TextUnbox vs Drag Your GAN

TextUnbox runs printed and handwritten OCR, DALL-E image generation, speech-to-text, translation, and background removal from one web portal and a shared REST API. Upload or paste an image in the browser to extract text, generate visuals from a text prompt or voice recording, transcribe WAV audio, or strip a photo background. A single license key unlocks every browser app and API endpoint, with usage counted per successful operation rather than by monthly seat.

Most OCR tools stop at text extraction, and most image generators ignore document scanning entirely. TextUnbox ties both workflows to the same prepaid unit pool, so a team can OCR a batch of receipts and generate marketing art without buying separate subscriptions. The REST API at hello.textunbox.app accepts multipart image uploads and JSON prompts, with OpenAPI specs and Postman collections published on the docs site.

Developers building document pipelines, mobile capture apps, or internal automation can call standardized HTTPS endpoints with an x-textunbox-licensekey header. Casual users get the same capabilities through responsive browser pages, including clipboard paste for quick scans. Free trial keys are available by email without a credit card, though ongoing use requires buying prepaid extraction units through Gumroad.

In the realm of synthesizing visual content to meet users' needs, achieving precise control over pose, shape, expression, and layout of generated objects is essential. Traditional approaches to controlling generative adversarial networks (GANs) have relied on manual annotations during training or prior 3D models, often lacking the flexibility, precision, and versatility required for diverse applications.

In our research, we explore an innovative and relatively uncharted method for GAN control – the ability to "drag" specific image points to precisely reach user-defined target points in an interactive manner (as illustrated in Fig.1). This approach has led to the development of DragGAN, a novel framework comprising two core components:

Feature-Based Motion Supervision: This component guides handle points within the image toward their intended target positions through feature-based motion supervision.

Point Tracking: Leveraging discriminative GAN features, our new point tracking technique continuously localizes the position of handle points.

DragGAN empowers users to deform images with remarkable precision, enabling manipulation of the pose, shape, expression, and layout across diverse categories such as animals, cars, humans, landscapes, and more. These manipulations take place within the learned generative image manifold of a GAN, resulting in realistic outputs, even in complex scenarios like generating occluded content and deforming shapes while adhering to the object's rigidity.

Our comprehensive evaluations, encompassing both qualitative and quantitative comparisons, highlight DragGAN's superiority over existing methods in tasks related to image manipulation and point tracking. Additionally, we demonstrate its capabilities in manipulating real-world images through GAN inversion, showcasing its potential for various practical applications in the realm of visual content synthesis and control.