Reasoning over images and videos: modular CV pipelines vs end-to-end VLMs



https://redd.it/1pxzrli
@r_opensource

Collaborative project APK editor plus

https://github.com/FabioSilva11/Apk-Editor-Plus/tree/master#-autor

https://redd.it/1pxywac
@r_opensource

Would this be useful for people distributing Python libraries? Looking for honest feedback

Hey folks,

I recently used a Python service that was available via pip. Most of the code was readable, but the core features were obfuscated. The package tracked usage using API keys and would limit functionality once a certain threshold was reached.

Honestly, I didn’t hate this approach. It felt like a reasonable middle ground between open code and sustainable monetization — free/visible parts stay open, and the high-value stuff is paid and usage-limited.

That got me thinking: why isn’t there a simple, standardized way for library authors to do this themselves?

So I started exploring an idea where:

You can distribute a normal Python package
Explicitly mark which functions are just tracked vs. paid
Track usage at the function level
Optionally obfuscate only the paid parts
Handle limits and plans without turning the library into a full hosted SaaS

I put together a small concept page to explain the flow with a Python example:
\[Link in comment\]

I’m not trying to sell anything — just genuinely curious:

Would this be useful if you maintain or distribute libraries?
Does this feel reasonable, or does it cross a line?
How have you handled monetization for code you ship?

Looking for honest feedback (even if the answer is “this is a bad idea”).

https://redd.it/1pxyctg
@r_opensource

Palettd - Open Source Color Palette Generator (Github Readme's, etc) - With Hosted CDN

I worked on a cool little open source project over the weekend (among many projects). I was working on a personal project, and wanted a quick way to display the intended color palette on my Github README file. Looked around, nothing incredibly easy to use -- so I made one (which frankly, is how most of my open source releases go).

This one is actually my first NPM release!

Then, I might've taken it a bit further, and launched a website where you can test it out (Playground), and even an open API where you can generate the images, on the fly, without even downloading from NPM (I mean, unless you want to -- that's cool as well)

https://palettd.com/

Go ahead, test it out. It simply includes a super easy way to generate a color palette via form, or via API - and hosted via CDN (supports both SVG and PNG)

(Btw, the display image you're seeing on LinkedIn, generated from Palettd automatically)


https://github.com/mpge/palettd

https://www.npmjs.com/package/palettd

https://palettd.com/api/palette/6366F1,EC4899,F59E0B,10B981.noscript?t=1766985014442
Example:

https://redd.it/1pyekk4
@r_opensource

Tool for personal project/task management that allows collaborating with one or two other people

I'm an entrepreneur and also active in various initiatives to help out people in need in the community, as well as being a busy parent. As a result, I have lots of tasks fighting for priority in my life, both at the pro and personal levels. I need a way to keep it all organized and track progress. Looking for something that is hopefully free, respects user data privacy, and allows to have at least one collaborator as I'd love to have some projects/tasks shared with my partner so we can see each other's tasks and have shared projects. Any ideas?

https://redd.it/1pyei90
@r_opensource

Which smartphone OS should I use?

I have been looking to move away from my android and have been looking into opensource OS such as GrapheneOS and others.

While LegacyOS supports many phone models, they are still out of budget. I have a VIVO V2027 phone lying around. Which opensource OS can I use with it? Can I install the OS as it is on the phone that current has android on it? Or does it need some other specifications?

Would like some help with this. Thanks

https://redd.it/1pyikxi
@r_opensource

Open-source security tooling: what monetization models stay community-friendly (open-core vs dual license vs services

I’m building an open-source runtime security tool and trying to design a sustainable business model without pulling the rug on indie users.

Constraints that matter to me:

Explanations over “magic scores” (teach devs why something is flagged)
Runs offline/on-device (CPU/edge), so sensitive data doesn’t have to leave the environment

I’m exploring: paid support/training, enterprise packaging (SSO/RBAC/audit/compliance), and/or dual licensing.

Questions for folks who’ve done this well

1. What models have you seen work that don’t “enshittify” the community edition?
2. If you did open-core, what did you keep paid without backlash?
3. If dual-licensing: how did you handle contributors + CLAs and avoid future pain?
4. Any “landmines” you wish you knew early?

