🤖 EZ Auto Renamer – ComfyUI Workflow
Created by Me – a filmmaker & AI workflow designer. This clever tool automatically renames your images using AI! Just drop a folder of images into ComfyUI, and it analyses each one with Qwen 3.5, generates a five‑word descriptive filename (e.g., mountain_road_cave_foliage_hills), and saves them with the new names. Perfect for organising large datasets, training sets, or just keeping your creative assets tidy – all without manual typing.
---
### 🧩 Overview
This workflow loads all images from a folder inside ComfyUI's input directory, sends each one to a Qwen 3.5 vision model, and asks it to output exactly five lowercase words describing the image, joined by underscores. It then checks the length of the generated name – if it's under 128 characters, it's used; otherwise, a fallback error message is applied. Finally, the images are saved with their new AI‑generated filenames in your output folder.
---
### 1️⃣ Required Models & Prerequisites
Make sure these files are placed in your ComfyUI models/ folders:
| Folder | Required Files |
|--------|----------------|
| text_encoders/ | qwen3.5_4b_bf16.safetensors (or qwen3vl_4b_fp8_scaled.safetensors) |
> 💡 That's it! Only one model is needed – no diffusion models, VAEs, or LoRAs. The workflow uses only the text encoder for image captioning and renaming.
---
### 3️⃣ How It Works (AI‑Powered Batch Renaming)
1. Load images – the LoadImageDataSetFromFolder node reads all images from a folder inside ComfyUI/input/ (e.g., a folder named Rename).
2. Analyse each image – each image is passed to the TextGenerate node with Qwen 3.5, which is prompted:
> "Output exactly five lowercase words describing the image, joined by underscores. Do NOT add any other text, punctuation, line breaks, or explanation. Only the five words."
3. Clean and validate – the output is cleaned (removing any unwanted tags like </think> and newlines), and its length is checked. If it's 128 characters or less, it's accepted; if longer (e.g., if the model produced extra text), it falls back to ERROR Renaming.
4. Save with new name – the image is saved with the generated filename (e.g., mountain_road_cave_foliage_hills.png) inside your output folder.
---
### 6️⃣ Advanced Settings (Optional)
- Seed – set a fixed seed for reproducible name generation (if you want the same description for the same image each time).
- Folder Name – change the folder inside ComfyUI/input/ that the workflow loads images from (default: Rename).
- Prompt – you can customise the text prompt to get different styles or more/fewer words (e.g., "Output exactly 3 words...").
- Length Check – the default limit is 128 characters; you can adjust this in the IFMatchCond node if you want longer or shorter filenames.
---
### 📦 Final Output
After running, all images from your input folder will be saved in your output folder with descriptive, AI‑generated filenames – perfect for organising datasets, making files searchable, or preparing training material without manual labelling.
---
Why I built this: As a filmmaker and VFX supervisor, I often work with hundreds of reference images, concept art, or frame grabs. Manually renaming them is tedious and time‑consuming. This workflow does it in seconds, giving me meaningful, consistent filenames that make my projects much easier to manage. I hope it helps you keep your creative files organised too! 🗂️✨
---
📬 Contact: If you’d like to donate, ask a question, or order a custom workflow, reach out to me on Civitai (SaeedKhalili), visit my website SaeedKhalili.com, or email me at [email protected]. I’m always happy to help fellow creators!
Description
🏷️ EZ Auto Renamer – ComfyUI Workflow
Created by Me – a filmmaker & AI workflow designer. This clever tool automatically renames your images using AI! Just point it to a folder, and it analyses each image with a vision model, generates a descriptive filename, and renames or moves the files – all in one click. Perfect for organising large datasets, training sets, or just keeping your creative assets tidy.
---
### 🧩 Overview
This workflow uses my custom ImageRenamer node to:
- Load all images from a source folder.
- Analyse each image with a CLIP‑based vision model (e.g., Qwen 3.5, Gemma, or any compatible encoder).
- Generate a filename based on your custom prompt (e.g., "Output exactly five lowercase words describing the image, joined by underscores").
- Optionally rename the files in place or move them to a new folder with the new names.
All logic is packed into a single custom node – no complex node spaghetti!
---
### 1️⃣ Required Models & Prerequisites
Make sure these files are placed in your ComfyUI models/text_encoders/ folder:
| Folder | Required Files |
|--------|----------------|
| text_encoders/ | Any CLIP‑based vision model (e.g., qwen3.5_4b_bf16.safetensors, qwen3vl_4b_fp8_scaled.safetensors, gemma4_e4b_it_fp8_scaled.safetensors, etc.) |
> ⚠️ Installation Required: This workflow uses my custom ImageRenamer node. You must install it first:
> ```bash
> git clone https://github.com/saeedkhalili/ComfyUI-Saeed-Nodes
> ```
> Place the cloned folder inside your ComfyUI custom_nodes/ directory and restart ComfyUI.
---
### 3️⃣ How It Works (AI‑Powered Batch Renaming)
1. Set source folder – enter the path to the folder containing your images (e.g., C:/Users/.../my_images/).
2. Choose your model – pick the CLIP model you have from the dropdown (e.g., qwen3.5_4b_bf16).
3. Define your prompt – customise what the AI should output (default: "Output exactly five lowercase words describing the image, joined by underscores").
4. Set max length – limit the generated filename length (default: 64 characters).
5. Choose operation – either rename (changes filenames in place) or move (renames and moves files to a new folder).
6. Run – the node processes all images, generates names, and renames/moves them accordingly.
A summary string is output, showing which files were processed and their new names.
---
### 6️⃣ Advanced Settings (Optional)
- Source Folder – the folder where your images are stored.
- Clip Name – select your CLIP model from the dropdown (auto‑populated from your text_encoders/ folder).
- Clip Type – choose the correct type for your model (e.g., qwen_image, stable_diffusion, lumina2).
- Prompt – change the instruction to get different naming styles (e.g., "Output exactly 3 words in Title Case").
- Max Length – set a character limit to keep filenames short and readable.
- Seed – fix the seed for reproducible name generation.
- Output Prefix – add a prefix to all new filenames (e.g., img_).
- Operation – choose rename (modifies files in place) or move (renames and moves to a new folder – you’ll be prompted for the destination).
---
### 📦 Final Output
After running, all images in your source folder will be renamed (or moved) with descriptive, AI‑generated filenames – perfect for organising datasets, making files searchable, or preparing training material without manual labelling. The node also outputs a summary string listing the changes.
---
Why I built this: As a filmmaker and VFX supervisor, I often work with hundreds of reference images, concept art, or frame grabs. Manually renaming them is tedious and time‑consuming. This workflow does it in seconds, giving me meaningful, consistent filenames that make my projects much easier to manage. And because it works with any CLIP‑based model, you can use whichever text encoder you already have. I hope it helps you keep your creative files organised too! 🗂️✨
---
📬 Contact: If you’d like to donate, ask a question, or order a custom workflow, reach out to me on Civitai (SaeedKhalili), visit my website SaeedKhalili.com, or email me at [email protected]. I’m always happy to help fellow creators!