Description
Methexis-Inc/img2prompt is a tool that produces approximate text prompts that correspond to a given image. It's specifically fine-tuned for stable-diffusion (clip ViT-L/14).
This tool leverages the open-source CLIP Interrogator notebook developed by @pharmapsychotic and employs OpenAI CLIP models to associate an image with different artists, media, and artistic styles.
The results of this comparison are then combined with BLIP captions to develop a text prompt which can then be used to generate new images that are visually similar to the original.
The tool can be used through an API, or the GitHub repository and related license can be reviewed for more details. Predictions usually finish within 24 seconds, running on Nvidia T4 GPU hardware.
Pricing Plans
Releases
Reviews
Pros & Cons
Pros
Optimized for stable diffusion
Leverages CLIP models
Performs comparative image analysis
Cons
Specifically optimized for stable-diffusion
Only runs on Nvidia T4 GPUs
Depends on BLIP captions
