NeSprite Series

Traditional Dot Conversion

Before

After

Simple dot-style conversion can be achieved by combining image downscaling using the nearest-neighbor algorithm with color quantization using k-means clustering.
Even more advanced dot conversion methods traditionally only added features like thickening edges for better outline retention when shrinking images, and basic adjustments to contrast or saturation.

While this was enough for decent quality image-based dot conversion, the same couldn't be said for video.

Early AI-generated videos were often easily recognized as AI-generated, primarily because they processed each frame independently.
As a result, faces or backgrounds would fluctuate between frames.
Modern AI now treats video as a coherent sequence, improving visual consistency dramatically.

In other words, natural-looking video requires consistent per-frame processing.
Traditional dot conversion algorithms, when applied to video, often result in flickering.
This is due to the randomness inherent in k-means clustering, which causes color variations frame-to-frame.

To solve this problem, the NeSprite algorithm was developed.

NeSprite Algorithm

Before

After

1. Outline enhancement via white pixel contraction
   This is a technique long used in dot conversion: if a white pixel is surrounded by dark ones, it gets converted to black.
   This thickens outlines and enhances edge clarity.
2. Contrast / Saturation / Brightness / Hue adjustments
   Essential for tuning the visual tone of pixel art.
   Increasing contrast improves clarity, and boosting saturation emphasizes the vivid style typical of pixel graphics.
3. Downscaling using nearest-neighbor
   Nearest-neighbor uses the value of the closest pixel, preserving the original color and shape.
   Unlike other interpolation methods that blend colors when shrinking, nearest-neighbor keeps hard edges intact — perfect for pixel art.
   It’s also fast to compute.
4. Dithering
   By applying a pre-defined dithering matrix, pixel positions are patterned to reduce color banding and provide a more visually appealing reduction.
5. Mapping each pixel to its nearest palette color
   Using a fixed palette limits the number of colors and keeps the result stylistically consistent. It also prevents unwanted color fluctuations between frames.
   Since retro game consoles also relied on palette systems, this method fits well with pixel art.

Creator-Oriented & High-Level

The greatest strength of NeSprite lies in its palette-based transformation.
For pixel artists and game developers, this concept should feel very familiar.
Have you ever wished for pixel art in exactly the colors you want?

NeSprite provides that — a system optimized for video conversion, and features that are especially appreciated by creators and engineers alike.