--- datasets: - rullaf/mtg-art pipeline_tag: text-to-image tags: - art --- # Magic Diffusion A text2img model derived from StableDiffusion 1.5, fine-tuned with [EveryDream-trainer](https://github.com/victorchall/EveryDream-trainer) on a dataset consisting of post-processed Magic the Gathering card art crops (32,159), and hi-resolution images of the art (13,048). Annotations are based on card metadata and various other sources, including art description. ## Comparison * Magic Diffusion v2 – this model * Fantasy Card Diffusion v1 – https://huggingface.co/volrath50/fantasy-card-diffusion * Open Journey v2 – https://huggingface.co/prompthero/openjourney-v2 * Stable Diffusion 1.5 – https://huggingface.co/runwayml/stable-diffusion-v1-5 All images were generated with identical settings: * 40 steps * 512x512 * seed `1111 + 1234*n` (`1111`, `2345`, `3579`, `4813`, `6047`, `7281`, `8515`, `9749`) 1111 2345 3579 4813 6047 7281 8515 9749 ### Taylor Prompt > mtg card art Taylor Swift wandering bard legendary creature human bard by chris rahn by volkan baga by zoltan boros armored bard taylor swift holding her weapons and instruments beautiful composition detailed realistic fantasy painting masterpiece best quality Negative Prompt > guitar lowres bad anatomy bad hands text error missing fingers extra digit fewer digits cropped worst quality low quality normal quality jpeg artifacts signature watermark username blurry
Magic Diffusion v2 | Fantasy Card Diffusion v1 | Openjourney v2 | Stable Diffusion 1.5 |
---|---|---|---|