Waifu-diffusion
Models by this creator
🔎
wd-1-5-beta2
127
wd-1-5-beta2 is a text-to-image diffusion model fine-tuned by the waifu-diffusion team on high-quality anime images. It is an updated version of the Waifu Diffusion v1.4 model, which was conditioned on anime-styled images through fine-tuning the Stable Diffusion 1.4 model. The current model, wd-1-5-beta2, has two versions: the base version and an "aesthetic" version that was further fine-tuned on popular aesthetic images. The model uses the same VAE (Variational Autoencoder) as the previous Waifu Diffusion v1.4 model, which can be found here. This VAE was originally trained on anime-style images. Model inputs and outputs Inputs Text prompt describing the desired image Outputs An image generated based on the input text prompt Capabilities The wd-1-5-beta2 model is capable of generating high-quality anime-style images based on text prompts. It can create a wide variety of scenes and characters, from portraits to landscapes, with a distinctive anime aesthetic. What can I use it for? The wd-1-5-beta2 model can be used for creative and entertainment purposes, such as generating anime-inspired artwork, character designs, and concept art. It could be utilized by artists, illustrators, and hobbyists to aid in their creative process or to generate unique and compelling images. Things to try One interesting aspect of the wd-1-5-beta2 model is the "aesthetic" version, which was further fine-tuned on popular aesthetic images. This version may be able to generate images with a more refined and polished anime style, potentially capturing the look and feel of high-quality anime illustrations. Experimenting with prompts that focus on aesthetic qualities, such as "masterpiece, best quality, highly detailed," could yield visually striking results.
Updated 5/28/2024
📶
wd-1-5-beta3
117
The wd-1-5-beta3 model, created by the waifu-diffusion team, is a text-to-image diffusion model trained on high-quality anime images. It builds upon the previous WD 1.5 Beta 2 and WD 1.5 Beta versions, with five new aesthetic variations - WD 1.5 Radiance, WD 1.5 Ink, WD 1.5 Mofu, and WD 1.5 Illusion. The base WD 1.5 Beta3 model is intended primarily for training use, while the aesthetic variants are recommended for generation. Model inputs and outputs The wd-1-5-beta3 model takes text prompts as input and generates corresponding anime-style images as output. The text prompts can describe a wide variety of anime-themed subjects, characters, and scenes. Inputs Text prompts:** Short to medium-length descriptions of the desired anime-style image Outputs Generated images:** Anime-style images that match the input text prompts Capabilities The wd-1-5-beta3 model is capable of generating a wide range of high-quality anime-style images from text prompts. It can create illustrations of characters, scenes, and more, with a distinct anime aesthetic. The five aesthetic variations offer different stylistic approaches, allowing users to explore diverse artistic interpretations. What can I use it for? The wd-1-5-beta3 model can be used for various creative and entertainment purposes, such as: Generating concept art and illustrations for anime-inspired projects Creating custom anime-style avatars or character designs Producing unique and personalized anime-themed artwork The model's versatility allows users to explore their creativity and potentially monetize their work, for example, by selling generated images or offering custom illustration services. Things to try One interesting aspect of the wd-1-5-beta3 model is the ability to fine-tune it or create custom variations using the provided WD 1.5 Base model. This allows users to further customize the model's outputs to their specific needs or preferences. Experimenting with prompt engineering and the aesthetic variants can also lead to unique and unexpected results.
Updated 5/28/2024
🔮
wd-1-5-beta
116
wd-1-5-beta is a beta version of the Waifu Diffusion model, which is a latent text-to-image diffusion model fine-tuned on high-quality anime images. It builds upon the Waifu Diffusion v1.3 and Waifu Diffusion v1.4 models, with further improvements and enhancements. This beta model is not yet finalized, but provides a preview of the upcoming Waifu Diffusion 1.5 release. Model inputs and outputs wd-1-5-beta is a text-to-image generation model, taking in text prompts and outputting corresponding images. The model leverages the same VAE as Waifu Diffusion v1.4, which can be found at https://huggingface.co/hakurei/waifu-diffusion-v1-4/blob/main/vae/kl-f8-anime2.ckpt. Inputs Text prompt describing the desired image Outputs Generated image corresponding to the input text prompt Capabilities The wd-1-5-beta model is capable of generating high-quality anime-style images from text prompts. It includes aesthetic embeddings to help improve the quality and consistency of the generated images. The model performs best when generating images at resolutions between 500 and 1000 pixels, and then using a 2x latent upscale hiresfix. What can I use it for? wd-1-5-beta can be used for a variety of creative and entertainment purposes, such as generating anime-style artwork, character designs, and illustrations. The model is released under the Fair AI Public License 1.0-SD, which allows for commercial use and distribution of derivative works, as long as the license terms are followed. Things to try With the wd-1-5-beta model, it's recommended to experiment with different prompting techniques and use the provided aesthetic embeddings to help improve the quality of the generated images. The model's capabilities are still in development, so users should expect some variability in the results, but the overall quality and consistency of the outputs is quite impressive.
Updated 5/28/2024