# ⭐ ComfyUI StarNodes **A big, friendly toolbox of 100 custom nodes that make ComfyUI easier, faster and more fun.** *Starters • Samplers • Image tools • Qwen & Flux helpers • Video • PSD export • Wildcards • and much more* [![Version](https://img.shields.io/badge/version-2.7.1-blueviolet?style=for-the-badge)](#) [![ComfyUI](https://img.shields.io/badge/ComfyUI-custom%20nodes-orange?style=for-the-badge)](#) [![License](https://img.shields.io/github/license/Starnodes2024/ComfyUI_StarNodes?style=for-the-badge)](LICENSE) [![GitHub](https://img.shields.io/badge/GitHub-Starnodes2024-black?style=for-the-badge&logo=github)](https://github.com/Starnodes2024/ComfyUI_StarNodes)

--- ## 🌟 What is StarNodes? StarNodes is a **swiss-army-knife node pack** for [ComfyUI](https://github.com/comfyanonymous/ComfyUI). It fills the gaps you hit every day: starting a workflow with one node instead of five, saving real layered **PSD files**, browsing **360° panoramas**, batching images into **grids with captions**, managing **wildcards**, **LoRAs**, **sampler presets** — plus a full set of **Qwen / Flux / LTX-Video** helpers. > 💡 **In one sentence:** fewer wires, fewer clicks, better defaults — everything lives under one **⭐StarNodes** menu. --- ## ✨ Recently Added & Updated — Spotlight These are the newest and most powerful nodes in the pack. If you try nothing else, try these:

### 🆕 ⭐ Star LTXV 2.5 All-in-One **The complete LTX-2.5 pipeline in a single node.** Text-to-video, image-to-video, image+audio, first/last-frame and audio-only modes — two-pass upscale rendering, baked sigma or plain-step schedules, model/LoRA/CLIP/VAE caching, and sound decoded from the high-step pass. ### 🆕 ⭐ Star Video Sound Enricher **Fix the scratchy AI-video soundtrack.** De-harsh bell, deep-bass and warmth boost, high-fizz taming and gentle analog saturation — four tuned presets plus a full Custom mode, always delivered at 44.1 kHz. ### 🆕 ⭐ Star Video Sound Enricher Option **Same sound magic, zero extra wires.** Outputs the enricher settings as a `sound_settings` bundle — plug it into the LTXV 2.5 All-in-One and the soundtrack comes out cleaned and enriched internally. ### 🆕 ⭐ Star SD Upscale Refiner Advanced **SD1.5 upscaling + refinement in one node.** Built-in LoRA support, tiled diffusion for low VRAM, and optional ControlNet tile guidance — a complete "make it big and beautiful" pipeline. ### 🆕 ⭐ Star Minimax All In One **The whole MiniMax H3 reference-to-video pipeline in a single node.** Models, text encoder and both VAEs load internally; feed reference images, videos and audios, then sample and decode video + audio — no sub-graph. Includes the new **2:1 (Panorama)** preset. ### 🆕 ⭐ StarSampler (Unified) **One sampler to rule them all.** Extensive configuration in a single node, with a tiled VAE decoder so even big images finish on modest GPUs. Replaces stacks of sampler plumbing. ### 🆕 ⭐ Star Split Sampler Option **Two samplers, one run.** Plug into the `options` input of ⭐ StarSampler or ⭐ Star SD Upscale Refiner to run the first N steps with one sampler and the rest with another — e.g. `euler` for 6 steps then `ddim` for 6, for 12 steps total. Mix sampler strengths without rewiring anything.

### 🆕 ⭐ Star Advanced Ratio/Latent **Aspect ratio + megapixels, done for you.** Pick a ratio and a target size — get perfect dimensions and a ready-to-use empty latent, no calculator needed. ### 🔄 ⭐ Star Output Cleaner **Your output folder, finally under control.** Browse thumbnails, select and delete old generations without ever leaving ComfyUI. ### 🔄 ⭐ Star Image Compare **Judge your results like a pro.** Interactive before/after wipe slider right inside ComfyUI — drag to compare two images pixel by pixel. Perfect for checking upscales, refiners and filter tweaks. ### 🔄 ⭐ Star Panorama Tools **Create, save and explore 360° worlds.** The 360 Parallax Viewer (and Pro) let you look around panoramas interactively with mouse parallax, overlays and depth maps — while Save Panorama JPEG/+ embeds proper XMP metadata, including stereoscopic 3D output. ### 🔄 ⭐ Star Video Compressor **Shrink videos to any target.** Compress to a quality level *or an exact file size* with H.264, H.265, VP9 or AV1 — no external tools needed. ### 🔄 ⭐ Star Slideshow Maker **Images in, movie out.** Transitions, motion effects and audio, rendered through FFmpeg straight from your workflow. ### 🔄 ⭐ Star Tiled Upscalers (SeedVR & PiD) **Huge upscales on small GPUs.** Tile-based upscaling with SeedVR2 (with color correction) or PiD/PixelDiT models — overlapping tiles mean no seams, no VRAM panic.

