50 free ComfyUI workflows to automate your image editing

If you spend hours editing images one by one, you’re leaving a lot of time on the table. With the right ComfyUI workflows, you can batch‑process entire folders in a single run: resize, rename, remove backgrounds, restore old photos, build storyboards, generate comics, and even auto‑caption your training sets.

This guide walks through 50 free automation workflows built on ComfyUI and the latest Pixeloma nodes, so you can turn your image pipeline into a mostly hands‑off system.

Before you start: update ComfyUI and Pixeloma

These workflows rely on the newest Pixeloma nodes. If you’re on an older version, some nodes will show up in red.

To check your version, double‑click the ComfyUI canvas and add the Pixeloma version check node. The Pixeloma row shows what you’re running, and this node works even on older builds.

For the latest version, go to pixeloma.com and open the Help page. It always lists the current release at the top, plus step‑by‑step instructions for updating based on how you installed ComfyUI (for example, the Easy Install version). There’s also a “Fix a problem” section if you see red nodes or missing models.

If you’re brand new to ComfyUI, it can help to first read a beginner guide like how to use ComfyUI step by step, then come back and plug in these automations.

How these workflows are organized

All episode 38 workflows live together under a dedicated section in your Pixeloma workflows list. Each one includes three key helper buttons:

Download models – lists every model used, the download links, and the correct folder paths.

How to use – shows a short, written walkthrough plus tips specific to that workflow.

Nodes info – explains what each node does and typical run times. Press Escape to close it.

Node colors are consistent across all workflows:

• Orange: model loaders
• Blue: user controls (usually numbered in the order you should check them)
• Green: processing steps
• Red: saving/export nodes

Most folder‑based workflows follow the same basic pattern:

1. Use Load images from folder and either browse to your folder or paste its path.
2. Click Pick images, choose specific files or Select All, and hit Done.
3. Adjust any blue nodes (prompts, sizes, options).
4. Press Run once. One run processes every selected image in sequence.
5. Find results in ComfyUI’s output folder, usually inside a subfolder named after the workflow.

Essential batch image utilities

1. Resize an entire folder in one run

This workflow resizes every image in a folder without touching your originals. You choose a target size for the longest side (for example, 1024), or open the gear icon to type custom sizes. Optional shape presets can also crop into squares or wide frames. Outputs are saved as JPGs in an output/resized folder, preserving filenames.

2. Create three social media sizes at once

Need square posts, vertical feeds, and stories from the same photos? This workflow takes each image and runs it through three Resize crop nodes (for example: square, tall post, and story). You can type any width/height combo; it crops from the center and never stretches. Each size saves to its own subfolder under output/social media.

3. Add a text watermark to every image

To protect your work, this workflow stamps a text watermark onto every image in a folder. You can set the corner, font size, colors, and opacity, and it’s best to leave the size in percent so it scales consistently across large and small images. Results go to output/watermarked.

4. Batch rename and convert formats

This utility gives every image a clean new base name plus a running number, and lets you convert formats to WebP, JPG, or PNG in one go. Type the base name once, pick the output format, and run. Copies are saved under output/renamed in numbered order.

5. Remove backgrounds and export two versions

Using a small background‑removal model, this workflow cuts out the subject in each image and saves two versions: a transparent PNG and a white‑background JPG (perfect for online shops). If you only need one version, just mute the extra save node. Outputs live in output/no background and output/white background.

6. Build a slideshow video from a folder

Turn a folder of images into a simple slideshow video. You set the duration per image, frames per second, and final video size (for example, a square post). A math node multiplies duration and FPS for you. The finished video is saved in output/slideshow and can be previewed right from the node.

7. Create contact sheets

This workflow arranges all images into a single contact sheet with filenames under each thumbnail. You can pick the number of columns, image size, background color, and optional title. The sheet is exported to output/contact sheets.

8. Upscale and enhance images

Using a compact upscale model, this workflow increases resolution (up to 4×) while sharpening and adding realistic detail. A scale factor lets you set the final size (for example, 0.5 for 2× instead of 4×). Upscaled images keep their original filenames and are saved under output/upscaled.

9. Stamp your logo on every image

Instead of text, this version uses your logo (ideally a PNG with transparency) and places it in a corner of every image. Math nodes calculate position from each image’s own size so the logo spacing looks consistent. Outputs go to output/with logo.

