my.cebas.com | Contact
 
| Home | News | Products | Online Store | Community | Service & Support | Downloads | Company | Login

 

finalRender™ 3.5

Hybrid Rendering

finalRender 3.5 offers true hybrid rendering technology. Besides the standard raytracing algorithm, another special "scanline rendering" algorithm has been added to the core engine of finalRender.

The main advantage of such a hybrid approach is improved rendering speed and better image quality. Both rendering technologies have their advantages and disadvantages. Combining the best of both leads to a superior, more robust rendering solution.

What can I expect from such a hybrid thingy - is it faster?

Yes and No.

A noticeable speed increase should be visible in everyday productions when using the scanline rendering option. Especially scenes without heavy secondary (raytracing) effects will perform at their best compared to pure raytracing scenes. Renderings with 10x improvements can easily be achieved using the Scanline Option.  

In some situations however, scanline rendering may not be faster. In the worst case, it might even be slower and use more memory than raytracing, for example with instanced geometry or with scenes showing many overlapping high poly objects.

So what's scanline rendering?

Scanline rendering is the algorithm of choice for all realtime-based or hardware accelerated rendering applications. Several flavors of scanline rendering methods exist. Z-Buffer and A-Buffer rendering methods, for example, are used to render complex 3D scenes.

The 3ds Max core renderer is a highly optimized scanline renderer, though it is aged now and people are moving towards true raytracing engines. The 3ds Max scanline rendering system is still the top choice with tight budgets and a high volume output requirement at the lowest possible cost.

In its simplest form, a scanline renderer would use a Z-Buffer to render pixels. A Z-Buffer is an array that stores the current Z-Depth (distance to the camera) of each pixel. As polygons are sent to the Z-buffer, only those nearest to the camera are rendered (shown in their color). This Z-Buffer rendering is done per scanline, hence the name scanline renderer.

Which method is used by finalRender?

finalRender 3.5 uses a special bucket-based rasterizer algorithm to implement the full advantage of a scanline rendering system in its raytracing environment. It is beyond the scope of this manual to explain the coding methods used by finalRender in detail. The basic methods are nevertheless explained below.

The main rendering process can be divided into two main steps, though in reality, there are many more. The first step is finding the closest pixel to the camera inside a bucket, then coloring (shading) it based on its polygon or face color. Pixels behind the closest one do not need to be processed because they are hidden.  

After this first "rasterizing" step is complete, the second step, "raytracing", is started. By shooting rays into the scene, starting at the position of the first (primary) position (nearest Z-Buffer value) reflections, shadows or GI are calculated. As described, finding the first hit on a surface visible to the camera is done by a scanline rendering approach (Z-Buffer). All further calculations are done through raytracing.

The renderings below show the full power of the hybrid rendering approach.
 

Scanline rendering means speed

Note: 
To illustrate the outstanding thin geometry rendering, all images shown below are zoomed in 2x. 

Shown below is a rendering of a benchmark scene used to stress and test any rendering technology. The image consists of extremely thin objects at large distances with a very low color contrast towards its surroundings.

 



Rendered in 4 seconds with finalRender Scanline Option enabled (zoomed 2x)

 



Alpha channel of the exact same rendering in 4 seconds with finalRender Scanline Option enabled. The images shown above illustrate the clean and smooth anti-aliasing along the fine detail cables on the bridge model. Note the render time as well as the Alpha channel.

 



Rendered in 20 seconds with the raytracer only option (Scanline Option turned off)

 



Rendered in 20 seconds with the raytracer only option (Scanline Option turned off). By using the raytracer only option, the thin lines in this scene are not captured well and show large gaps when using the same quality settings as the scanline rendering option.

To capture such fine detail, low contrast objects, the raytracer needs to shoot a large amount of rays resulting in a much higher rendering time.

 

 

Rendered in 34 seconds with increased anti-aliased settings

 

 

Rendered in 34 seconds with increased anti-aliased settings

 

Conclusion

To obtain the same render quality with raytracing, as shown in the first image, the anti-aliasing render settings must be turned up very high, which increases render time dramatically.

This simple test already shows an increase of more than 8x when compared to the Scanline Option.

 

 

©  2013 cebas Visual Technology Inc.  |  Follow:  Twitter   YouTube   LinkedIn   Facebook   RSS  |  Privacy Policy   |  Terms & Conditions  | Terms of Sale Sitemap