(Not linking anything here—happy to share details if someone asks.)

https://redd.it/1pyj128
@r_opensource

Blinter The Linter - A Cross Platform Batch Script Linter
https://github.com/tboy1337/Blinter

https://redd.it/1pyiisx
@r_opensource

SnippetMotion - Turn code into animated videos/GIFs in seconds inspired from carbon.sh

I have been using carbon a lot for the past few years and suddengly though of why cant we do short gifs/videos with simple animated typing with syntax highlighting, so I created SnippetMotion, an open-source tool to generate animated code videos and GIFs for social media, README files, and tutorials.

Key features:
\- Type-in animation with customizable speed
\- mobile and desktop aspect ratios with nice mockups
\- 5+ syntax themes (cyberpunk, ocean, sunset, etc..,
\- Export as MP4 or GIF
\- syntax highlighting
\- per line pause like checkpoints
\- No login required just try it

Built with React + TypeScript, inspired by carbon.now.sh

Use_cases:

\- GitHub README tutorials
\- Twitter/LinkedIn code posts
\- Documentation and blog posts
\- Educational content

Live demo: https://snippetmotion.vercel.app
GitHub: https://github.com/Gururagavendra/snippetmotion

roast it, Open to feedback and contributions!

https://redd.it/1pylhw3
@r_opensource

I built a Python-native rendering framework with JIT CSS and State Management that lets you manipulate templates as Python objects. also usable with django

Hey everyone,

I wanted to share a project I’ve finished working on past 5 days, called Mastodon UI (mui in short).

I'm Django-oriented as one of my backend stacks, I love its backend features like (ORM, Forms, Auth), but I’ve always found the template system to be a bit of a context switch. I’d be writing complex business logic in Python, then have to jump into an HTML file to write Jinja-style loops and conditionals. I missed the type safety and autocomplete, and the OOP capabilities I had in my Python files. so I decided to build a middle ground: A Python-native template rendering framework in Python that can be a meta-framework for Django.


It’s called Mastodon UI, and it’s now live on PyPI (mastodon-ui==1.0.6).


What it does
Instead of writing this in template.html:


raw HTML:
<div class="card">
{% if user.is_authenticated %}
<h1>Welcome, <span>{{ user.username }}</span></h1>
{% else %}
<h1>Ooops your not regestred!</h1>
{% endif %}
</div>


You write this in your views.py (or a components/ file):



from mui import (

div,

h1,

ElementState,

StateProps,

)


def card(request):
# not a view but more of simplified example
props= StateProps(required=True, prop_is_truthy=['username','is_authenticated'])
input_props ={
'username':request.user.username,
'is_authenticated':request.user.is_authenticated,
}


content = {'welcom_noscript':f'Welcome {ElementState("span", d_state="username",props=props).change_state(input_props,input_props).state_placeholder}'}
return div(
(  ElementState('h1',d_state='welcom_noscript', props=props).change_state(content,input_props).state_placeholder or h1('Ooops your not regestred!')),
Class="card",
)

is_authenticated?:
yes:
<div class="card"><h1>Welcome <span>Olman</span></h1></div>
no:
<div class="card"><h1>Ooops your not regestred!</h1></div>

The Cool Engineering Stuff (Under the Hood)
Building this taught me a ton about system architecture. Here are the main features:


JIT CSS Engine: I didn't want to manage a huge CSS file. MUI scans your Python components at runtime, finds the styles you defined in Python, and generates a minified CSS string on the fly. It only ships the CSS you actually use.


"state management" Architecture: I decoupled the Logic (State) from the Rendering (Elements). This means components can "fail safely"—if data is missing for an element, it just doesn't render, rather than breaking the page.


Django Forms Integration: This was the hardest part. I built a bridge (RequestDataTransformer) that takes a standard Django Form class and renders it using MUI elements automatically (handling CSRF, errors, and widgets).


HTMX Native: Since I wanted a SPA-like feel without JS, I built a fluent API for adding hx-* attributes to any element.


