The translated data: Image synthesis is a significant research area in the field of computer vision, where researchers have proposed a new image synthesis model called LCM. This model can generate high-resolution images with fewer inference steps. LCM is derived from a pre-trained latent diffusion model (LDM) and can directly predict the solution of the probability flow ODE in the latent space, reducing the number of iterations and computational load. The paper also introduces a new fine-tuning method called LCF, which allows for fine-tuning the pre-trained LCM on customized image datasets. LCM achieves state-of-the-art performance in text-to-image generation tasks and boasts advantages such as speed, high fidelity, low distortion, and low noise.