Automatic

promptgen-lexart

AUTOMATIC

47

promptgen-lexart is a text generation model created by AUTOMATIC and fine-tuned on 134,819 prompts scraped from Lexica.art, the Stable Diffusion 1.5 checkpoint. This model is intended for use with the Stable Diffusion WebUI Prompt Generator tool, allowing users to generate new text prompts for Stable Diffusion image generation. It builds upon the pre-trained DistilGPT-2 model, resulting in a more specialized and efficient prompt generation system. Model inputs and outputs promptgen-lexart takes in a seed text prompt as input and generates a new, expanded prompt text as output. This can be useful for quickly ideating new prompts to use with text-to-image models like Stable Diffusion. Inputs A seed text prompt, e.g. "a cat sitting" Outputs A new, expanded prompt text, e.g. "a tabby cat sitting elegantly on a plush velvet armchair, detailed fur, intricate texture, highly detailed, cinematic lighting, award winning photograph" Capabilities promptgen-lexart can generate diverse and detailed text prompts that capture a wide range of visual concepts and styles. By leveraging the knowledge gained from the Lexica.art dataset, the model is able to produce prompts that are well-suited for use with Stable Diffusion. What can I use it for? The promptgen-lexart model can be a valuable tool for text-to-image workflows, allowing users to rapidly explore new prompt ideas and refine their prompts for higher quality image generation. It can be used in conjunction with Stable Diffusion or other text-to-image models to streamline the ideation and prompt engineering process. Things to try Try seeding the model with different starting prompts and observe how it expands and refines the text. Experiment with different temperature and top-k settings to control the diversity and quality of the generated prompts. You can also try incorporating the model into your own text-to-image pipelines or webapps to automate the prompt generation process.

Read more