Consistent crops
Apply one crop and composition workflow across a full batch of training images.
LoRA Dataset Builder
Build cleaner LoRA datasets, faster.
Batch crop, resize, rename and pair training images with matching Caption TXT files—without uploading your dataset to a server.
Free · No account · Local processing · Images + Caption TXT in ZIP
Matching Caption TXT
Create one text file per image using your own reusable caption template.
Private by design
Your images stay on your device while SwiftCrop processes and packages the dataset.
A faster LoRA dataset workflow
Go from an unorganized image folder to a consistent export in three steps.
1. Add images
Drop JPEG, PNG or WebP files into SwiftCrop and review the batch.
2. Set dataset options
Choose size, format, filename pattern, Dataset Name and Caption Template.
3. Export the ZIP
Download processed images with optional matching Caption TXT files.
Built for LoRA training
- Batch image resizing and cropping
- Portrait-aware composition adjustment
- Sequential, consistent filenames
- Image and Caption TXT pairs in one ZIP
Flexible output sizes
Square outputs such as 1024×1024, 768×768 and 512×512, plus custom sizes.
SwiftCrop prepares files only; you remain in control of the training configuration and caption content.
Explore the AI Creator Hub
Prepare datasets for other popular image-generation workflows.
LoRA dataset FAQ
Does SwiftCrop generate AI captions?
No. SwiftCrop creates TXT files from the Caption Template you provide. It does not run a captioning model.
Are my LoRA training images uploaded?
No. Image processing and ZIP creation run locally in your browser.
Can I change the output resolution?
Yes. Choose a preset or enter a custom width and height for your workflow.