Bootstrap support for common classes: ucan (1) add bs5 classes/attributes to an element or create an element with the classes instead.


Aaand more.


Why I'm sharing
I’m a self-taught developer (shoutout to any ALX peers here!), and this is my first major open-source framework. I’d love for you to roast the code, try it out, or just tell me if I’m crazy for trying to replace templates. Also, the software kinda passed 380 + tests.


If you’re a Django dev who hates context-switching, this might be for you.


pip: pip install mastodon-ui


PYPI: https://pypi.org/project/mastodon-ui/


GitHub Repo: https://github.com/MojahiD-0-YouneSS/mastodon-ui

https://redd.it/1pyn3lh
@r_opensource

Behind the Scenes of OSS Vulnerability Response
https://www.utam0k.jp/en/blog/2025/12/29/oss-vuln-response/

https://redd.it/1pym71c
@r_opensource

GPU-accelerated node-based image compositor with Python automation API (MIT License)

I've released PyImageCUDA Studio - an open-source tool for designing image processing pipelines visually and automating batch generation through Python.

Demo videos:

https://github.com/user-attachments/assets/6a0ab3da-d961-4587-a67c-7d290a008017

https://github.com/user-attachments/assets/f5c6a81d-5741-40e0-ad55-86a171a8aaa4

The problem it solves:

Most image processing libraries (Pillow, OpenCV) require writing code for everything - no visual feedback until you run the noscript. Visual tools (Photoshop, GIMP) have limited or no automation APIs. This creates a gap: design complex effects visually OR automate batch processing, but not both.

PyImageCUDA Studio bridges that gap.

How it works:

1. Design your image pipeline in the node editor (40+ nodes: generators, effects, filters, transforms)
2. See results in real-time via GPU-accelerated preview
3. Save as .pics project file
4. Automate batch generation with simple Python API

Example - generating personalized certificates:

from pyimagecuda_studio import LoadProject, set_node_parameter, run

with LoadProject("certificate.pics"):
    for name in ["Alice", "Bob", "Charlie"]:
        set_node_parameter("Text", "text", f"Certificate for {name}")
        run(f"certs/{name}.png")

Technical highlights:

- Built on PyImageCUDA, my custom CUDA library with native kernels for all image operations
- GPU acceleration: 10-350x faster than CPU alternatives on complex operations
- Real-time preview using CUDA-OpenGL interoperability
- Zero-copy GPU-to-GPU display for instant visual feedback
- No CUDA Toolkit installation required (just NVIDIA drivers)
- Cross-platform: Windows and Linux

Use cases:

- Batch thumbnail generation for content creators
- Personalized graphics at scale (certificates, badges, social media posts)
- Video frame processing (apply effects to entire clips)
- Data augmentation for ML datasets
- Motion graphics and procedural art
- Marketing automation (generate thousands of ad variations)

Tech stack:

- Core: PyImageCUDA (my custom CUDA kernels in C++/CUDA)
- GUI: PySide6
- Preview: PyOpenGL
- I/O: PyVips
- License: MIT

Project status:

Beta release (v0.1.0) - core features stable, gathering feedback for v1.0. Contributions welcome!

Requirements:

- Python 3.10+
- NVIDIA GPU (GTX 900 series or newer)
- Windows 10/11 or Linux

Installation:

pip install pyimagecuda-studio

Run:

pics
# or
pyimagecuda-studio

Links:

- GitHub: https://github.com/offerrall/pyimagecuda-studio
- PyPI: https://pypi.org/project/pyimagecuda-studio/
- Core library (PyImageCUDA): https://github.com/offerrall/pyimagecuda
- Documentation: https://offerrall.github.io/pyimagecuda/
- Performance benchmarks: https://offerrall.github.io/pyimagecuda/benchmarks/

Contributing:

The project is in active development and I'm open to:
- Bug reports and feature requests
- Code contributions (GPU optimization, new nodes, UI improvements)
- Use case adaptations for different workflows
- Documentation improvements

All contributions go through GitHub Issues and PRs. I respond quickly to community feedback.

