Overview Following up from my previous article, I continued to finalize the artwork by iterating between manual and AI rendering. This time, I aimed to render in a realistic style using 3D. …

Overview I explored a workflow that utilizes ControlNet to generate images aligned with specific intentions. In this study, I considered a method where drawings and colorings are gradually refined with iterations of hand corrections and AI enhancements. Below are the results at each stage of the workflow. Objective Drawing using Stable Diffusion…

Abstract I have experimented with PanoHead, which can generate a full 360-degree 3D model from a single image of a head. GitHub - SizheAn/PanoHead: Code Repository for CVPR 2023 Paper "PanoHead: Geometry-Aware 3D… Code Repository for CVPR 2023 Paper "PanoHead: Geometry-Aware 3D Full-Head Synthesis in 360 degree" - GitHub …github.com Using a image of a head generated by Stable Diffusion, I confirmed the possibility of 3D conversion. …

日本語 (Japanese): https://note.com/takumi__ncr/n/n120130e9add5 I am currently carrying out personal research and have reached a point where I can share some preliminary results. Overview The main tasks performed were as follows: 1: Character Design using AI Using ControlNet, I designed an original character. 2: Manual 3D Modeling Based on the character design…