【AIイラスト】ふにゃふにゃの線画しか描けなくても美麗イラストが作れる方法を解説(Stable Diffusion / Scribble)

This paragraph discusses the process of creating an illustration of a popular character named Hoshino Ai using AI. The challenge lies in replicating a specific pose that is difficult for AI to generate due to its complexity. The speaker decides to use a rough sketching tool called 'Scribble' to create the illustration. The paragraph highlights the limitations of AI in replicating certain poses and the importance of learning methods to specify poses accurately. It also introduces the concept of 'Image2Image', a technique used in recent AI-generated videos, and its role in creating realistic illustrations.

The speaker delves into the use of the 'Scribble' function, which allows for the creation of illustrations from rough sketches. The paragraph explains how to input a prompt and use AI to generate a clean illustration, even for those who cannot draw well. It describes the process of refining the illustration by adjusting the pose and proportions, and the use of 'Control Net' for more precise control over the final image. The paragraph also touches on the ability to extract poses from photographs and illustrations, and the potential for further refinement using AI.

In this paragraph, the speaker concludes the illustration process by discussing the final steps of refining the AI-generated image. They explore the option of adjusting the noise level and seed value to achieve the desired look. The paragraph also covers the creation of a base painting for the illustration and the use of 'Control Net' to extract lines from the original sketch. Finally, it mentions the possibility of enhancing the illustration by adding details such as eye highlights and adjusting the color scheme to better match the original image.