01
The vendor console
A strong fit when the answer is “one instrument, one supported path, and a support contract behind it.” If that is your entire problem, it can be a good answer.
The comparison
Where the current generation stops.
We are not comparing products by name — you know which one you use, and you can place it in the columns below in about four seconds. What matters is not which logo is on the window. It is which stages of the session the software was ever designed to hold.
First, the fair part
Each of these is very good at something.
A comparison table where the other side scores zero is marketing, not information. So here is what each approach earns before we get to the matrix — and those wins are marked in the table too.
02
The scripting workstation
Offers unusually deep instrument control and has enabled many important methods in the field. Nothing here is a substitute for that community.
03
The node-graph pipeline
Got the architecture right first — distributed nodes, a real database, on-the-fly processing — years before anyone else took it seriously.
04
The stack you assembled
Fits your lab exactly, because you built it to. The problem was never the fit. The problem is that its maintainer is also a scientist with their own deadlines.
The matrix
Twenty-six capabilities, five approaches.
Read down the closest archetype. These editorial ratings describe broad product surfaces, not tested versions or every configuration. Confirm current capabilities with the relevant supplier and validate CryoMaestro in your own environment.
Which are you?
Product surfacePossible with extra workNot typical
| Capability | Vendor consoleShips with the column | Scripting workstationMacros, one machine | Node-graph pipelineDistributed, earlier decade | Assembled in-houseBest-of-breed plus glue | CryoMaestroOne record, end to end |
|---|---|---|---|---|---|
| Before the grid is loadedThe stages that happen in a freezer, a shipping dewar, and a calendar — and that almost no acquisition software models at all. | |||||
| Shipment, dewar, puck and vial inventoryUsually a spreadsheet | Not typical | Not typical | Not typical | Possible with extra work | Product surface |
| Chain of custody for each grid | Not typical | Not typical | Not typical | Possible with extra work | Product surface |
| Projects, teams, customers, ownership | Not typical | Not typical | Possible with extra work | Possible with extra work | Product surface |
| Instrument booking and scheduling | Not typical | Not typical | Not typical | Possible with extra work | Product surface |
| On the columnThis is where the current generation is genuinely strong. The differences are less about whether automation exists and more about whether the plan behind it is inspectable. | |||||
| Atlas → squares → square maps → holes → template | Product surface | Product surface | Product surface | Possible with extra work | Product surface |
| Deep low-level instrument and camera controlScripting is the benchmark here | Possible with extra work | Product surface | Product surface | Possible with extra work | Product surface |
| Mixed fleet from more than one manufacturer | Not typical | Product surface | Product surface | Possible with extra work | Product surface |
| Autofocus, astigmatism, coma, drift stabilization | Product surface | Product surface | Product surface | Possible with extra work | Product surface |
| Calibration state shown as a preflight answer | Possible with extra work | Possible with extra work | Possible with extra work | Not typical | Product surface |
| Acquisition plan compiled, versioned and snapshotted | Not typical | Not typical | Possible with extra work | Not typical | Product surface |
| Per-target provenance — why this hole was rejected | Possible with extra work | Possible with extra work | Possible with extra work | Not typical | Product surface |
| Queue editable mid-run without restarting the plan | Possible with extra work | Possible with extra work | Possible with extra work | Not typical | Product surface |
| While the session runsThe difference between a quality metric and a quality signal is whether anything downstream is allowed to act on it. | |||||
| On-the-fly motion correction and CTF estimation | Possible with extra work | Possible with extra work | Product surface | Possible with extra work | Product surface |
| Ice thickness and contamination scoring | Possible with extra work | Possible with extra work | Product surface | Possible with extra work | Product surface |
| Quality gates that pause, skip or reroute the queue | Not typical | Possible with extra work | Possible with extra work | Not typical | Product surface |
| Live run dashboard tied back to atlas coordinates | Possible with extra work | Not typical | Possible with extra work | Not typical | Product surface |
| After the frames landWhere the acquisition-only tools hand off to a shared drive, and where the processing-only tools have already lost the session context. | |||||
| Motion, CTF, picking and 2D at session scale | Not typical | Not typical | Product surface | Possible with extra work | Product surface |
| Result-specific views, not raw output files | Not typical | Not typical | Possible with extra work | Possible with extra work | Product surface |
| Distributed GPU node fleet with health visibility | Not typical | Not typical | Possible with extra work | Possible with extra work | Product surface |
| Results linked back to the exact target they came from | Not typical | Not typical | Possible with extra work | Not typical | Product surface |
| Extending and operating itThe long-run question: what happens when your method changes, your fleet grows, or the person who set it up leaves. | |||||
| Add a reviewed method through a stable contractTyped extension contract → consistent UI | Not typical | Possible with extra work | Not typical | Possible with extra work | Product surface |
| Protocols stored as versioned, reviewable objects | Not typical | Not typical | Possible with extra work | Not typical | Product surface |
| Browser-based, multi-user, usable remotely | Possible with extra work | Not typical | Not typical | Possible with extra work | Product surface |
| Facility analytics — throughput, KPIs, GPU, ice trends | Not typical | Not typical | Possible with extra work | Not typical | Product surface |
| Can run on your own hardwarePrivacy still depends on configuration and integrations | Product surface | Product surface | Product surface | Product surface | Product surface |
| Documented extension and interoperability surface | Not typical | Possible with extra work | Product surface | Possible with extra work | Product surface |
Method Rows describe intended product surfaces, not verified performance, fitness, availability, or a complete market survey. “Partial” means it may be reached with a script, add-on, or person. Features and integrations can change. See the Terms.
What the columns show
The gaps are not random. They cluster.
Every approach above is dense in exactly one band and thin everywhere else. That is not a failure of engineering — it is the scope each one was built for. It only becomes your problem because a real session crosses all five bands whether the software does or not.
Acquisition tools end at the disk
They are dense on the column and empty on either side. Everything before the grid loads and everything after the frames save becomes someone’s manual job.
Processing tools start too late
By the time data reaches them, the session context — which square, which hole, which template, which operator decision — has already been flattened into filenames.
The seam is where the loss happens
Re-entered metadata, broken parent links, decisions nobody recorded, and a quality signal that arrives too late to change anything.
The CryoMaestro column
It is dense all the way down for one structural reason: sample logistics, the column, the compute, and the facility around them share a single data model. Nothing has to be re-entered at a boundary, because there is no boundary.
Get started
Check the matrix against your own session.
Open the bundled demo, follow one grid from import to selected 2D classes, and see which rows your current setup actually covers.