> 🆕 = recently added • 🔄 = recently updated --- ## 🚀 Installation ### Option A — ComfyUI Manager (recommended, 1 minute) 1. Open **ComfyUI Manager** 2. Search for **`Starnodes`** 3. Click **Install** → restart ComfyUI ✅ ### Option B — Manual install ```bash cd ComfyUI/custom_nodes git clone https://github.com/Starnodes2024/ComfyUI_StarNodes cd ComfyUI_StarNodes pip install -r requirements.txt ``` Restart ComfyUI. Done. 🎉 > **Where are the nodes?** Press **double-click** on the canvas and search **`star`** — > or open the **⭐StarNodes** category in the node browser. --- ## 🧭 Quick Start (60 seconds) | I want to… | Use this node | |---|---| | Start a Flux / SDXL / Qwen workflow fast | ⭐ **Star Start(t) Settings** nodes | | Upscale a latent with a model | ⭐ **Star Model Latent Upscaler** | | Save my image with all metadata | ⭐ **Star Save Image+** | | Export a layered PSD file | ⭐ **Star PSD Saver** family | | Use wildcards in prompts | ⭐ **Star Wildcards Advanced** | | Stack multiple LoRAs | ⭐ **Star Dynamic LoRA** | | Debug any connection | ⭐ **Star Show Everything** | | Make a grid of images with captions | ⭐ **Star Grid Composer** | --- ## 📦 What's Inside Click a category to expand it. All categories appear in ComfyUI exactly as named below.

🚀 Starters — one node to kick off a whole workflow

| Node | What it does | |------|--------------| | ⭐ Star FLUX Start(t) Settings | Load UNET + 2 CLIPs and create an empty latent in one step | | ⭐ Star SD(XL) Start(t) Settings | Load checkpoint + CLIP + VAE and create an empty latent | | ⭐ Star Qwen Image Start(t) Settings | Load a Qwen model and create an empty latent |

🎛️ Sampler — sampling, LoRAs, inpainting, refinement

| Node | What it does | |------|--------------| | ⭐ StarSampler (Unified) | Advanced all-in-one sampler with tiled VAE decode for low VRAM | | ⭐ Star Split Sampler Option | Run two different samplers in one pass — e.g. euler 6 steps + ddim 6 steps | | ⭐ Star Split Sampler Info | Split sampler info into its individual components | | ⭐ Star 3 LoRAs | Apply up to 3 LoRAs with individual weights | | ⭐ Star Dynamic LoRA | Flexible multi-LoRA node with dynamic slots | | ⭐ Star Dynamic LoRA (Model Only) | Same, without touching CLIP | | ⭐ Star Dynamic LoRA Weight | Normalize and fine-tune LoRA weights | | ⭐ Star Detail Daemon | Enhance details for Flux & all SD models | | ⭐ Star FluxFill Inpainter | Inpainting for Flux with optimized conditioning | | ⭐ Star Flux2/Qwen-Image-Edit Inpainter | All-in-one inpaint: crop-and-stitch, ref image, Differential Diffusion | | ⭐ Star Distilled Optimizer (QWEN/ZIT) | Two-pass distilled refinement for Z-Image-Turbo / Qwen Turbo |