10. Resize by percentage

If you want to scale images proportionally without changing aspect ratio, this workflow resizes every image by a percentage (for example, 50% to halve dimensions). Math nodes compute new width and height per image, and the resized copies go to output/scaled with the new size in the filename.

11. Make a long photo strip

Without any AI models, this utility builds a horizontal photo strip from all images in a folder. It first normalizes each image to the same size, then uses a loop to stitch them side by side into one long image. The final strip is saved under output/photo strip.

Flux Klein: batch edits, styles, and creative tools

Many of the more advanced workflows use Flux Klein (and sometimes Flux "Klein"/"client" in the transcript), a powerful image model used for both generation and editing. Once you’ve downloaded the Flux models via the Download models button, most of these workflows reuse them without extra setup.

12. Apply the same Flux edit to a whole folder

Here, every image in a folder gets the same Flux Klein edit in one pass. You write a single prompt (for example, “turn each portrait into a cute 3D character”), and each image is processed in about three seconds after the first run loads the model. Results appear under output/folder edit.

13. Give each image its own prompt

Instead of one prompt for all, this workflow uses Prompt multi so each image gets a unique prompt. Prompt 1 goes to image 1, prompt 2 to image 2, etc. This is ideal for editing mixed content like mugs, guitars, and sneakers, each going into a different scene. Outputs are stored in output/paired edit.

14. Turn photos into coloring book pages

Using Flux Klein, this workflow converts photos into clean black‑and‑white line art with no shading or color—perfect for printable coloring pages. The prompt is tuned for crisp outlines on white. Kids love pages made from pets, family photos, or toys. The finished pages are saved in output/coloring pages.

15. Outpaint to wider formats

When you need a 16:9 image from a narrow photo, this workflow adds a gray border to the sides and asks Flux to extend the scene. Outpaint stitch then places your original image back in the center pixel‑perfect, surrounded by AI‑generated extensions. Wide results are stored in output/wide.

16. Consistent product photos for e‑commerce

If you sell online, this Flux Klein workflow turns plain product shots into consistent studio scenes. A shared prompt defines the background and podium, and a text watermark node stamps your shop name on each image. An optional notify node alerts you when processing is done. Outputs go to output/product photos.

For a broader view of how AI can streamline online selling, you might also like this guide to automating e‑commerce workflows with AI agents.

17. Restore and colorize old photos

This workflow takes a folder of faded or damaged photos and uses Flux Klein to clean them up and colorize them. You simply paste the folder path, select all images, and run. Each photo comes back sharper, cleaner, and in color, saved under output/restored. If you prefer to keep them black and white, the How to use panel shows exactly which words to change in the prompt.

18. Generate one image per prompt

For simple text‑to‑image runs, this workflow lets you write one prompt per line in Prompt each, choose an image size, and generate a sequence of images—one per prompt. Outputs go to output/prompt each.

19. Create prompt combinations with brackets

Here, you write a single prompt that includes choices in square brackets, split by vertical bars. For example, three colors and two vehicle types generate six combinations. A helper node lists every expanded prompt, and Flux Klein renders each one to output/combinations.

20. One subject, many styles

This workflow lets you type a subject once (for example, “a bunny reading a book”) and then list different styles (watercolor, pixel art, clay, etc.) in Prompt each. Each style is appended to the base subject, and you get a grid of consistent content in varied looks. Results land in output/styles.

21. One picture, many edits

Load a single source image, then write one edit per line in Prompt each (“at the beach”, “in Tokyo at night”, “in a magic forest”, etc.). Flux Klein keeps the character consistent while changing environments or moods. All versions go to output/many edits.

22. Process a big list of prompts

Paste a large batch of prompts into Prompt pack, leaving a blank line between each. The workflow automatically queues one run per prompt, using Flux Klein to draw every image. Outputs are saved under output/prompt pack.

Prompt automation and experimentation

23. Random prompts with curly brackets

In this setup, curly brackets hold choices for each part of the prompt. With Dynamic prompts turned on, every run randomly picks one option from each set, builds a new prompt, and saves it along with the image. Re‑running gives you endless variations. Images go to output/random prompts.

24. Prompt stack with on/off switches

Rather than retyping prompts, Prompt stack keeps each phrase on its own row and lets you enable or disable lines like switches. For example, turning off “in a library” and enabling “in a sunny park” instantly changes the scene. A Show text node reveals the final combined prompt, and the result goes to output/prompt stack.

