Online bike fitting: calculator, video analysis or a fitter?
Choose a bike-fit method by the question it can answer: body-based estimates, recorded movement, remote human review or an in-person service. Compare inputs, outputs, privacy and the checks a screen cannot perform.
Published 22 October 2026 · Updated 23 October 2026 · OpenBikeFit
Different inputs. Different evidence.
- Calculator
- Declared body inputs → Model estimate
- No observation of pedalling.
- Video software
- A recorded view → Movement estimate
- No automatic human review.
- Human review
- Material actually reviewed → Scoped interpretation
- Remote is not a physical inspection.
These are independent evidence types, not three required stages or an accuracy ladder. In-person work can add direct physical observation within the agreed service. No row proves ideal fit, component permission or a health outcome.
On this page
- 01The quick choice: start with the question, not the technology
- 02A bike-fit calculator models your inputs; it does not watch you ride
- 03A frame-size suggestion is not a completed bicycle position
- 04Video software estimates the movement present in its recording
- 05Check whether you can make a usable recording before choosing video
- 06Remote human review is a service scope, not a report badge
- 07An in-person fitting includes work that a browser cannot perform
- 08When numbers disagree, compare the evidence before the subtraction
- 09One clip, an independent repeat and a re-check answer different questions
- 10Keep a hand-off record that another person can interpret
- 11Compare privacy, cost and responsibility before sharing or paying
- 12Take the next step without confusing an example with your result
Inspect before recording
Read a fictional report first
A public generated example, not your result or the interactive demo. No camera, input, saved fit or acceptance is required to read.
Open toolOnline bike fitting can mean entering body dimensions into a calculator, recording movement for software to analyse, sending material to a human reviewer, or booking an appointment online that happens in person. The shared word online describes access, not what was measured. Before comparing results or paying for a service, identify the question, the material supplied and the person or model responsible for the conclusion.
This guide is for adults using educational bicycle-setup tools without an active health or injury concern. It does not assess symptoms, decide whether riding is medically safe, prescribe treatment or approve a mechanical adjustment. Its job is narrower: help you choose an appropriate information workflow and keep its limitations beside the result. Reading needs no account, camera, personal measurement, upload or acceptance of an interactive tool's terms.
The quick choice: start with the question, not the technology
If you want to understand what a formula returns from height and cycling inseam, choose a body-input calculator and inspect its assumptions. If you want a record of the bicycle you already own, measure named physical points and use a worksheet. If you want to inspect movement in a side-view recording, first check whether you can produce a suitable stationary recording. These are separate tasks; you do not have to complete every one to use a useful free tool.
If you want someone to interpret a particular situation, establish whether the service includes an identified human reviewer, what they receive and how questions or follow-up are handled. A tool that generates a report automatically is not thereby a remote consultation. When the question requires physical inspection, a different view or an examination outside an educational app's scope, collecting more numbers from the same screen does not supply that missing work.
- Only body measurements available: read the input protocol, then inspect a calculator estimate. Keep it labelled as calculated.
- A bicycle and measuring tools available: preserve its existing setup with exact references. There is no need to loosen a component just to make a record.
- A suitable stationary recording available: inspect a fictional report and the camera protocol before deciding whether to use the separate recording workflow.
- A question that needs a person or physical inspection: clarify the reviewer's scope and the information they need. Do not treat a green software indicator as a substitute.
A bike-fit calculator models your inputs; it does not watch you ride
A calculator can apply a declared rule consistently to the numbers you enter. Its input might include height, inseam, a riding category and selected component dimensions. The returned millimetres are model outputs, not a tape reading from your installed bicycle. More decimal places do not add observations of your limbs, shoes, pedalling or bicycle. Two people can enter the same available inputs while differing in characteristics that the particular model never receives.
OpenBikeFit's calculator compares named historical references and explicitly labelled app conventions. Its app-specific approximation is not a measured knee angle: a knee angle is not an input to that calculation. The source audit explains the actual rule, source type and reading limits. A study about measured joint angles does not automatically validate every coefficient or other result item. Use the existing methodology to inspect those details rather than assuming every number carries the same evidence.
Begin with your own measured inputs and deliberate units. Cycling inseam is not a trouser label, and a current saddle coordinate is not a body measurement. The separate body-measurement guide owns that protocol. If you lack an optional observation, leave it unknown instead of copying a plausible example. Changing a preference can change a model's output without establishing that your bicycle or movement changed. Preserve the input choices if you later compare two calculations.
A frame-size suggestion is not a completed bicycle position
A nominal frame size, a frame's stack and reach, a bar-clamp coordinate and a hood coordinate end at different places. A body-based estimate does not establish a complete build, usable adjustment envelope or component compatibility. Even a correctly identified frame chart cannot tell you whether the actual receiving parts, steerer, post, saddle rails or controls permit the configuration you have in mind. Those questions need their own exact evidence.
If you are comparing bicycles, keep the manufacturer's actual model, year, size and geometry state with each chart. Use the existing frame-reference guides to identify the endpoints, then the separate comparison workspace only for a task its model supports. Do not move the saddle to make a frame label agree with a formula. A size chart, calculator band or successful arithmetic comparison is not a reason to order an unseen assembly or assume it will reproduce a particular position.
Video software estimates the movement present in its recording
A recording adds observations of a visible rider and bicycle under specific conditions. Software may estimate landmarks, angles, cycle events and variation. That is a different evidence source from an anthropometric formula. It is still conditional on the view, tracking and analysis method. A photograph is one instant; a clip contains time, but the existence of video does not establish usable cycles, independent repeatability or physical accuracy.
