A Survey on Various Algorithms to Improve the Efficiency and Features of Text to Image Generation

Generation of realistic images is a versatile domain that has several applications in picture editing, video games, and computer-aided design. These applications include creating realistic visuals that semantically match the provided text descriptions. Text-to-image generator is used for generating images by collecting all the related images. It has a significant influence on many domains and different sets of applications. Existing solutions only help in generating an image from the given text with low accuracy and do not contain many features. Animators are in need to view their creative ideas in efficient manner. Considering these issues, the design for the proposed system is carried out. The proposed solution takes images from the internet according to the text provided for generation and enhances the generated image quality. 3D model and animated video construction are added as the additional features. Deep neural network model helps in creating images from text captions and constructing 3D model for an extensive range of topics that can be expressed using natural language. By combining various models, a new algorithm is proposed for generating images with high resolution from the given text prompt. Animated video is generated from different frames of an image. The proposed system stands as a unique solution for the users to envision their ideas.