Design of Printed Pattern Generation Method Based on Diffusion Models
It is a crucial aspect to design printed patterns in the field of clothing production.However,manually designed patterns often have the shortages of content similarity and low design efficiency.Therefore,a printed pattern generation method based on a diffusion model was designed and implemented.This method employs a deep learning technology to extract and expand existing prin-ted pattern datasets,generate the textual descriptions of these patterns from the dimensions of color and category,and complete the production of the dataset for printed patterns.The pre-made dataset is used to make small adjustments to the diffusion model and flat-ten the processing of the image feature space,which makes the generation patterns at the boundary transition edges meet the require-ments of the textile industry.This paper analyzes the characteristics of local diffusion,and achieves a image generation effect with variable detail from images and text.Experimental results demonstrate that the designed pattern generation method is capable of pro-ducing high-quality patterns,and its feature space tiling method makes the transition of the printed pattern boundary more smooth.
万方数据
维普