🖼️ Image & Latent — filters, upscalers, grids, panoramas, PSD export

| Node | What it does | |------|--------------| | ⭐ Star Model Latent Upscaler | Complete latent upscale pipeline with model choice | | ⭐ Star Image2Latent | Convert an image to a latent | | ⭐ Star Latent Resize | Resize latents with ratio / megapixel selector | | ⭐ Star Advanced Ratio/Latent | Aspect ratio + latent megapixel helper | | ⭐ Starnodes Aspect Ratio Advanced | Resolution selection with aspect-ratio math done for you | | ⭐ Star Adaptive Detail Enhancement | Sharpen / denoise via edge, face & texture analysis | | ⭐ Star Apply Overlay (Depth) | Blend filtered images using depth or masks | | ⭐ Star Simple Filters | Sharpen, blur, saturation, contrast, brightness, color matching | | ⭐ Star HighPass Filters | High-pass sharpening for fine detail | | ⭐ Star Black & White | Flexible B&W conversion with tonal control | | ⭐ Star Radial Blur | Focus / zoom motion blur effects | | ⭐ Star HDR Effects | HDR-style enhancement | | ⭐ Star Realistic Film Grain | Analog grain with real film-stock profiles | | ⭐ Star Image Compare | Interactive before/after wipe slider | | ⭐ Star Tiled PiD Upscaler | Upscale with PiD/PixelDiT models in overlapping tiles | | ⭐ Star Tiled SeedVR Upscaler | SeedVR2 tiled upscaling with color correction | | ⭐ Star Lucida RMBG | High-quality background removal (BiRefNet fine-tune) | | ⭐ Star 360 Parallax Viewer | Interactive 360° panorama with mouse parallax | | ⭐ Star 360 Parallax Viewer Pro | Panorama + video export, overlays, depth maps | | ⭐ Star Save Panorama JPEG | JPEG with XMP panorama metadata | | ⭐ Star Save Panorama JPG+ | Panorama export with stereoscopic 3D output | | ⭐ Star Grid Composer | Compose images into a grid with captions, fonts, colors | | ⭐ Star Grid Image Batcher | Batch up to 16 images for the Grid Composer | | ⭐ Star Grid Captions Batcher | Batch up to 16 captions for grid layouts | | ⭐ Star PSD Saver (Dynamic) | Save layered PSD files | | ⭐ Star PSD Saver 2 (Optimized) | Faster PSD export | | ⭐ Star PSD Saver Adv. Layers | PSD export with advanced layer handling | | ⭐ Star Watermark | Text / image watermarks with placement control | | ⭐ Star Meta Injector | Copy PNG metadata from one image to another | | ⭐ Star Icon Exporter | Multi-size PNG/ICO export with shaping & shadow | | ⭐ Star Random Image Loader | Random images from a folder, seed-controlled | | ⭐ Star Image Loader 1by1 | Sequential image loading with state memory | | ⭐ Star Qwen Image Edit Inputs | Stitch up to 4 images for Qwen editing |

🎬 Video — LTX-Video, loops, compression, slideshows

| Node | What it does | |------|--------------| | ⭐ Star LTXV All-in-One (2-Pass) | Half-res pass → 2x upscale → full-res pass in a single node | | ⭐ Star LTX Video Settings | Video dimension & frame calculator | | ⭐ Star VAE LTXV Save / Load | VAE encode/decode for LTX video | | ⭐ Star LTX Image Cut | Smart cropping for LTX video with aspect-ratio preservation | | ⭐ Star LTXV Get Last Frame | Extract the last frame from video latents | | ⭐ Star LTXV Load Last Image From Folder | Load the last generated image from a folder | | ⭐ Star Video Joiner | Join multiple videos into one | | ⭐ Star Video Loader | Decode video to frames + audio + fps — no extra suite needed | | ⭐ Star Video Compressor | Compress to a target size/quality (H.264/H.265/VP9/AV1) | | ⭐ Star Slideshow Maker | Slideshows with transitions, motion effects and audio | | ⭐ Star Image Loop | Seamless looping frames from panoramic images | | ⭐ Star Video Loop | Seamless looping frames from video | | ⭐ Star Frame From Video | Pick first / last / specific frame from a batch |

📝 Text & Data — wildcards, prompts, storage

| Node | What it does | |------|--------------| | ⭐ Star Wildcards Advanced | Wildcards with folder paths, nesting and multi-prompt support | | ⭐ Star Text Inputs (Concatenate) | Merge multiple text inputs | | ⭐ Star Text Filter | Remove words, whitespace and empty lines | | ⭐ Star Easy-Text-Storage | Save & reuse text snippets across workflows | | ⭐ Star Prompt Picker | Pick prompts from a file or folder (random or sequential) | | ⭐ Star Web Scraper (Headlines) 📰 | Scrape news headlines as prompt inspiration |

🎨 Prompts & Conditioning — Qwen, Flux2, regional control

| Node | What it does | |------|--------------| | ⭐ Star Image Edit for Qwen/Kontext | Dynamic prompt builder with customizable templates | | ⭐ Star Qwen-Rebalance-Prompter | Intelligent prompt rebalancing | | ⭐ Star Qwen Edit Encoder | CLIP encoder tuned for Qwen image editing | | ⭐ Star QwenEdit+ Conditioner | Enhanced conditioning for Qwen models | | ⭐ Star Qwen Regional Prompter | Region-based prompting for precise control | | ⭐ Star Flux2 Conditioner | Text + up to 5 reference images for Flux2 | | ⭐ Star Conditioning Saver / Loader | Save and reuse conditioning between workflows |

💾 IO & Save — smarter saving and loading

| Node | What it does | |------|--------------| | ⭐ Star Save Image+ | PNG/JPG/WEBP/PSD with format chips, mask embedding, 16-bit PNG, 5 metadata fields | | ⭐ Star Load Image+ | Load from input/output folder, clipboard paste, metadata fields, invert mask | | ⭐ Star Metadata Saver Option | Labeled key/value metadata for Star Save Image+ | | ⭐ Star Image Loader Options | Show metadata from Star Load Image+ with copy buttons | | ⭐ Star Save Folder String | Flexible path builder with date-based organization |