Happy to answer questions about the architecture, CUDA implementation, or potential use cases!

https://redd.it/1pysfef
@r_opensource

Kiorg v1.4.1 - A modern battery included file manager with vim inspired keybind
https://github.com/houqp/kiorg

https://redd.it/1pyrwd7
@r_opensource

WenLang – Open Source (AGPL v3) Web App for Learning Languages Through Texts

Hi everyone,

I’d like to share an open source project I’ve been working on called **WenLang**.

**WenLang** is a web application focused on **learning languages through texts**. The main idea is to provide language learners with a practical environment where they can study real texts, interact with content, and gradually improve comprehension and vocabulary in a natural way.

# ✨ Main features

* Users choose:
* Their **native language**
* The **language they are learning**
* (Optionally) other languages they already have some proficiency in
* Study languages through **written texts**
* **Text-to-Speech (TTS)** support to practice listening and pronunciation
* Users can:
* Comment on texts written by other authors
* Publish their own texts in their **native language** or in a language they have good command of
* Community-driven content and interaction

# 🎯 Project goal

The focus of WenLang is to become a **free, open, and collaborative tool** where language students can find everything they need to study languages through reading, listening, and discussion — all in one place.

The project is released under the **GNU AGPL v3 license**, so it stays open and benefits the community, even when deployed as a web service.

# 🤝 Open source

This is an **open-source project** are very welcome:

* Frontend / backend development
* UX/UI improvements
* Accessibility
* Performance
* Language-learning features
* Documentation
* Ideas and feedback

If you’re interested in **languages**, **education**, or **open-source web development**, I’d love to hear your thoughts or have you contribute.

Thanks for reading, and feel free to ask questions or give feedback!
— *WenLang*



https://redd.it/1pywfxy
@r_opensource

corroded: so unsafe it should be illegal

corroded is a library that removes everything Rust tried to protect you from.

It's so unsafe that at this point it should be a federal crime in any court of law.

But it's still blazingly fast 🗣️🦀🔥

Repo is here.

https://redd.it/1pz0kr0
@r_opensource

1100+ open source cloud projects for learning and resume building


A quick follow up to a previous post about this: https://www.reddit.com/r/Cloud/s/89KNntjVCZ

The open source repository for cloud projects (https://github.com/mzazon/cloud-projects) crossed 1100 (!!) projects recently. AWS, Azure, and GCP all covered. With so many projects, the community contributed suggestions and feedback and being able to search and filter was at the top of the list…

So a couple community members threw together a prototype/beta single page, GitHub pages hosted, no login required, no membership required, all session data stored on your browser page that was just approved and merged into the main branch of the repo: https://cloudprojects.dev

Have a look, give it a star if you like it, open an issue with any suggestions. Hope it is helpful.

Happy holidays to all you cloud professionals and aspiring professionals.

https://redd.it/1pz6gr7
@r_opensource

I got sick of ad-bloated utility sites, so I built an open-source alternative (and it looks good).

Hey everyone,

Like many of you, I use random tools daily (JSON formatters, diff checkers, etc.). I got tired of the top search results being ugly, slow, and ridden with ads/trackers.

So I built VoidCraftr (tools.voidcraftr.com).

It’s a collection of common utilities, but designed to actually be pleasant to use.

Let me know if you wanna know about the tech stack.

The Goal: To build the "Swiss Army Knife" of the web that the community actually owns.

It’s fully open source. I’m looking for contributors who want to add their own weird/niche tools to the suite.

Roast it, love it, or fork it. Let me know what you think.

https://redd.it/1pz829s
@r_opensource

DevConsole - a fun open-source task tracker I originally built for fun

Hi everyone!

I originally started DevConsole as a for-fun side project, without any big plans. I mostly wanted to experiment with a retro terminal-style UI and gamified task management for developers.

Along the way it turned out more interesting and useful than I expected, so I decided to open-source it and share it with anyone who might enjoy or find it helpful.

