image

Image Pipeline Builder

Build reusable image-processing flows with nodes and apply the same steps across multiple JPEG, PNG, and WebP images.

Local processing

About your filesFiles and content selected in this tool are processed in your browser and are not uploaded to a Browser Kitty backend.

Operation demo See the basic workflow in a short video.

About this tool

Image Pipeline Builder turns Resize, Crop, Rotate, Flip, Canvas, color adjustment, Grayscale, Blur, Sharpen, Border, Rounded Corners, Text Watermark, and other operations into connected nodes. Add multiple JPEG, PNG, or WebP images, inspect intermediate results with a representative-image Preview, then apply the same Pipeline sequentially across the full batch.

One output port can connect directly to several branches, so shared upstream work can feed outputs such as a main image and a thumbnail without a separate Split node. Frequently used graphs can be saved as browser-local Recipes and run directly with newly selected images, while Pipeline JSON provides a portable graph/settings backup. Results can be saved individually or collected into a ZIP. Selected images and generated outputs are not uploaded to an application server.

Good for

Apply the same resize, crop, adjustment, and watermark sequence to recurring batches of photos
Create a main web image and a thumbnail from the same source through separate branches
Tune node order and settings while inspecting intermediate Preview results
Save recurring image-processing work as a Recipe and reuse it with new image sets
Back up or transfer an image-processing graph as Pipeline JSON for another device or Git repository
Collect several JPEG, PNG, or WebP Outputs into structured ZIP folders

What it can do

Key capabilities available in this tool.

Connect Resize, Crop, Rotate, Flip, Canvas, Adjust, Grayscale, Blur, Sharpen, and other processing nodes
Apply one Pipeline sequentially across multiple JPEG, PNG, and WebP images
Fan one output port into multiple branches and reuse shared upstream work for several Outputs
Inspect Before / After intermediate Preview at the selected node before running the full batch
Save browser-local Recipes and run them directly on new images or apply them back to the Canvas
Export Pipeline JSON with nodes, connections, and settings for portable backup or transfer
Configure JPEG / PNG / WebP, quality, filename templates, and ZIP subfolders per Output
Download files individually or as ZIP; failed images are separated while the remaining batch continues

How to use

Open the tool and add one or more JPEG, PNG, or WebP images through the Images node or image panel.
Add the processing nodes you need, such as Resize, Crop, Rotate, Adjust, or Text Watermark, and connect their ports in execution order.
When one processed image should produce several sizes or formats, connect the same output port directly to multiple branches.
Select a node and inspect Before / After Preview for the representative image while tuning its settings.
For each Output, choose JPEG, PNG, or WebP, quality, filename template, and an optional ZIP subfolder.
Run the Pipeline across all selected images. If one image fails, the remaining batch continues and failed results are shown separately.
Save successful images individually or collect them with Save ZIP.
Register recurring workflows as Recipes, or save Pipeline JSON when you need a portable backup or transfer.

Supported

Input: static JPEG, PNG, and WebP. Animated GIF/WebP, HEIC/HEIF, TIFF, RAW, SVG, and PSD are outside the v1.0.0 input scope.
Processing nodes: Resize, Crop, Rotate, Flip, Canvas, Adjust, Grayscale, Blur, Sharpen, Border, Rounded Corners, Text Watermark, and Output.
Branching: one image output port can connect directly to several downstream branches, with shared upstream work reused during evaluation.
Preview: switch Before / After at the selected node to inspect the representative image; Output also reports the actual encoded size.
Batch: images are processed sequentially to reduce peak memory. A failed image is separated while the remaining batch continues.
Output: JPEG, PNG, or WebP with per-Output quality, filename templates, optional ZIP subfolders, individual downloads, or ZIP export.
Persistence: Recipes keep the graph/settings in the current browser, while Pipeline JSON stores nodes, connections, settings, and layout as a file. Source and generated images are excluded.
Runtime networking: the standalone build uses connect-src 'none'; decode, Preview, transforms, Batch, encoding, Recipe execution, and ZIP generation run in the browser without uploading selected images.

Limitations and notes

v1.0.0 accepts static JPEG, PNG, and WebP and outputs JPEG, PNG, or WebP; AVIF output is not included.
Crop is batch-oriented and uses ratio, size, and anchor settings rather than freehand per-image crop handles.
Text Watermark uses system fonts and does not load external web fonts.
JPEG output flattens transparent pixels against the configured background color.
Outputs are newly encoded in the browser, so original EXIF/GPS metadata is not carried over and embedded ICC profile preservation is not guaranteed.
Very large images, large batches, many branches, or high-resolution Preview/Output work can use substantial device memory. Batch processing is sequential to reduce peak usage.
Cancellation is checked between major processing stages; a synchronous Canvas operation already in progress cannot be interrupted in the middle of that operation.
Recipes and current-Pipeline recovery depend on browser storage and can disappear when site data is cleared. Use Pipeline JSON for portable backup.

Frequently asked questions

Are selected images uploaded to a server?

No. The hosted version downloads the app HTML initially, but decode, Preview, transforms, Batch execution, encoding, Recipe execution, and ZIP generation run in the browser. The standalone build uses connect-src 'none'.

What is the difference between a Recipe and Pipeline JSON?

A Recipe is for quickly reusing a workflow in the same browser. Pipeline JSON is a portable backup for transfer to another environment. Neither stores source or generated images, so input images must be selected again when running the workflow.

Can one source image produce both a main image and a thumbnail?

Yes. One output port can connect directly to multiple branches, allowing shared processing followed by separate Resize and Output nodes. A built-in Main image + thumbnail template is also available.

How is this different from Image Compressor & Converter?

Image Compressor & Converter is optimized for quickly reducing or converting images from use-case and target-size settings. Image Pipeline Builder is for composing ordered, branching multi-step processing such as resize, crop, and watermark and reusing the workflow itself as a Recipe.

Are EXIF or GPS metadata preserved in output images?

No. Original EXIF/GPS metadata is intentionally not copied because outputs are newly encoded in the browser. Embedded ICC profile preservation is also not guaranteed, so the app is not intended for strict color-managed print production.

Offline version and source code

The repository generates a single HTML build with the image-processing runtime and Node Editor Core embedded. Open dist/index.html directly to add JPEG/PNG/WebP images, edit Pipelines, preview, run batches and Recipes, and save ZIP output without a network connection.

View source on GitHub