🛠️ Helpers & Tools — the little things that save your day

| Node | What it does | |------|--------------| | ⭐ Star Show Everything | Universal debug node — connect anything, see type/shape/stats/preview | | ⭐ Star Stop And Go | Interactive pause / preview / continue control | | ⭐ Star Output Cleaner | Browse & clean your output folder with thumbnails | | ⭐ Star Ollama Prompt Helper | Local Ollama prompts with 15 presets and vision support | | ⭐ Star Duplicate Model Finder | SHA256 duplicate model scanner | | ⭐ Star FP8 Converter | Convert checkpoints to FP8 | | ⭐ Star Model Packer | Merge split safetensors, convert precision | | ⭐ Star Divisible Dimension | Keep dimensions divisible by a given value | | ⭐ Star Size Calculator by Side | Resize longest/shortest side, keep aspect ratio | | ⭐ Star Denoise Slider | Simple denoise-strength slider | | ⭐ Star Box Drawer | Draw filled or outlined rectangles | | ⭐ Star Image Shifter | Shift images with seamless wrapping (panoramas/textures) | | ⭐ Star Krea2 Unbound | Prompt-adherence enhancer for Krea2 models | | ⭐ Star Multi Inputs To One | Combine dynamic inputs into a single output | | ⭐ Star Show Last Frame | Extract the last frame from video latents | | ⭐ Star Save / Load / Delete Sampler Settings | Reusable sampler presets | | ⭐ Star SD Upscale Refiner Advanced | SD1.5 upscale + refine with LoRA, tiled diffusion, ControlNet tile | | ⭐ Star Palette Extractor | Extract the dominant color palette from an image |

--- ## 🃏 Wildcards On first start, StarNodes copies a ready-made wildcard collection to `[ComfyUI]/wildcards/`. Add your own `.txt` files there — one option per line. ```text a photo of __animal__ in __place__, {golden hour|blue hour|midnight}, __style__ ``` - `__name__` → random line from `wildcards/name.txt` - `subfolder\__name__` → use subfolders - `{a|b|c}` → quick random choice - Nesting works up to 10 levels deep 🪆 --- ## 🎨 Theme System StarNodes ships its own **color themes**. Pick a theme in the ComfyUI settings, then apply theme presets to any node via the **right-click menu** (multi-select supported). Perfect for keeping big workflows readable. --- ## 📚 Example Workflows Ready-to-use workflows live in **`example_workflows/`** — drag any `.json` onto your canvas. Highlights include: - 🖼️ **Image upscaling & refinement** (SD upscale, SeedVR/PiD tiled upscalers) - 🎬 **LTX-Video** pipelines (2-pass all-in-one, loops, slideshows) - 🌌 **360° panorama** generation & interactive viewing - 🧩 **Qwen image editing** (regional prompting, multi-image edit) --- ## 📦 Dependencies Everything is listed in `requirements.txt` and installs automatically. **Good to know:** if a package is missing, the rest of the pack still loads — only the affected node is skipped, and ComfyUI tells you exactly what to install. | Package | Needed for | |---------|------------| | `psd-tools` | PSD Saver nodes | | `opencv-python` | Detail Enhancement, Panorama JPG+, Slideshow | | `scikit-learn`, `webcolors` | Palette Extractor | | `color-matcher` | Simple Filters (color matching) | | `imageio-ffmpeg` | Video Loader / Compressor / Slideshow | | `requests`, `beautifulsoup4` | News Scraper | | `ollama` | Ollama Prompt Helper | --- ## 🆘 Troubleshooting | Problem | Fix | |---|---| | A node is missing | Check the ComfyUI console — it prints which package to `pip install` | | Nodes don't appear at all | Make sure you restarted ComfyUI after installing | | Wildcards don't resolve | Confirm the files exist in `[ComfyUI]/wildcards/` | | Ollama node can't connect | Start the Ollama app/server first (`ollama serve`) | 📖 Detailed per-node docs are built right into ComfyUI: select a node and open its **help panel** — StarNodes ships a doc page for every node (`web/docs/`). ---

**Made with ⭐ by [Starnodes2024](https://github.com/Starnodes2024/ComfyUI_StarNodes)** If StarNodes helps your workflows, consider leaving a ⭐ on GitHub — it helps others find the pack!

Name	Type	Default	Description
model_path	STRING		—
save_name	STRING	_fp8_scaled_e4m3fn.safetensors	—

⭐ Star FP8 Converter

Inputs (2)

Outputs (1)