25. Turn short ideas into full AI prompts

This workflow uses a small local “writer” model plus Flux Klein. You type short ideas (one per line) into Prompt pack. The AI expands each idea into a detailed prompt, saves all generated prompts to a text file, and then Flux Klein draws the images. Everything is stored in output/ideas.

26. Turn a story into a storyboard

Write a short story in the AI prompt box, and the writer model splits it into six scenes, each with its own prompt. A style phrase is appended to all of them, then Flux Klein draws each scene. An Image grid node assembles the scenes into a single storyboard image. Both individual frames and the full board are saved in output/storyboard.

27. Character turnaround sheets

From one character image, this workflow generates a four‑view turnaround: front, 3/4, side, and back. You load your character, and Prompt each holds the descriptions for each view. Flux Klein renders each angle and an image grid combines them into a single sheet under output/turnaround.

28. Loop basics: multiple seeds, same prompt

This loop workflow shows how to create multiple variations of the same prompt. Loop start controls the number of rounds, and math nodes add the loop index to the seed so each round uses a different seed. The images are combined and saved under output/loop basics.

29. Loop build‑up: step‑by‑step edits

Here, each loop round edits the result of the previous one. You load a starting image and list changes in Prompt multi (for example, add a lemon tree, then string lights, then a cat, then a sunset). Loop end feeds each round’s output into the next, and every stage is saved in output/loop build up.

30. Four‑panel comic strips

Using a character image as the base, this loop workflow creates a four‑panel comic. Each row in Prompt multi is one panel’s description (waking up, breakfast, bike ride, sunset goodbye). After each round, Stitch images adds the new panel to the right. The final strip is stored in output/comic strip.

Testing settings and styles with XY plots

31. Compare samplers and steps

The first XY plot workflow creates a 4×3 grid (12 images total) to test how different samplers and step counts affect the result. The X axis cycles through step counts (for example, 1, 2, 4, 8), and the Y axis cycles through samplers. All use the same seed so differences are easy to see. When the grid is complete, you click Save output to export it to output/XY plot.

32. Compare styles and subjects

The second XY plot workflow tests words instead of settings. You put styles across the X axis (for example, “watercolor painting”, “digital 3D”, “pixel art”, etc.) and subjects down the Y axis (fox, lighthouse, robot). Flux Klein draws each combination, and you again manually save the grid when it’s done.

Draft, control panels, and dual‑job workflows

33. Draft first, then upscale the winner

This workflow speeds up experimentation. It first generates a small, fast draft image from your prompt. The run pauses at a Pause image node so you can preview it. You can hit Regenerate to get a new draft with a new random seed. When you find one you like, click Continue and Flux Klein redraws it at full resolution. Drafts and finals are saved under output/draft/finish.

34. A single control panel for key settings

This workflow turns a standard Flux Klein setup into a simple app‑like interface. A Control panel node holds the prompt, image size, steps, and seed, and each row is wired to the relevant nodes behind the scenes. You only tweak values in one place, then press Run. Images go to output/control panel.

35. One workflow, two jobs (text vs. image edit)

Here, Flux Klein loads once but serves two separate jobs: A is pure text‑to‑image, B edits an existing image with an edit prompt. A Switch source node flips all the necessary connections between A and B with a single click. Run with A selected to generate a new image, then switch to B and run again to edit a source image. Results are saved under output/two jobs.

36. Turn steps on or off per image

This modular workflow lets you selectively enable or disable steps like background removal, watermarking, and cropping to story size. Each step has its own preview and save node, and a Mute switch controls whether it runs. Only enabled steps execute, and each step writes results into its own subfolder under output/on off.

37. Style presets via dropdown

Instead of manually typing style phrases, this workflow offers a dropdown of style presets (for example, anime, 3D clay). You write the subject in a text node, pick a style, and a Text join node combines them into a full prompt. Switching styles is as easy as changing the dropdown and running again. Outputs go to output/style presets.

38. Cleaner graphs with set/get nodes

This is a standard Flux Klein workflow that demonstrates using Set and Get nodes to avoid long wires across the canvas. The loaders (model, text encoder, VAE) are stored under names in Set nodes, then retrieved wherever needed via matching Get nodes. Functionally it’s the same as direct wiring, but far easier to read. Images are saved under output/set get.

