🔧 TL;DR: Depix is a proof-of-concept tool that attempts to recover plaintext from pixelized screenshots using block-wise matching against De Bruijn-based search images. ✨ KEY FEATURES: • Block-matching against a prepared De Bruijn sequence search image. • Support for gamma-encoded averaging and linear sRGB averaging modes. • Auxiliary box-detection utility to locate pixel blocks. 🚀 USE CASES: • Recovering short, redacted strings from screenshots when pixelization used a linear box filter. • Demonstrating limitations of naive pixelation as a redaction technique. 💡 BENEFITS: • Shows concrete failure modes of simple box-filter pixelation. • Provides a repeatable PoC for research and awareness. Depix is a targeted research tool that highlights how pixelization by a linear box filter can leak recoverable information when paired with a crafted search image. #tool #depixelization #image_processing https://lnkd.in/dw6KkMRQ
Depix: A Tool for Recovering Text from Pixelized Screenshots
This title was summarized by AI from the post below.
More Relevant Posts
-
Excited to share my latest project: LCQuant 0.9 – a perceptual command line color quantizer built for uncompromising visual quality. LCQuant is a small tool that reduces the number of colors in an image (reducing its file size) while minimizing quality loss. It’s designed to preserve contrast and color diversity in logos, photos, and gradients, supports alpha transparency, and even allows palettes beyond 256 colors for impressive file size optimizations. In most tests it beats famous quantizers such as Wu, Neuquant, Dennis Lee 3 and K-Means. This tool comes from my years of experience in design, illustration, and image optimization — and it’s lightweight, fast, and ready for modern workflows. 👉 Learn more and try it here: https://lnkd.in/dv7CRD94 And I'd love to read your feedback! :)
-
-
In this experiment, I generated random grayscale images of various sizes (4×4, 8×8, 16×16, 32×32, and 64×64) and applied the five standard PNG filters: None · Sub · Up · Average · Paeth For each filtered image, I computed the Shannon entropy (H = −Σp·log₂p) — a measure of how much information (or randomness) remains after filtering. Result: While random images show similar entropy across filters, structured or correlated images (like natural photos) show a clear trend: Up and Paeth filters typically yield the lowest entropy, meaning better compression with the same quality. None filter shows the highest entropy — no redundancy reduction at all. This demonstrates how PNG intelligently adapts line-by-line, choosing the filter that minimizes entropy → thus, optimal lossless compression. Entropy isn’t just theory — it’s what makes your images smaller, without losing a pixel.
-
-
𝗥𝗲𝗳𝗶𝗻𝗲𝗱 𝗖𝗵𝗮𝗿𝘁𝘀. 𝗦𝗺𝗮𝗿𝘁𝗲𝗿 𝗜𝗻𝘀𝗶𝗴𝗵𝘁𝘀. IC just got a visual upgrade. From enhanced crossplot filtering with depth tolerance to improved chart key panels sorted stratigraphically, we're making interpretation smoother and more intuitive. ✅ Default dynamic templates now show all interpretations ✅ Zonal coloring and pen styles preserved in Ties Manager ✅ Large image files display with improved clarity ✅ Text wrapping now handles extensive comments with ease 𝗕𝗲𝘁𝘁𝗲𝗿 𝘃𝗶𝘀𝘂𝗮𝗹𝘀. 𝗕𝗲𝘁𝘁𝗲𝗿 𝘄𝗼𝗿𝗸𝗳𝗹𝗼𝘄𝘀. 𝗕𝗲𝘁𝘁𝗲𝗿 𝗱𝗲𝗰𝗶𝘀𝗶𝗼𝗻𝘀. #Geoactive #SubsurfaceIntelligence #GeoscienceSoftware #ReservoirCharacterization #DataVisualization #ICPlatform
-
Interpolate a Video from the First and Last Frames with Veo 3.1 and Nano Banana Google released Veo 3.1 with new features and one of them is using the first and last frames to interpolate a video. When leveraging the image generation of Gemini 2.5 Flash Image model (a.k.a Nano Banana), Veo 3.1 model can use the first and last images to generate a short video. This interpola... https://lnkd.in/e2gWqrRM By Connie Leung
-
We’re excited to announce #Dyad Studio v1.2.0 This update introduces a powerful new Graphical Model Builder, letting you #design and #visualize models using the enhanced Diagram View. Quickly auto-layout components, explore libraries in the new Dyad Components View, and enjoy improved compilation output visibility. Try the latest version and experience faster, more intuitive #modeling. https://lnkd.in/e2vwGriV #JuliaLang #Dyad #SystemSimulation #ModelBasedDesign #ScientificComputing
