Clip Studio Paint (CELSYS)

Clip Studio Paint offers a Remove Tones feature that uses AI to detect and remove or convert screentone dot patterns from scanned or flattened manga pages. Artists can use it to strip away printed halftone dots or convert them to grayscale shading.

The Remove Tones feature uses machine learning technology and sends image data to a remote AI server for processing, like the Colorize and Pose Scanner features. It operates on a selected raster layer and can either delete screentone patterns entirely or convert them to grayscale, which is useful when importing traditionally drawn manga or working with flattened files. The feature was introduced alongside Colorize in Clip Studio Paint Ver. 1.8.4 in 2018 and is restricted to the EX edition.

Clip Studio Paint offers a Smart Smoothing feature that uses AI deep learning to reduce pixelation and noise when an image is resized or its resolution is changed. Artists use it to clean up low-resolution artwork so it can be printed or to smooth lines that have become blurry after scaling.

Smart Smoothing uses a deep learning model developed by CELSYS in partnership with the ailia SDK framework from ax Inc. and Axcell Corporation, and runs locally on the user's device without requiring an internet connection. It takes a raster image layer as input and produces a smoothed version with reduced jagged edges (jaggies) and pixel noise. The feature was introduced in Clip Studio Paint Ver. 1.9.1 in July 2019 and is available across DEBUT, PRO, and EX editions on Windows, macOS, and iPad.

Clip Studio Paint offers an AI-powered Colorize feature that automatically fills color into a user's black-and-white line art drawings. The artist provides the line drawing and optional color hints, and the AI generates a colored layer on top.

The Colorize feature uses machine learning technology to automatically colorize line art merged into a single layer, sending image data to a remote AI server for processing. Users can run a basic 'Colorize all' pass or supply a hint layer with rough color blobs to guide the AI's output. According to a 2024 official statement, the feature's model was built from illustration data provided by creators who gave explicit consent for use in machine learning.