Painterly rendered portraits from photographs using a knowledge-based approach

Abstract

Portrait artists using oils, acrylics or pastels use a specific but open human vision methodology to create a painterly portrait of a live sitter. When they must use a photograph as source, artists augment their process, since photographs have: different focusing -everything is in focus or focused in vertical planes; value clumping -the camera darkens the shadows and lightens the bright areas; as well as color and perspective distortion. In general, artistic methodology attempts the following: from the photograph, the painting must 'simplify, compose and leave out what's irrelevant, emphasizing what's important'. While seemingly a qualitative goal, artists use known techniques such as relying on source tone over color to indirect into a semantic color temperature model, use brush and tonal "sharpness" to create a center of interest, lost and found edges to move the viewers gaze through the image towards the center of interest as well as other techniques to filter and emphasize. Our work attempts to create a knowledge domain of the portrait painter process and incorporate this knowledge into a multispace parameterized system that can create an array of NPR painterly rendering output by analyzing the photographic-based input which informs the semantic knowledge rules.Keywords: Non Photorealistic Rendering, Stroke based Rendering, Painterly Rendering, Human Vision INTRODUCTIONNon Photorealistic Rendering (NPR) is a computer graphics technique which creates imagery with a wide variety of expressive styles inspired by painting, drawing, technical illustration, and cartoons. This is in contrast to typical computer graphics which focuses on photorealism. Within 3D and 2D NPR techniques, stroke based or painterly rendering typically uses a 2D source such as a photograph and creates a list of strokes to be rendered on a new canvas. NPR already has applications in video games, movies, architectural and technical illustration, animation and rising fields such as computational photography. NPR also has applications in learning and medicine (e.g. communication systems for autistic children) where filtering out un-needed detail is important, as is true in technical illustration.Since computer systems cannot typically hold a semantic representation of the object to be rendered (e.g. a face for a portrait, or an emotional emphasis), many current NPR techniques rely on computer imaging approaches (e.g. edge detection, image segmentation) that model at the physical level such as blobs, strokes and lines. We propose a novel approach to painterly rendering which relies on parameterizing a knowledge space of how a human painter paints -that is their open methodology to the process. In general, artistic methodology attempts the following: from the photograph or live sitter, the painting must 'simplify, compose and leave out what's irrelevant, emphasizing what's important'. Since human painters have knowledge of the source imagery, we are limiting this approach to portraiture and therefore take advantage of portrait and facial...

Rembrandt's Textural Agency: A Shared Perspective in Visual Art and Science

“…(Figure 4). Second, we rendered these photographs in the style of Rembrandt using a knowledge-based computer painterly rendering system [24] where approximately 50 parameters of brush details, color palette and other painterly attributes were matched as close as possible to the original Rembrandt portraits [25]. Third, we selected four regions in each rendered portrait for selective manipulation with regard to textural detail: one region centered about each eye and one region centered on each side of the chin, where the material of the collar meets the skin of the neck, as illustrated in figures 5 and 6.…”

Exploring the Cognitive Correlates of Artistic Practice Using a Parameterized Non-Photorealistic Toolkit

“…The process of representing a scene, whether from life or a photograph as the source, must by necessity involve simplification, a loss of irrelevant detail in order to better highlight what's important to the artist (DiPaola, 2007). For instance, when using a photograph as source, artists must select the focus that will be rendered in the painting (e.g., will everything be in focus when directly fixated or only selected regions?…”

“…This process allows us to test subjects on how they view the sitter photo compared to the final painting and its cognitive based variants. (DiPaola, 2007).…”

“…Using our parameterized painterly NPR system, we report on four experiments in which we monitor viewer gaze using portraits where these factors are decoupled from relative detail. Portraits were rendered with our parameterized non-photorealistic system to mimic Rembrandt (DiPaola, 2007(DiPaola, , 2009. Participants first viewed and then assigned artistic ratings to each portrait.…”

Learnable Stroke Models for Example-based Portrait Painting

“…DiPaola [9] attempts to map the knowledge domain of the human portrait painter. However, this preliminary work places emphasis on methodology rather than delivering a concrete rendering system.…”