Working with existing prompts and images

39. Extract prompts from old images

If your images were created in ComfyUI, Automatic1111, Forge, or downloaded from Civitai, they often contain embedded prompts. This workflow uses a Prompt reader node to read those prompts from every image in a folder and save them to a single text file via Save text. The file is stored in output/old prompts.

40. Recreate old images with Flux Klein

This workflow takes those embedded prompts and uses Flux Klein to recreate the images in the same shape. You can optionally add extra words via Text join to tweak all prompts at once. New versions are saved in output/recreated, and an Image compare node lets you overlay old and new.

41. Generate images from text files

Here, each text file in a folder holds one prompt. A Load texts from folder node reads them in name order (you can number them if order matters). Flux Klein creates one image per file, naming each output after its text file. Results are saved under output/from text files and displayed in a grid with filenames.

AI captioning, naming, and sorting

42. Auto‑caption a training set for LoRA

Training a LoRA usually requires a text caption per image. This workflow uses a local vision‑capable model to look at each picture and write a descriptive caption. You can add a trigger word that’s prefixed to every caption. Each caption is saved as a text file with the same base name as its image, and a copy of the image is stored alongside it. Everything goes into output/training set.

43. Name images based on their content

Using the same seeing model, this workflow writes a short, meaningful name for each image and renames the files accordingly. A Text join node creates a mapping list of old name + new name, which is saved to a text file. Renamed images are written under output/named by AI.

44. AI‑generated product scenes

For product shots, this workflow has a local AI look at each image and write an edit prompt tailored to that product. These prompts are saved in text files, then Flux Klein uses them to generate new scene images. Before/after comparisons can be viewed with Image compare, and results are stored in output/AI scenes.

45. Reimagine a folder in a new style

This variation uses the same two models, but you provide the target style (for example, “cute 3D animated character”) in the AI prompt box. The AI rewrites each image’s description into that style, saves the prompts, and Flux Klein redraws the images. Outputs go to output/reimagined, with optional before/after comparisons.

46. One edit prompt, grid of before/after

Here, a single edit prompt (for example, “turn every product into a clay toy”) is applied to the whole folder with Flux Klein. Each result is saved individually under output/clay toys, and an Image grid node builds a large grid showing each original above its edited version, complete with filenames.

47. Yes/no sorting with AI

This workflow asks the AI a simple yes/no question about each image, such as “Is there a person in this picture?”. Based on the answer, a Text match node routes the image to one of two save nodes. The “yes” images go to output/with a person, and the rest to output/without a person.

48. Sort images into category folders

Using the same vision model, this workflow sorts images into subfolders based on a list of categories you provide (for example, animal, food, place). The AI picks one word from your list for each image, and Text join builds a path like “animal/filename”. Save image writes each file into its category folder under output/sorted.

Caption review and re‑rendering from captions

49. Review and fix captions in a dedicated window

Caption review is a new node that opens its own window for cleaning up captions. You paste the folder path, click Review captions, and see each image with its corresponding caption text file. You can filter by images without captions or search for words, then click a caption to edit it (for example, change “yellow t‑shirt” to “blue t‑shirt”). Clicking Save changes updates the text file. This node does not run with the workflow; it’s just an interactive editor.

50. Recreate images from fixed captions

Once captions are corrected, another workflow reads the images and their text files with Load images from folder and Load texts from folder (both pointing to the same path). You add a shared style phrase via Text join (for example, “soft watercolor painting”), and Flux Klein recreates each image from its caption. Results go to output/from captions, and an image grid shows old vs. new—like the t‑shirt now correctly rendered in blue.

Handling large folders in parts

When you’re working with big folders, re‑running the entire set every time is wasteful. One of the workflows is designed to run in parts by skipping images that already have results.

After you set the output folder (for example, winter products), enable Skip pictures already done on the folder node. The first run might process only a subset of images. Subsequent runs will automatically ignore any images that already have matching outputs, stopping immediately when there’s nothing left to do.

Wrap‑up

These 50 ComfyUI workflows turn repetitive image tasks into one‑click automations. Whether you’re preparing assets for an online shop, building datasets for training, or experimenting with Flux Klein for creative projects, you can batch almost everything: resizing, renaming, background removal, restorations, style transfers, captions, and smart sorting.

Once you’ve updated Pixeloma and downloaded the required models, you can mix and match these workflows to build your own end‑to‑end image pipeline—and spend your time on creative decisions instead of manual edits.