RaceLabs · Performance

Up to 7,000
images per minute.

Analyse motorsport event photographs at up to 7,000 images per minute in the cloud. RaceLabs sends lightweight previews for analysis while your full-resolution originals stay on your disk, ready for editing and delivery.

Cloud photo analysis rate; throughput varies with workload and service availability. Preparation, uploads, review, editing and publication have their own timings. See the complete workflow below.

PREVIEWS FOR ANALYSISORIGINALS ON YOUR DISKREVIEW BEFORE DELIVERY
10,000
photos in the
illustrative calculation
2.5 GB
assumed preview payload
0.25 MB per photo
0.5%
of a 500 GB payload
in this example
0
full-resolution originals
required for analysis
01 · Know where the time goes

From files
to finished galleries.

Cloud analysis is one part of the job. Time these stages on your first event to build a delivery schedule you can use.

RaceLabs event workflow and timing dependencies
StageWhat happensWhat affects the time
PrepareBack up the shoot and prepare analysis previewsComputer, file count, size and format
Upload previewsSend lightweight files for analysisPayload and sustained upload speed
AnalyseAnalyse photos at up to 7,000 images per minute and group the event by rider or vehicleWorkload and service availability; preparation and transfer are separate
Review and editCorrect assignments and prepare the files you want to sellUncertain matches, split groups and your editing choices
Upload and publishTransfer finished gallery files, publish and check deliveryFull-resolution photo/video payload, connection and checks

A processing rate does not measure the time to a reviewed, edited, published gallery. See the complete RaceLabs workflow.

02 · The upload calculation

Smaller previews.
A smaller transfer.

An illustration, not a measured product benchmark: 10,000 photographs at 50 MB per original make 500 GB. If each analysis preview is 0.25 MB, the previews total 2.5 GB. Actual file and preview sizes vary.

Transfer-only arithmetic using decimal MB/GB and an ideal sustained 200 Mbps upload
Assumed payloadSizeCalculated transfer time
10,000 originals × 50 MB500 GB20,000 s = 5 h 33 min 20 s
10,000 previews × 0.25 MB2.5 GB100 s = 1 min 40 s

2.5 ÷ 500 = 0.005, or 0.5%: the preview payload is 1/200 of the original payload in this example. Transfer seconds = GB × 8,000 ÷ Mbps. These calculations exclude protocol overhead, retries and every other workflow stage; they are not a comparison with another product.

Your connection still matters

At a sustained 2 Mbps, the same 2.5 GB of previews takes 10,000 seconds, or 2 h 46 min 40 s, before overhead. Check the upload available where you will work, especially at a circuit.

Allow for RAW and video

Preparation depends on your source files and computer. Video also needs analysis, clip review and final transfer. A calculation for still-photo previews cannot establish video delivery time.

03 · Prepare work you can sell

Your photographs.
Your delivery standard.

Review the whole collection

Inspect the rider or vehicle groups and the Unclear bin. Correct wrong assignments, merge split groups where appropriate and check the final set. For shared cars or karts, confirm driver labels with session or stint information.

Follow the trackday photo grouping and review steps on a complete motorcycle event.

Keep your editing workflow

Save ordinary local folders and finish your edits. The Lightroom Classic export plugin uploads finished JPEGs into the event. Uploading and publishing are separate steps, with time to check the gallery before customers use it.

For a branded photo and video store, allow time to transfer the finished full-resolution files as well as the sorting previews.

Understand the benchmark

The June 2026 internal rider benchmark reports retrieval mAP 0.991, B³ precision 0.996 and B³ recall 0.863. These measure ranking and grouping within that sample, not delivery time or a universal accuracy rate.

Judge previews by the results

The published benchmark does not include a controlled comparison of compression levels. Check difficult frames and the correction work on your own event before judging whether the workflow fits.

The TrackDayPhoto customer story shows published rider galleries and photo/video packages. These illustrate the customer experience after publication; store listings are not timing measurements.

04 · Questions

Speed, answered.

How fast does RaceLabs analyse motorsport photos?
RaceLabs supports cloud photo analysis at up to 7,000 images per minute. Throughput varies with workload and service availability. This rate covers the analysis stage; preparation, uploads, review, editing and publication are timed separately.
Do I upload my full-resolution originals for sorting?
RaceLabs prepares lightweight previews for analysis while your full-resolution originals stay on your disk. Uploading finished gallery files for publication is a separate stage.
Does my computer or connection affect the time?
Yes. Your computer and file formats affect preview preparation, and your sustained upload speed affects transfer time. Cloud analysis is another stage. Smaller uploads reduce transfer work; they do not remove these dependencies.
What does the 10,000-photo upload example measure?
It is an illustrative transfer calculation, not a measured event. Assuming 50 MB per original and 0.25 MB per preview, the payloads are 500 GB and 2.5 GB. At an ideal sustained 200 Mbps, transfer alone takes 20,000 seconds and 100 seconds respectively, before overhead.
What changes when I shoot RAW or video?
File format and size affect preparation and upload. Measure your own RAW/JPEG files on your computer. Include video duration, format, clip review and final transfer separately; the photo example does not predict video turnaround.
Does the benchmark prove that compression has no effect?
No. The June 2026 internal rider benchmark reports retrieval and grouping results. It does not publish a controlled comparison of preview compression levels or establish identical outcomes for every source file.
How much review should I allow?
Inspect the groups as well as the Unclear bin. Correct wrong assignments, merge split groups where appropriate and check the finished collection. Review time depends on the event; a benchmark precision score is not a fixed customer error rate.
How should I compare complete delivery time?
Use the same complete event and record preparation, upload, analysis, review and corrections, editing, finished-file upload, publication and delivery checks. Keep each stage separate and include any waiting or retries. Try your first event free before setting a customer deadline.

Time your own event.
Set your own standard.

Your first event is free. Bring a complete shoot and record the work from preparation through review, editing and publication. See what RaceLabs adds to your day, then use the photo-sorting rates and volume examples to plan your costs.

Speed and matching quality answer different questions. Read the published rider-matching research alongside your own processing and review measurements.