What it is:

Open-source task management system
Retro / terminal-inspired UI
Designed with developers in mind
A bit playful, a bit cyberpunk

Links:

🔗 Live demo: [https://d371l.github.io/devconsole/](https://d371l.github.io/devconsole/)
💻 GitHub repository: https://github.com/D371L/devconsole

Feedback, ideas, and contributions are very welcome.
Hope someone finds it useful or at least fun 🙂

https://redd.it/1pz5n05
@r_opensource

Email Bulk Attachment Downloader - Open Source App

# What My Project Does:

A powerful desktop application for bulk downloading email attachments from Gmail and Outlook with advanced filtering, auto-renaming, and a modern GUI.

It is desgined to minimize the annoying times, when you are looking to download bulk of invoices or bulk of documents and automate the whole process with just few clicks.

The app is perfect even for non-developers, as i have created a Setup Installer via Inno Setup for quick installation. The GUI is simple and modern.

# Source Code:

TsvetanG2/Email-Attachment-Downloader: A powerful desktop application for bulk downloading email attachments from Gmail and Outlook with advanced filtering, auto-renaming, and a modern GUI

# Features:

Multi-Provider Support - Connect to Gmail or Outlook/Hotmail accounts
Advanced Filtering - Filter emails by sender, subject, and date range
File Type Selection - Choose which attachment types to download (PDF, images, documents, spreadsheets, etc.)
Calendar Date Picker - Easy date selection with built-in calendar widget
Auto-Rename Files - Multiple renaming patterns (date prefix, sender prefix, etc.)
Preview Before Download - Review and select specific emails before downloading
Progress Tracking - Real-time progress bar and detailed activity log
Threaded Downloads - Fast parallel downloads without freezing the UI
Modern Dark UI - Clean, professional interface built with CustomTkinter

# Target Audience

Accountants, HR Department, Bussines Owners and People, that require bulk attachment downloads (Students at some cases, office workers)

# Usage Guide

1. Connect to Your Email



Select your email provider (Gmail or Outlook)
Enter your email address
Enter your App Password
Click Connect

2. Set Up Filters



From: Filter by sender email (e.g., invoices@company.com)
Subject: Filter by keywords in subject (e.g., invoice)
Date Range: Click the date buttons to open calendar picker

3. Select File Types



1. Check/uncheck the file types you want to download:
2. PDF
3. Images (PNG, JPG, GIF, etc.)
4. Documents (DOC, DOCX, TXT, etc.)
5. Spreadsheets (XLS, XLSX, CSV)
6. Presentations (PPT, PPTX)
7. Archives (ZIP, RAR, 7Z)

4. Search Emails



Click Search Emails to find matching emails. The results will show:
Number of emails found
Total attachment count

5. Preview Results (Optional)



Click Preview Results to:
See a list of all matching emails
Select/deselect specific emails
View attachment names for each email

6. Configure Renaming



Choose a rename pattern:

|Pattern|Example Output|
|:-|:-|
||
|Keep Original|invoice.pdf|
|Date + Filename|2024-01-15\_invoice.pdf|
|Sender + Date + Filename|john\_2024-01-15\_invoice.pdf|
|Sender + Filename|john\_invoice.pdf|
|Subject + Filename|Monthly\_Report\_data.xlsx|

7. Download



Set the download location (or use default)
Click Download All Attachments
Watch the progress bar and log

# Installation

Installation steps left in the Github Repo.

You can either set up a local env and run the app, once the requirements are downloaded or you can use the "Download" button in the documentation.

https://redd.it/1pz9ren
@r_opensource

Seemake - CMake project analyzer written in Kotlin

I've started a project called Seamake (which later on I thought hey Seemake makes more sense! but then forgot to change it everywhere).


It's a CMake analyzer, you give it the root project directory and it tries to give you a report on the dependencies and options within that software. Mainly because I have multiple projects that needs package management, building from source and binding Cpp software in Kotlin.


This is a very bad draft but I'm hoping to make this better overtime. You can find it at

https://github.com/thisismeamir/seemake


I would love you to hate on it, comment on it, and discuss this with me so feel free to reach me as well.

https://redd.it/1pzbo7l
@r_opensource

I tried wrap for developers and here is what I learned

I recently tried the evaluator of Github developers and it showed that I am in the top 1% of developers in the world.


I think there is something wrong with that app.


What do you think about that?


https://redd.it/1pzdbe2
@r_opensource