Continuum
What if no step was ever final?
A new approach to scientific image processing.
Not all workflows remember their past.
One commits. The other evolves.
Each new step replaces the previous one. Earlier decisions are locked and cannot be revisited without starting over.
Every operation creates a new state. Change any node, and all dependent results update. Nothing is ever lost or final.
A causal structure, not a pipeline.
Decisions create states. States remain alive.
What Continuum actually does.
Continuum is built around a fully non-destructive processing model. Every operation — background extraction, color calibration, stretching, noise reduction, star handling — is evaluated as a state, not baked into the image.
In practical terms, this means you never “commit” a step. If you change your background model, your color calibration, stretch, star reduction and final image all update automatically. You are always working on the cause, not reprocessing the consequences.
This is fundamentally different from traditional astrophotography software, where processing is linear and fragile: once a step is applied, everything downstream depends on that irreversible decision.
Continuum lets you branch workflows naturally. You can test multiple stretches, color calibrations or background models side by side, compare them in context, and refine earlier decisions without duplicating images or restarting.
The result is not just a cleaner workflow, but a different way of thinking about astrophotography processing: iterative, exploratory, and scientifically honest.
Nothing is ever final. There is only the current state of your signal.
Easy to learn. Embarrassingly powerful.
Most astrophotography software is powerful but difficult to learn. Complexity is pushed onto the user through rigid workflows, fragile processing order, and irreversible steps.
Continuum takes the opposite approach. The interface is intentionally simple, while the underlying model is a full computational graph evaluated continuously.
You don’t need to understand the graph to use Continuum effectively. But the graph is always there — precise, consistent, and scientifically rigorous.
The result is a tool that is fast to learn, difficult to outgrow, and almost uncomfortable in how much control it gives you without asking for effort in return.
Early access will be limited.
Continuum will not be released to everyone at once. The first versions will be available through a gradual, invite-based early access phase.
Joining the waitlist is the only way to receive updates, access invitations, and detailed information as development progresses.
If you’re interested in using Continuum from the beginning, this is the moment to stay in the loop.
It’s called Continuum because nothing in the process is ever discrete, final, or broken into irreversible steps.
In traditional image processing, each operation replaces the previous one. Decisions are applied, committed, and left behind. The workflow moves forward by cutting off alternatives.
Continuum treats image processing as a continuous structure. Every decision remains part of the system. Change one element, and all its consequences update coherently.
There is no “before” and “after”. No frozen history. No final image.
Only a continuous relationship between signal, decisions, and their effects.