Wiki History Listing
| V4 |
NovelAI, a fork of the Stable Diffusion image generation software. It differs from Stable Diffusion in that it focuses on anime-style images labeled with booru tags. It takes a string of words as an input prompt, runs them through an AI algorithm trained on a database of labeled images, and tries to output an image that fits the inputted prompt. The input can be programmed in various ways beyond the words themselves, including using a base image as the starting point ("image-to-image"). Specific parts of images can be targeted for modification ("inpainting"). Due to the software's aptitude for front or rear portraits, along with the low barrier of entry, most NAI images that are not added by the bot are self-uploads. If you are uploading an image you generated, please make sure of the following: -If it's image-to-image and it's only a minor modification, such as inpainting or a new background, tag the original artist and tag it as a [[third-party_edit]]. -If it's image-to-image and you are not the artist of the seed image, [i]do not tag yourself as the artist of the output image[/i]. Until the administration confirms whether image-to-image art should be credited to the seed artist when it is unrecognizable, this is to avoid edit wars. -Do not add tags that were part of the prompt unless they are visible in the image. However, if the image is published elsewhere with a description, you may include canonical characteristics such as [[humanoid]] and [[incest]]. See also: [[stable_diffusion]] Updated by Mukokuseki Sat, May 20 '23, 22:39 |
| V3 |
NovelAI, a fork of the Stable Diffusion image generation software. It takes a string of words as an input prompt, runs them through an AI algorithm trained on a database of labeled images, and tries to output an image that fits the inputted prompt. The input can be programmed in various ways beyond the words themselves, including using a base image as the starting point ("image-to-image"). Specific parts of images can be targeted for modification ("inpainting"). It differs from Stable Diffusion in that it focuses on anime-style images labeled with booru tags. Due to the software's aptitude for front or rear portraits, along with the low barrier of entry, most NAI images that are not added by the bot are self-uploads. If you are uploading an image you generated, please make sure of the following: -If it's image-to-image and it's only a minor modification, such as inpainting or a new background, tag the original artist and tag it as a [[third-party_edit]]. -If it's image-to-image and you are not the artist of the seed image, [i]do not tag yourself as the artist of the output image[/i]. Until the administration confirms whether image-to-image art should be credited to the seed artist when it is unrecognizable, this is to avoid edit wars. -Do not add tags that were part of the prompt unless they are visible in the image. However, if the image is published elsewhere with a description, you may include canonical characteristics such as [[humanoid]] and [[incest]]. See also: [[stable_diffusion]] Updated by Mukokuseki Fri, Nov 11 '22, 22:47 |
| V2 |
NovelAI, a fork of the Stable Diffusion image generation software. It takes a string of words as an input prompt, runs them through an AI algorithm trained on a database of labeled images, and tries to output an image that fits the inputted prompt. The input can be programmed in various ways beyond the words themselves, including using a base image as the starting point ("image-to-image"). Specific parts of images can be targeted for modification ("inpainting"). It differs from Stable Diffusion in that it focuses on anime-style images labeled with booru tags. Due to the software's aptitude for front or rear portraits, along with the low barrier of entry, most NAI images that are not added by the bot are self-uploads. If you are uploading an image you generated, please make sure of the following: -If it's image-to-image and it's only a minor modification, such as inpainting or a new background, tag the original artist and tag it as a [[third-party_edit]]. -If it's image-to-image and you are not the artist of the seed image, [i]do not tag yourself as the artist of the output image[/i]. Until the administration confirms whether image-to-image art should be credited to the seed artist when it is unrecognizable, this is to avoid edit wars. -Do not add tags that were part of the prompt unless they are visible in the image. However, if the image is published elsewhere with a description, you may include canonical characteristics such as [[humanoid]] and [[incest]]. Updated by Mukokuseki Fri, Nov 11 '22, 22:47 |
| V1 |
NovelAI, a fork of the Stable Diffusion image generation software. It takes a string of words as an input prompt, runs them through an AI algorithm trained on a database of labeled images, and tries to output an image that fits the inputted prompt. The input can be programmed in various ways beyond the words themselves, including using a base image as the starting point ("image-to-image"). Specific parts of images can be targeted for modification ("inpainting"). It differs from Stable Diffusion in that it focuses on anime-style images labeled with booru tags. Due to the software's aptitude for front or rear portraits, along with the low barrier of entry, most NAI images that are not added by the bot are self-uploads. If you are uploading an image you generated, please make sure of the following: -If it's image-to-image and it's only a minor modification, such as inpainting or a new background, tag the original artist and tag it as a third-party_edit. -If it's image-to-image and you are not the artist of the seed image, [i]do not tag yourself as the artist of the output image[/i]. Until the administration confirms whether image-to-image art should be credited to the seed artist when it is unrecognizable, this is to avoid edit wars. -Do not add tags that were part of the prompt unless they are visible in the image. However, if the image is published elsewhere with a description, you may include canonical characteristics such as humanoid and incest. Updated by Mukokuseki Fri, Nov 11 '22, 22:46 |