etc.. And machinima is certainly a good start to expose your talent as such.

On my side it's a bit different as I'm already a professional, what I'm trying to do is to find the right platform/pipeline/workflow to be able to propose, to the professional community, an alternative way to create 3d animation series, movies, etc.. I've also worked in the movie industry on set, and my ultimate goal is to bring back the on set 'magic' into the 3d animation world with a certain approach of realtime animation, wich is a bit what iClone propose in its 'Director' mode, but need to be opened with a SDK or scripting ( like Daz did ) in order to include procedural animation rather than being purely based on mocap or predefined animations.

So far, the only 'hard limit' or 'wall' or 'showstopper' if you want I've found with iClone are :

- the impossibility to keyframe smooth animation ( due to the lack of smooth key interpolation, only linear and the ease in/out modes are available ). I sincerely hope they will fix this because this is totaly unacceptable for any animation software.

- the need of a SDK/Scripting to deal with procedural animation ( crossing fingers.. ).

Regarding the rendering side, I'm pretty confident it's possible to get the needed quality results and that's what I'm experimenting with those tests.

On a side note, regarding the anti-aliasing problems I get in the animation : there are also some flickering, not due to aliasing problems but to modeling problems( sorry StoneMason.. ). In the UrbanFuture3 set, there are a lot of overlapping polygons that create flicking artifacts when the camera move, so I have to fix those polygons. For the antialiasing, there is still another possibility wich is an old trick too : rendering 4x time the size of the image then scale down after rendering. I did not tested this yet, but I will. Obviously, a native 'enhanced' anti-aliasing mode would be preferable ( StudioGPU is doing this without any problem ), but this is not a showstopper per say though.

So far on the rendering side, I'm quite happy with the possibilities, and the hints you gave about SSAO and layers show promises. The next step for me is to test this multi-passes rendering approach with characters.

Cheers,
Guy.

--
guy rabiller | GOETZIWOOD STUDIOS
"N.O.E." (Nations Of Earth) Sci-Fi TV Show, Showrunner.

No at all blue, all I'm saying is that what he did to get that image\animation is beyond anyone that just started using 3d or doing compositing for that matter, making all those layers work just right is not easy or intuitive.
Of course with proper training anyone can do this. But the key word here is knowhow. Not only of 3d but also in the art of compositing.

Cheers,

Stuckon3d



Product Specialist

My iClone 7 gallery

My IMDB

Hmm it does LOOK complex... however I think if I knew how to seperate the layers it'd be a bit easier... it'd also be easier if I knew how to set up the UVs in XSI... however I think that's only for something you're importing from an external source. If I'm making the scene within iClone I don't think I should hav a real issue with UVs since most of it is a complex maze of 3D blocks anyway.

But I can't quite understand how you get all those different layers out of iClone... is that possible to do already? (multipassing I mean) hmm...

*caugh*tutorial*caugh*

and what's this I hear about 4.2?

----------------------------------------------------
***BEST OF THE CORE**
:A music reel demonstrating the different types of music I produce for multimedia. Hope everyone enjoys.


Player not working? Use this link.

My other audio work.

Facebook page:

Creation begins at the core.

you saying we are thick ?:D



I Make My Mates Do The CanCan For The Amusement Of Killer Robots:D

" Also, when 4.2 comes out, layering will become much easier. :w00t: "

Is that a hint? Is that a hint?? LOL :w00t:

Cathy:hehe:

Just click on the graphic to go there.

Hi Cathy,
if you are going to render in layers you are better off saving them as image sequences with alphas, it is much cleaner output than trying to do green screen and then trying to key out the green. ;)
Also, when 4.2 comes out, layering will become much easier. :w00t:

Cheers,

Stuckon3d



Product Specialist

My iClone 7 gallery

My IMDB

Hi Guy-

I do understand what you are doing now and realized I could probably achieve something similar to break the 4 light limit by shooting multiple passes with green screen in iClone and compositing the shots in After Effects. BRILLIANT! I don't know why this never occurred to me! This opens-up whole new worlds of ideas.

Thanks so much!
Cathy :hehe:

Just click on the graphic to go there.

Excellent job Guy, you do have some fantastic skills man, but you do realize that 95 % of the people here will not be able to do that, right? Not because they are not able but because in order to micromanage an image like that, requires a lot of experience in both 3d and 2d. There are way too many complex techniques here, from creating the compositing layers , to the remapping of uvs, to the baking, etc, etc, etc...
It is great that you are trying to use iclone as your rendering platform for making the layers, and the 2d/3d pros will see this as an advantage in some situations. I also believe you are correct about the antialiasing issue, at render time they should use a better algorithms to avoid the crawling semihorizontal artifacts. This is great experimentation. Thanks for sharing your findings.

sincerely,

Stuckon3d



Product Specialist

My iClone 7 gallery

My IMDB

Hello,





Yes indeed, sorry, I should have been more explicit on the procedure.

I've rendered several times the same scene/animation, but each time I change the 'appearence' of what's on screen to get specific information per render ( pass ). Then the passes are loaded in a compositing software ( this can be done in Blender, for instance ) and assembled in a specific way to obtain the final image.

Here is a little animation showing the involved passes ( I did not show the transparent surfaces though, that has to be treated separately, but I guess you'll get the idea ):



This is a technic largely used in production ( vfx, commercials, 3d animation movies, etc.. ) wich gives a lot of flexibility. You can for instance, re-render only the 'Albedo' passe if you are unhappy with your textures, then composite again. Re-rendering this pass is extremely quick as no lighting is involved. You can also add an unlimited number of 'Lighting' passes to overcome the 4 shadowed lights limitation in iClone, etc..

Here I only used a few passes, but I plan to add more for local lighting near some visible light sources. The 'Ambient Lighting' uses IBL, albeit IBL can also be used for main lighting ( without shadows though ).

The color coded lighting pass deserves some explanation I guess : With 3 lights in your scene that overlap somhow, it is very convenient to color code each light, one red, one green, one blue. This way, in the compositing software, you can separate the three ( by extracting each color component ) and adjust them independently ( level, color, .. ) but you only have to render once. This is an old production trick as well.

If you have more questions, don't hesitate.

Cheers,
Guy.

--
guy rabiller | GOETZIWOOD STUDIOS
"N.O.E." (Nations Of Earth) Sci-Fi TV Show, Showrunner.