STALP: Style Transfer with Auxiliary Limited Pairing

Abstract

source style source frame Jamriška et al. target frame our approach Texler et al. Figure 1: An example of style transfer with auxiliary pairing-an artist prepares a stylized version (source style) of a selected frame from input video (source frame). Then a network is trained to transfer artist's style to remaining video frames (target frame). During the training phase a subset of input video frames as well as the source frame and its stylized counterpart are taken into account. Once the network is trained, the entire sequence can be stylized in real-time (our approach). In contrast to current state-of-the-art in example-based video stylzation (Jamriška et al. [JST * 19] and Texler et al. [TFK * 20]) our approach better preserves important visual characteristics of the style exemplar even though the scene structure changed considerably (head rotation). Input video frames and source style c Zuzana Studená, used with permission.

Efficient Interpolation of Rough Line Drawings

“…In addition, these methods were not trained on rough drawings, and would thus require a large training database that must encompass different drawing styles. In contrast, example‐based stylization approaches [TFK*20, FKL*21] can transfer arbitrary styles to a video sequence with only a few stylized exemplars. Nevertheless, as shown in our experiments (Section 5.1), when applied to rough drawings, their results look more like “advected” textures than hand‐drawn animations.…”

“…Instead, the goal of the interpolation process is to generate a set of strokes for each intermediate frame that (1) appropriately conveys the interpolated contour motion, (2) interpolates the stroke distributions of the key drawings, and (3) introduces a minimum amount of temporal artifacts. One possible solution for such a problem would be to rely on example‐based raster synthesis (e.g., [JST*19,TFK*20,FKL*21]). However, as discussed in Section 5.1 and shown in Figure 11, ignoring the vectorial nature of strokes does not preserve stroke consistency, which leads to severe visual artifacts in the resulting animations.…”

A Stylistic Study of the Hand-Painted Winter Panorama Maps of Pierre Novat

“…Jenny and Jenny (2013) used an early style transfer method, example-based texture synthesis, to apply Berann's style to 3D rendered terrains. This area of computer graphics has made significant progress since that time and continues to be a promising avenue of research-see the stylized panorama in Figures 11 and 12 of Futschik et al (2021). Still, as these methods are automatic, they lack artistic control, a crucial feature for creating expressive renderings.…”

Style Transfer of Chinese Art Works Based on Dual Channel Deep Learning Model

“…Reference [3] proposes an example based image styling method based on the consistency of the target image during training.…”

“…Reference [ 3 ] proposes an example based image styling method based on the consistency of the target image during training. This method keeps the statistical data of neural response compatible with the data extracted from programmed sources and is widely used in video stylization, style conversion to panorama, face, and 3D models.…”