diff --git a/docs/getting-started.md b/docs/getting-started.md index 289d19f8..8a221a2a 100644 --- a/docs/getting-started.md +++ b/docs/getting-started.md @@ -86,9 +86,9 @@ mark-of-the-web check to ensure the server runs successfully. Upon extraction, i file. Once model conversion is complete, the server will commence, opening a browser window for immediate access to the web UI. -### Cross platform setup +For more details, please see [the setup guide](./setup-guide.md#windows-all-in-one-bundle). -Link to the other methods. +### Cross platform setup Users who have a working Python environment and prefer installing their own dependencies can opt for a cross-platform installation using a Python virtual environment. Ensure a functioning Python setup with either pip or conda. Begin by @@ -96,6 +96,8 @@ cloning the onnx-web git repository, followed by installing the base requirement platform. Execute the launch script, patiently waiting for the model conversion process to conclude. Post-conversion, open your browser and load the web UI for interaction. +For more details, please see [the setup guide](./setup-guide.md#cross-platform-method). + ### Server setup with containers For server administrators, onnx-web is also distributed as OCI containers, offering a streamlined deployment. Choose the @@ -103,6 +105,8 @@ installation method that aligns with your proficiency and system requirements, w bundle, the intermediate cross-platform setup, or the containerized deployment for server admins. Each pathway ensures a straightforward onnx-web installation tailored to your technical needs. +For more details, please see [the server admin guide](./server-admin.md#containers). + ## Running Running onnx-web is the gateway to unlocking the creative potential of Stable Diffusion in AI art. Whether you are a @@ -140,42 +144,56 @@ The txt2img tab in onnx-web serves the purpose of generating images from text pr descriptions and witness the algorithm's creative interpretation, providing a seamless avenue for text-based image generation. +For more details, please see [the user guide](./user-guide.md#txt2img-tab). + ### Img2img Tab For image-based prompts, the img2img tab is the go-to interface within onnx-web. Beyond its fundamental image generation capabilities, this tab introduces the ControlNet mode, empowering users with advanced control over the generated images through an innovative feature set. +For more details, please see [the user guide](./user-guide.md#img2img-tab). + ### Inpaint Tab The inpaint tab specializes in image generation with a unique combination of image prompts and masks. This functionality allows users to guide the algorithm using both the source image and a mask, enhancing the precision and customization of the generated content. +For more details, please see [the user guide](./user-guide.md#inpaint-tab). + ### Upscale Tab Addressing the need for higher resolutions, the upscale tab provides users with tools for high resolution and super resolution. This feature is particularly useful for enhancing the quality and clarity of generated images, meeting diverse artistic and practical requirements. +For more details, please see [the user guide](./user-guide.md#upscale-tab). + ### Blend Tab Enabling users to combine outputs or integrate external images, the blend tab in onnx-web offers a versatile blending tool. This functionality adds a layer of creativity by allowing users to merge multiple outputs or incorporate external images seamlessly. +For more details, please see [the user guide](./user-guide.md#blend-tab). + ### Models Tab Central to managing the core of onnx-web, the models tab provides users with the capability to manage Stable Diffusion models. Additionally, it allows for the management of LoRAs (Latents of Random Ancestors) associated with these models, facilitating a comprehensive approach to model customization. +For more details, please see [the user guide](./user-guide.md#models-tab). + ### Settings Tab Tailoring the user experience, the settings tab is the control center for managing onnx-web's web UI settings. Users can configure server APIs, toggle dark mode for a personalized visual experience, and reset other tabs as needed, ensuring a user-friendly and customizable environment. +For more details, please see [the user guide](./user-guide.md#settings-tab). + ## Image parameters In onnx-web, image parameters play a pivotal role in shaping the output of the Stable Diffusion process. These @@ -184,6 +202,8 @@ height, collectively govern the characteristics of the diffusion model's trainin ### Common image parameters +These parameters are part of the Stable Diffusion pipeline and common to most tools. + - Scheduler - Role: The scheduler dictates the annealing schedule during the diffusion process. - Explanation: It determines how the noise level changes over time, influencing how the diffusion process resolves @@ -226,6 +246,8 @@ height, collectively govern the characteristics of the diffusion model's trainin ### Unique image parameters +These parameters are unique to how onnx-web generates images. + - UNet tile size - One such parameter is the UNet tile size. This parameter governs the maximum size for each instance the UNet model runs. While it aids in reducing memory usage during panoramas and high-resolution processes, caution is needed. @@ -269,6 +291,8 @@ networks can be effectively utilized at higher or lower values, spanning from -1 sliders. This flexibility provides users with nuanced control over the influence of LoRA networks on the generated images. +For more details, please see [the user guide](./user-guide.md#lora-and-lycoris-tokens). + ### CLIP skip `` for anime. @@ -279,6 +303,8 @@ image results. For instance, skipping 2 levels refines the output by bypassing s optimizing the creative outcome. This combination of tokens and functionalities enables users to precisely tailor prompts in onnx-web for expressive image generation. +For more details, please see [the user guide](./user-guide.md#clip-skip-tokens). + ## Highres onnx-web supports a unique highres implementation, a powerful tool for super-resolution upscaling followed by img2img @@ -300,6 +326,8 @@ recursive image features. Highres prompts are separated from the base txt2img prompt using the double pipe syntax (`||`). These prompts guide the upscaling and refinement processes, enabling users to incorporate distinct instructions and achieve nuanced outputs. +For more details, please see [the user guide](./user-guide.md#prompt-stages). + ### Highres iterations `scale ** iterations` @@ -311,6 +339,8 @@ original size. This scaling effect continues exponentially with each additional prompts and iterations allows users to progressively enhance image resolution while maintaining detailed and refined visual elements. +For more details, please see [the user guide](./user-guide.md#highres-iterations-parameter). + ## Profiles The onnx-web web UI simplifies user experience with the introduction of a feature known as profiles. When you find @@ -370,6 +400,8 @@ functionality proves invaluable for adding characters to backgrounds, introducin precise control over where elements appear. It becomes a powerful tool for avoiding crowded or overlapping elements, offering nuanced control over image composition. +For more details, please see [the user guide](./user-guide.md#region-tokens). + ### Region seeds `` @@ -379,6 +411,8 @@ times within the same image. To prevent hard edges and seamlessly integrate thes region prompts and region seeds include options for blending with the surrounding prompt or seed. It's important to note that these region features are currently exclusive to the panorama pipeline. +For more details, please see [the user guide](./user-guide.md#reseed-tokens-region-seeds). + ## Grid mode onnx-web introduces a powerful feature known as Grid Mode, designed to facilitate the efficient generation of multiple