Firsthand vs EgoScale

Short answer

EgoScale demonstrates that downstream policy success scales log-linearly with egocentric hours out to roughly 20,854 hours [VERIFY]. Firsthand targets the other lever — fitness per hour: spec-fit capture with pose, depth and frame-accurate labels, billed per validated hour.

Side by side

Firsthand vs EgoScale, dimension by dimension.

Dimension	Firsthand	EgoScale
Hours	Custom to your spec; 10,585 validated hours across the current catalog	~20,854 hours [VERIFY]
Capture setup	Head + two wrist cameras, active-stereo depth, 3× IMU, hardware-triggered on one PTP clock	Large-scale aggregated egocentric capture [VERIFY]
Streams	Ten synchronized streams incl. depth, IMU, 3D hand & body pose, segmentation, action segments	Primarily RGB at scale [VERIFY]
Sync	Hardware-triggered; median 1.12 ms, 2.0 ms ceiling, drift written to metadata	Not a sub-2 ms hardware-sync guarantee [VERIFY]
Hand pose	3D, handJoints[2][21][3], metric, in the head-camera frame	Not manipulation-grade 3D hand pose throughout [VERIFY]
Depth	Metric depth, 848×480 at 30 Hz, with per-pixel confidence	Not universal metric depth [VERIFY]
Action labels	Frame-accurate action segments (verb, noun, ns timestamps)	Scaling-study labeling [VERIFY]
Failure-case coverage	Transit, occlusion and low-light shipped — not trimmed — with a published reject log	Scale over per-skill curation
License	Perpetual, irrevocable, buyer-owned commercial license	Per the EgoScale release [VERIFY]
Commercial use allowed	Yes	[VERIFY]
Consent documentation	Written release per participant; consent artefacts + data card per batch	[VERIFY]
Delivery formats	RLDS, LeRobot, WebDataset, HDF5, zarr, Rerun .rrd	[VERIFY]
Cost	Per validated hour, quoted per engagement; free sample pack, no form	Per the release [VERIFY]

Third-party figures marked [VERIFY] are confirmed against the sources below before publishing. Firsthand figures are measured on-site.

When EgoScale is the right choice

You want maximum pretraining volume and the scaling curve is your priority.
Research into data-scaling laws for egocentric policies.

When Firsthand is the right choice

You have hit diminishing returns on raw volume and need higher-fitness hours.
You need synchronized pose and depth, not just RGB at scale.
You want to pay only for hours that pass QA against your spec.

FAQ

Common questions.

How much egocentric data do I actually need?

EgoScale reports no saturation out to ~20,854 hours [VERIFY], so more hours keep helping — but fitness dominates in practice. A useful start is 50–100 validated hours against one named skill; see /guides/how-much-egocentric-data-do-you-need.

Is volume or fitness more important?

Both matter, but teams typically discard most bulk footage as unusable for manipulation. Firsthand raises the fitness of every hour so fewer hours train further.

Last reviewed: 2026-09-23

Sources

EgoScale paper [VERIFY]
Firsthand evaluation methodology

Judge it on your own skill spec.

Get the free 40-episode sample pack, or send us your target skills and hours for a scoped quote.

Get the free sample pack →Request a quote