
An experiment produces events, not only files
In situ TEM and SEM experiments evolve as the sample environment changes. Images, videos, spectra, diffraction patterns, and 4D-STEM datasets are easier to interpret when they remain connected to the conditions present at that moment.
Temperature, pressure, gas composition, liquid flow, voltage, current, tilt, position, timing, sample identity, calibration, and user-defined events may all form part of the scientific record.
The image shows what happened. The experiment record helps explain why.

What is an in-situ experiment record?
An in situ experiment record is a structured account of what happened during an experiment. It may include commanded and measured values, hardware states, timestamps, alarms, notes, identifiers, calibration information, and references to acquired files.
No single source necessarily contains the complete record. Depending on the product and workflow, information may come from Hummingbird Control™, microscope or OEM acquisition software, operator input, and customer-selected data systems.
Information that may form part of the record:

Your data, your workflow
Colibri Platform™ supports an open approach to data management for in situ microscopy. Laboratories can use their experimental data across the software and data environments selected for their research.
Depending on verified product capabilities and project scope, experimental data may be used with OEM microscope software, Python workflows, open-source analysis tools, facility databases, LIMS or ELN systems, institutional repositories, and customer-selected automation or AI environments.
Colibri Platform™ does not replace a laboratory’s LIMS, ELN, repository, or institutional database. Instead, its open data approach supports the use of experimental data within the systems the laboratory already selects and manages.
These workflows can support liquid-cell TEM, gas-cell catalysis, electrical biasing and EBIC, 4D-STEM, tomography, and other in situ microscopy applications. Specific data fields, exports, and integration pathways vary by Hummingbird product and configuration.
This approach gives laboratories the flexibility to build data workflows around their research, existing infrastructure, and future needs.

Data formats and integration pathways
Data fields, export formats, and integration pathways vary by Hummingbird product and configuration. Product pages and technical documentation identify the capabilities supported by each system.
Currently available
Product-specific CSV logs where supported.
Supported or project-specific
Customer- or OEM-specific exports, file references, handoff workflows, and defined software integrations.
Future direction
JSON experiment records, HDF5/Zarr-compatible exports, and PDF experiment summaries.

Relating experimental-state data to microscope data
The appropriate method depends on the systems involved and the required level of coordination.
Manual comparison
Users can compare supported Hummingbird logs with microscope files using acquisition times, operator notes, file names, or other references.
Timestamp and identifier alignment
Where both systems provide usable timestamps, session identifiers, or file references, these fields can help relate experimental-state data to microscope data.
Supported or project-specific integration
For defined configurations, Hummingbird Connect™ may support coordinated data exchange or a project-specific workflow pathway. Support depends on the OEM environment, microscope model, software version, Hummingbird system, available interfaces, and project requirements.

Experimental context for AI workflows
Image-only models may identify patterns without knowing the conditions that produced them. Temperature, pressure, electrical state, position, timing, calibration, alarms, and operator actions can provide important context for customer-selected analysis, automation, and AI environments.
For adaptive experiments, external software may also need to determine what the hardware is doing, whether a command succeeded, and whether the observed response matches the intended state.
Explore AI-Ready TEM & SEM Workflows

Frequently Asked Questions
Hummingbird software (via Colibri Platform™) records a comprehensive set of experimental metadata across multiple domains, including:
- Thermal: temperature, setpoints, ramp rates, dwell times, heater power
- Gas: pressure, flow, gas ratios, purge events, analyzer data
- Liquid: flow state, pump settings, liquid volume, heating state
- Electrical: voltage, current, compliance limits, sweep steps
- Motion: holder/stage position, tilt, manipulator position
- Safety: alarms, limits, interlocks, emergency stops
- Session metadata: timestamps, operator notes, sample ID, holder ID, microscope ID
- Timing + user-defined events
This creates a structured in-situ experiment record capturing the full experimental context behind each dataset.
Yes. Colibri is designed around open, exportable data.
Currently available:
- CSV logs
Planned formats:
- JSON experiment records
- HDF5/Zarr-compatible exports
- PDF experiment summaries
It may also support:
- Custom customer/OEM-specific export formats
The goal is to enable use in external tools, pipelines, and long-term data systems, not to lock data into proprietary software.
Yes. The platform supports multiple levels of alignment:
- Manual alignment
- Post-experiment matching of control logs with image timestamps
- Timestamp-based alignment
- Shared timestamps or identifiers connect datasets
- Synchronized workflows (advanced)
- Integration via APIs or OEM pathways
- Coordinated control + acquisition
This allows users to correlate experimental conditions with image/video data at different levels of precision.
Yes. Colibri is explicitly designed to:
- Integrate with microscope software ecosystems
- Work with OEM pathways and APIs
- Support custom workflows and automation setups
It does not replace microscope software but instead connects with existing systems.
Yes.
Because Colibri:
- Exports open formats (CSV, planned JSON, HDF5/Zarr)
- Emphasizes interoperability and external analysis
It is compatible with:
- Python-based analysis pipelines
- Open-source tools
- AI and data science workflows
Yes.
Colibri experiment records provide:
- Structured, time-resolved telemetry
- Multi-domain metadata aligned with datasets
This supports:
- Feature detection
- Model training
- Adaptive experiments
- AI-driven microscopy workflows

Discuss your data management workflow
Share your software requirements, automation goals, and microscopy workflows to receive recommended control, integration, and data management solutions.