2D Modeling & Rendering
Youyou Wang, Ozgur Gonen and Ergun Akleman
Gradient Domain Rendering
Abstract
We present an approach to allow artists to create 3D-looking stylized depictions with complete visual control. The core of our approach is to compute rendering effects directly in gradient domain. This approach allows us to use 2D vector fields that may not necessarily correspond to any 3D shape, but that can still help to compute a 3D appearance. To use our approach, an artist provides two images and a corresponding 2D vector field. Final images are created by interpolating between the two images using shading information derived from the 2D vector field.
We show that 2D vector fields is sufficient to obtain qualitatively convincing visual effects. Rendering methods we have developed include diffuse shading, ambient occlusion, soft \& hard shadows; and environment reflection using rectangular images. Although these methods do not directly correspond to underlying physical phenomena, they can provide results that are visually similar to 3D realistic rendering. One of the major advantages of our approach is the ability to treat images as ``mock-3D'' shapes that can be visually acceptable even when images do not correspond to real 3D shapes.
Another advantage of our approach is that the artists can interactively control the illumination and rendering processes to intuitively obtain desired visual results. Since 2D vector fields, themselves, are images, so they can be represented using any convenient 2D raster or vector image format like normal maps. The fact that the 2D vector field is itself an image makes it a very painter-friendly representation, subject to creation and manipulation by both algorithmic and direct approaches.
- To download Gradient Domain Rendering Paper click here.
- To download the video click here.
- To download Warhol's Campbell Soup video click here.
- To download Picasso's Self-Portrait video click here.
Gradient Domain Compositing
Abstract
One of the main issues in 2D digital painting systems is that reflections and refractions are not embedded into the compositing operations. In other words, the current 2D digital painting systems does not provide simple mechanism to obtain reflection and refraction effects. In current practice, deformations caused by reflection and refraction must be (1) computed in a separate 3D rendering software, (2) imported to the 2D digital painting system and (3) combined using masks.
In this paper, we present a compositing approach to obtain stylized reflections and refractions with a simple control. Our approach does not require any mask or separate 3D rendering. Moreover, only one additional image is sufficient to obtain a composited image with convincing qualitatively reflection and refraction effects. We have also developed linearized methods that are easy to compute. Although these methods do not directly correspond to underlying physical phenomena of reflection and refraction, they can provide results that are visually similar to 3D realistic rendering.
The major advantages of our approach is the ability to treat images as ``mock-3D'' shapes that can be inserted in any digital paint system without any significant structural change. The core of our approach is the shape map that encodes 2D gradient and thickness information for all visible points of an image of a shape. This information does not have to be complete or consistent for obtaining interesting composites. The shape maps can be directly sketched or painted by artists, since they can reflect the artist's intention.
- To download Gradient Domain Compositing Paper click here.
- To download the compositing video click here.
Gradient Domain Modeling
Ozgur's Thesis
In Preparation
D/R Shading Model
Fermi's Thesis
vimeo.com/222249570
https://vimeo.com/215480826