The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Back to feedPost
Just wrapped up building flux/, an interactive UI with real-time WebGL rendering, custom GLSL shaders, and scroll-driven animation, built entirely from scratch with vanilla JS, what do you think?
It's been a super fun shift for me, most of my background is Python automation, AI integration, and data pipelines, but diving into frontend/UI work like this has been genuinely rewarding.
Link's in my portfolio if you want to see it in action. Always open to talk automation, AI integration, or interactive web work.
Moving from Python pipelines into custom GLSL is an interesting shift. With scroll driving the experience, I’m curious how you keep shader work and scroll updates coordinated on slower devices. Did you build a lighter rendering mode for mobile?
Honestly it is more of a new interest than a whole shift
And right now the shader/scroll sync is tuned mainly for desktop performance, but it is very optimized that it works nicely on an alright cellphone. I haven't built a dedicated lighter rendering mode for mobile yet,...
The vanilla JS route is the part I would defend - straight to GLSL instead of inheriting a scene graph you never needed. The one that caught me on scroll-driven work was portrait: visible width is 2*z*tan(fov/2)*aspect, so at 0.75 aspect you see less than half of what you...
Yeah only the middle parts of the models show on portrait when using the phone, edges get cut off. Didn't scale the fov/camera distance for aspect ratio, just assumed desktop dimensions. Gonna fix that soon.
And vanilla + GLSL was intentional, didn't want scene graph overhead for...
The network for creativity
Join 1.25M professional creatives like you
Connect with clients, get discovered, and run your business 100% commission-free
Creatives on Contra have earned over $150M and we are just getting started
Related posts
👋 I'm new to Contra and I did my first post without realizing I hadn't even introduced myself!
Hey I'm Katya, I'm an AI Systems Architect, marketer and yoga teacher.
I build AI systems that bring in business and handle the work behind it. A voice agent that calls new leads back within seconds and books the appointment. A pipeline that traced 16,900+ leads to the ads that brought them. Listing videos made from the photos a real estate agent already has.
The marketing side came first. I built the multi-channel campaign for Jay Shetty's launch (22 million followers). After that I discovered the tech world and when AI came along, I realized I'm a builder at heart and never looked back! Building is so much fun 🤍
Here's what I've shared:
My first post: on building a fully AI-made YouTube video, end to end, because I'm camera shy 🙈 https://contra.com/community/MP9mFRuT-im-building-my-first-fully-ai-made
Last thing, if you're building with AI video or automation, I'd like to see what you're working on
And if you're hiring: I may be new to Contra but definitely not new to this work. Reach out for AI agent builds, apps, websites or anything in between.
P.s The video below is my AI avatar. My profile video is the real me :)
Nice Portrait....Has everything from a great pose, great composition and lighting and perfect environment setting.
Credit card landing page design for a fintech startup — an interactive 3D claw machine that turns the offer into the hero instead of another floating card at 15 degrees.
Every card brand ships the same hero: a card hovering over a gradient. Clawd sells one card, not a marketplace, and the entire pitch is "stop comparing — we already did the comparing". So the landing page had to make a single-product offer read as a win rather than a shortage of options, and pull early-access sign-ups from people who are deliberate about money: the ones who actually track spending and read the APR line.
The metaphor does the positioning work. A claw machine that never misses, plus "we already picked the best one", means the visual is the argument and the headline doesn't have to explain itself.
I built the scene in Blender as an interactive element rather than a still render — the claw follows the cursor, the card tilts in the grip, and the pile underneath reacts, so a visitor plays with the product's core idea for a few seconds instead of scrolling past it.
Hero plus full page structure, desktop and mobile designed separately, the Blender scene with web-ready exports, interaction notes, and a developer-ready Figma file.
nice work
