在像素層面( Pixel level ) 作畫是一種特殊的藝術風格,需要靠畫家手動調整每一個像素,而每個像素是用來保留原始形狀的重要元素。這樣的作畫過程較為繁複,因此我們提出一個系統,能夠自動將任意的線條圖( Line Drawing ) 轉成像素圖( Pixel Drawing )。首先會將原始的線條圖分成一組方形的格子(Grid),每個格子的尺寸為n乘n,視輸入與輸出解析度之比例而定。這些格子代表的將是輸出結果的「複數像素」(mega pixels)。接著依照原始線條所經過的比率計算出每個格子的選擇權重;最後遵從畫家們所歸納的一些像素藝術之規則,如:像素排列需為單調遞增或遞減以及線條需為單一像素寬度。照著這些規則來選擇需要填滿之格子,最終格子的選擇結果即為輸出結果。這些規則使輸出結果較為接近畫家之作品。此結果亦可提供給畫家作為作畫之初稿,畫家可再依個人作畫習慣調整成自己滿意的作品。
Drawing on the pixel level is a specific art style. It needs artist to adjust each pixel manually. Every pixel on the drawing is important to preserve the original shape as much as possible under different resolutions. This process is tedious and complicated, so we propose a novel system which converts arbitrary line drawings to pixel drawings automatically. First, the input line drawing image is partitioned into a set of rectangular grids. Each grid is consisting of n by n pixels, depends on the ratio of input and output resolution. These grids represent “mega pixels” at output resolution. Then, the weight of each grid is computed according to the coverage of original line pixels within the grid. Finally, some rules of pixel art which generalized by artists such as monotonic in generated pixel spans and single pixel line width, are used to choose the grids which need to be filled. These rules further improve the output that is closed to the artist’s work. The artists can then do some minor modifications, based on the final output result, to meet their personal stylization.