How will the flower bloom in 2100?
We collaborated with artist Annelie Berner to create a generative web tool that visualizes how the Circaea Alpina flower might evolve in response to future climate conditions.
I've finally put together a v1 of full case study including info about data, algorithms, behind the scenes and whole look dev process. Video content to follow!
#variablenature
@EhrlichMartin Maybe a different question:
2025 - editable ai file with named layers
2026 - here is js blob, prompt your agent to edit, gl
Or have you found a way to have tool internals equally organised?
@EhrlichMartin@Aptos@ashfallstudio Love how clean the design is. How do you do those animations? Yes we can prompt these days and we something but is eg html canvas/no-webgl/webgpu performant enough these days?
Well I do not have proper documentation but I do have 20k screenshots and putting them on timeline works be easy. The thing is most of my work is commercial and briefs and brands get involved. Even within one project linear timeline is not enough to show the process which is DAG graph.
Concept render from installation we delivered earlier this year. I'm working on full documentation and we have so much content i'll have to ration my posts here.
As we are working with @dmnsgn on transitioning #PEX to WebGPU I'm documenting our last big project made with WebGL. So much good stuff here i'll need to split it into several case studies.,,
I wonder if canvas is a hard requirement t or just best way to quickly access multiple versions of project and multiple projects. Zoom out, pan, zoom in. I'm toying with the idea in @nodes_io (one project graph multiple experiments) and you quickly realise that assets and dependencies and resources in general become a problem. Files update, libraries upgrade, data changes. Full clone on each fork and all living in the same live canvas and RAM are hard to imagine or at least wasteful. Maybe you would be able to freeze unused parts of canvas to free resources or there would be some last recently used cache. The alternative SOTA (multiple projects, vscode sessions, LLM chats) and constant window switching and git juggling is crazy though. I do wonder if I could run ALL my code of all projects (100k+ nodes) all at once would it be come kind of god mode or just a mess. It resembles a bit digital garden, second brain, obsessive note linking where in the end it was one big maintenance job and benefits never materialised.
@bnj@contraben@contralabs What's your take on slop detector promoting their product with slop copy of @Orkhan visions. I had pleasure to work on this and I'm confused. Yes nobody owns blob detection but src painting style, outlines, labels... obvious to naked eye, yet not to the agents it seems.
The AI industry has a benchmarking problem.
We're letting random people on the internet decide which models are best
even for capabilities like design...
So we built something different.
Introducing The Frontier Design Benchmark from @contralabs
https://t.co/Eiw47GSxEk
@simonfarussell It's easy to feel productive and still procrastinate and avoid hard things. No tears, no sweat, most work, feels disposable. I have some bigger ambition ideas but they require discipline.
https://t.co/Pne9P8XGXB
Things I did today with agents
- wrote LIDAR map rendering from data that agent, found, downloaded, parsed and enchanted with POI
- sped up our website build times 10x
- decoupled client project from Sanity io CMS so it runs offline with static data
- updated our pex-renderer engine so it support multiple views and shader effects graphs inside RenderGraph per frame so I can share resources
Each on if those would take 1d by hand and yet I feel like nothing has happened.
@thespite But like explain the lighting to 10yo? LLM speech too hard 4 me.
"Light also scatters through the mass: from the shadow map, the thickness between a grain and the first grain the light meets gives how much gets through, averaged over a small disc so it spreads sideways."
Our project for Dentally is not just a design tool but full Brand Algorithm - a recipe for various assets creation. Data -> Grid -> Blur -> Distort -> Refraction
Grid is 12 calendar months based on top level stats from client data. Distort is driven by 3d mesh and boids animation. Refraction is a stack of lenses. All working in tandem to give unique results and full control.
New case study: In collaboration with FORM Brands studio we designed a data-driven brand algorithm manifesting Dentally's drive for effortless efficiency and stress-free dentistry.
Our algorithms transform statistics into fluid gradients providing a dynamic base for the whole brand and assets across different platforms from print to video.
Check out the full case study for more info
https://t.co/eyBJJnTET1
New case study: In collaboration with FORM Brands studio we designed a data-driven brand algorithm manifesting Dentally's drive for effortless efficiency and stress-free dentistry.
Our algorithms transform statistics into fluid gradients providing a dynamic base for the whole brand and assets across different platforms from print to video.
Check out the full case study for more info
https://t.co/eyBJJnTET1