Abstract: Techniques are provided for modifying coloring of images utilizing machine learning. A trained model is generated utilizing machine learning with training data that includes images of a plurality of different scenes with different illumination characteristics. New original images of a scene may each be downsampled and transformed to a corresponding output image utilizing the trained model. A color transformation from each original image to its corresponding output image may be determined. In an embodiment, the color transformation is determined utilizing a spline fitting approach. The determined color transformations may be applied to each of the original images to generate corrected images. Specifically, the color transformation that is applied to a particular original image is the color transformation determined for the input image that corresponds to the particular original image. The corrected images are utilized to generate a digital model of the scene, and the digital model has accurate model texture.

Abstract: The systems and methods described herein may allow diffusion curve images to be displayed by a variety of applications without requiring specialized graphics hardware to render the images. The system may provide mechanisms to convert a diffusion curve representation to an alternate representation in a scalable, portable format. The conversion may include a tessellation operation, and may produce a mesh of shapes (e.g., triangles) for which diffusible attribute values (e.g. color and blur) are assigned to each node. Tessellating an image may include generating a piecewise linear approximation of the diffusion curves and triangulating the image to generate the mesh. Color values for each channel may be assigned by solving a sparse linear system in the triangle domain. The alternate representation may be displayed by another application using various interpolation techniques.

Abstract: A system and method for adding vector textures to images may employ a unified vector representation for images and textures based on extensions to diffusion curve representations, along with tools to facilitate the draping of texture maps over an image. This vector representation may support user-configurable parameters that allow artists to design and manipulate complex textures at a high-level, and to deform them directly in two-dimensional images. The vector representation may provide material descriptors and visual cues of three-dimensional shape in two-dimensional images. It may facilitate the creation of lighting and shading effects in such images. The system may include tools to manipulate the configurable parameters to create texels, to distribute texels in a texture map, to automatically generate new texture elements, to generate 2.5D surface approximations of two-dimensional images based on diffusible normal attributes, and/or to apply a texture by mapping or projecting it onto such 2.

Abstract: The systems and methods described herein may allow diffusion curve images to be displayed by a variety of applications without requiring specialized graphics hardware to render the images. The system may provide mechanisms to convert a diffusion curve representation to an alternate representation in a scalable, portable format. The conversion may include a tessellation operation, and may produce a mesh of shapes (e.g., triangles) for which diffusible attribute values (e.g. color and blur) are assigned to each node. Tessellating an image may include generating a piecewise linear approximation of the diffusion curves and triangulating the image to generate the mesh. Color values for each channel may be assigned by solving a sparse linear system in the triangle domain. The alternate representation may be displayed by another application using various interpolation techniques.