StableDiffusionVN / SDVN_Comfy_node

SDVN Comfy Node **Smart node set, supporting easier and more convenient ways to use ComfyUI** ___ **Hướng dẫn tiếng việt** **Installing** **Guide** • ✨ Base Node • 🏞️ Image • 📥 Download • 🧬 Merge • 💡 Creative • 👨🏻‍💻 Dev • 💬 API • ℹ️ Info_check • ✨ Preset • 🎭 Mask **Example** ___ Overview • A full toolkit of smart ComfyUI nodes covering loading/downloading, merging, masks, layouts, and API-based generation for SD15, SDXL, Flux, and more. • Built-in conveniences: auto model-type detection (CFG/sampler/scheduler), translation + dynamic prompt support across text fields, ControlNet/Apply Style presets, and curated model/style libraries ( , , ). • Download helpers for checkpoints, LoRA, ControlNet, upscalers, IPAdapter/InstantID, and image sources (Pinterest/URL/Instagram) with aria2c acceleration when available. • Utilities for metadata inspection/editing, streamlined wiring (Pipe/Switch/Any nodes), mask tools, and API nodes for Gemini, OpenAI, Hugging Face, and Deepseek. • Workflow examples live in the folder; preview images are in to quickly show what each node set does. Todo • [x] Workflow Example • [x] Guide • [x] Install ___ Quick start • Clone into , then install Python deps from the ComfyUI root: • macOS/Windows: install manually for download nodes. • (Optional) Rename to and fill in keys; copy to to customize style presets. • Start ComfyUI — nodes appear under SDVN categories; recommended companion nodes are listed below. ___ Install Install with simple commands: • • • From your ComfyUI root, install dependencies: • *For Windows or macOS machines, users are required to manually install in order to use nodes that automatically download models.* Also you should install the following nodes to be able to use all functions: • Dynamicprompts • Inpaint crop • TiledDiffusion • IPAdapter_plus • Controlnet_aux ___ Guide • **For all nodes with download:** Supports direct download from **civitai** and **huggingface** with model address link and model download link • **For all dodes load photos with the URL:** Automatically download photos to the url of the image. Can automatically search for the highest quality image with the Pinterest link . See also the support list • **For all nodes capable of entering the text:** Support **Google Translate** and **Dynamic Prompt function** ((Wildcards folder: /ComfyUI/wildcards)) ___ BaseNode *A collection of smart nodes that replace basic tasks, helping users build processes smarter and faster* **📀 Load checkpoint / 🎨 Load Lora** Supports 2 methods of loading checkpoint and downloading checkpoint directly for use. • If you leave the Download_url information, checkpoint will be selected according to the Ckpt_name • If you enter the checkpoint download url and leave the Download - True option, the checkpoint will be downloaded to checkpoints/loras folder and named Ckpt_url_name Additionally, these two nodes support a curated list of commonly used checkpoints and LoRAs. If the models are not already available in the folder, they will be automatically downloaded. **🏞️ Load Image / 🏞️ Load Image Url** • Support 2 images download methods from input folders and URL links / Image Path • Support sub-folders in the input folder **🏞️ Load Image Folder** • Load a list with number images from a folder. • If random = True, the images will be completely random; otherwise, they will be a sequence of images that are adjacent to each other. • If , the entire folder of images will be loaded. **🏞️ Load Pinterest** Automatically search and download images from Pinterest. Images will be automatically downloaded to the input folder, organized into separate subfolders. • Url: • Accepts any Pinterest image link (single image, board, board section, or profile page) (e.g., https://www.pinterest.com/...). If the link starts with a slash /, the prefix https://www.pinterest.com will be automatically added (e.g., /abc/vintage ⇨ https://www.pinterest.com/abc/vintage). • Accepts any keyword and will automatically search for images on Pinterest using that keyword. • Range: Location – Number of images that will be downloaded to the input folder. • Number: The number of images that will be loaded and returned. If random = True, the images will be randomly selected from the downloaded list; otherwise, the images will be selected based on their position determined by the seed. **🏞️ Load Image Ultimate** This is a powerful node that combines 5 different loading modes (Input Folder, Custom Folder, Pinterest, Insta, URL) to enhance flexibility in workflows. The options will automatically adjust based on the selected mode, and the usage is similar to the corresponding image loading nodes above. **🔡 CLIP Text Encode** • Simultaneously support both Positive, Negative • Support Random ability with Dynamic Prompt (Request installed node Dynamicprompts) • Support Translate function • Support Style Card. • Includes helper text encoders for advanced models: **Qwen Edit Text Encoder / Plus** and **Kontext Reference** (pass reference images/masks into conditioning/latent). **🗂️ Prompt Styles** • Supports easier prompt writing with pre-saved and curated style presets. • You can add or edit style cards by renaming and modifying the file my_styles.csv.example to my_styles.csv. **🎚️ Controlnet Apply** Provide full option to use ControlNet in a single node (Request installed node Controlnet Aux) • Can choose Controlnet Model, Preprocessor (Automatically detect Aux Preprocessor Aux Preprocessor + Add Invert Image option), Union Type • Show preview pictures Preprocessor when running • Supports automatic downloading of most popular ControlNet models for SD15, SDXL, and Flux. • Supports direct usage with ControlNet Inpaint Alimama Flux. • Supports exporting parameters for integration with the AutoGenerate node. • Extras: **Diffsynth Controlnet Apply** and **Diffsynth Union Lora Apply** for users who work with Diffusynth workflows. **🌈 Apply Style Model** • Supports autom…