OpenBikeFit's side-view workflow uses projected two-dimensional estimates. It is not a calibrated three-dimensional scan of every contact point or a physical component inspection. A side view cannot reliably settle a question about an unseen side, hidden interface or another movement plane merely by reporting a confident-looking angle. A missing or rejected observation is unavailable, not a measured zero. Keep the original quality and recording-condition information when interpreting any reported change.
Provider workflows also differ. MyVeloFit's public process describes recording and uploading a riding video and separately describes an expert-review service. That is evidence of how that provider presents those processes, not independent proof of accuracy or a description of every online tool. OpenBikeFit's public analysis example is instead a generated sequence you can read without uploading a rider record. A service name or an AI label alone does not tell you the recipient, processing location or reviewer.
Check whether you can make a usable recording before choosing video
For the existing camera workflow, the bicycle must be supported securely for the intended stationary recording and the complete moving rider must remain visible. Identify the actual trainer and bicycle reference, keep the camera fixed, use a true side view and follow the stated capture protocol. Do not balance precariously against a wall or improvise a mounting arrangement to satisfy an app. If suitable support or framing is unavailable, the non-camera reading and worksheet paths remain useful.
Check device readiness separately from recording quality. Permission, camera settings and a loaded pose runtime do not prove that the particular rider clip is suitable. Likewise, a good-looking preview does not establish independent agreement between takes. Read the existing camera setup, angle convention and repeatability guides for those tasks; this comparison does not replace their protocols. Never add an undocumented frame correction or edit the view simply to make an output move toward a preferred band.
Remote human review is a service scope, not a report badge
A human reviewer can ask questions, examine supplied material and explain why an observation matters within the service they actually provide. They remain dependent on the completeness, identity and quality of that material. Establish whether the reviewer sees the original recording, only selected images, an automatically generated summary or a questionnaire. These are not equivalent inputs, and the word reviewed should identify an actual person and an actual review, not your own opening of a file.
Before using a remote service, ask what the deliverable contains, who reviews it, whether follow-up questions are included, how conflicting or unusable material is handled and what cannot be assessed remotely. MyVeloFit explicitly presents expert review separately from its software analysis. That distinction illustrates a question to ask; it is not a ranking, endorsement or promise about another provider. OpenBikeFit's free automated outputs do not include a personal human fitting consultation or a scheduled response from a fitter.
An in-person fitting includes work that a browser cannot perform
An in-person service can include direct observation, discussion, physical measurement and hands-on work, depending on the practitioner and agreed scope. Do not assume every appointment includes the same equipment, checks or qualifications. Retül's own process description, for example, separates a fitter's initial assessment, marker-based motion capture and a final map of the bicycle. Its description supports that distinction between measurement and interpretation; its marketing outcomes are not guarantees adopted by this guide.
Ask in advance which bicycle, clothing, shoes, pedals and previous records are relevant, and what the final report defines. Also establish whether mechanical inspection, actual adjustment and any follow-up are included or require another competent service. Paying for a motion-capture label does not by itself answer those questions. An online booking page is merely a booking channel; it does not turn a physical appointment into a remote or automated fit. An educational tool is not a clinical assessment.
When numbers disagree, compare the evidence before the subtraction
Consider three independent fictional records: a calculated saddle estimate of 740 mm, an installed bicycle reading of 733 mm, and a projected knee interior angle of 145 degrees. These are not observations of one real rider. The first is a model output, the second needs a named saddle point and BB datum, and the third needs a landmark convention and a recording event. They do not form a vote in which the most confident-looking number wins.
The numerical difference 740 minus 733 is 7 mm, but it is not an instruction to raise a saddle by 7 mm. You have not established that the two lengths use the same endpoint, bicycle state or intended meaning, or that the estimate is a desired installed coordinate. Separately, 180 minus 145 equals 35 degrees of flexion under that planar convention. That conversion changes the label, not the recording, accuracy or bicycle. Do not combine these examples into a body-size pairing, target or universal correction.
Write a small evidence ledger before comparing reports: quantity, source, raw value, units, origin and endpoint, condition and method. Mark the output as calculated, manually observed or video-estimated. If the references do not match, retain both records and state why a difference is withheld. If they do match, a subtraction still describes the records; it does not prove the cause, significance or benefit of a physical change. Preserve disagreements instead of deleting inconvenient observations.
One clip, an independent repeat and a re-check answer different questions
Several cycles within one clip describe variation inside that recording. Another take of the unchanged setup adds an independent recording attempt under the declared conditions. A later recording after a controlled change is a different comparison again. Treating repeated frames as independent experiments, or calling a second attempt an after measurement when nothing changed, confuses the evidence before any analysis begins.
Keep the same named configuration, view and relevant recording conditions with the material. If unchanged takes disagree, resolve the observation problem rather than selecting the pair that produces a preferred result. OpenBikeFit's existing repeated-take and before/after gates remain separate from this article. A successful local save confirms persistence of the record, not its physical truth. A reported difference beyond the software's observed-noise comparison is not a clinical, handling or performance outcome.
Keep a hand-off record that another person can interpret
A useful hand-off identifies the question, the actual equipment and condition, the method used and the unresolved observations. Preserve the report's units, angle convention and named bicycle endpoints. Ask a receiving reviewer what material is necessary before sending it; do not attach personal video or a completed customer record to an unrelated general inquiry. A screenshot without its source context may omit precisely the information needed to explain a disagreement.
For a non-camera start, the free blank bicycle worksheet can preserve your own technical reference questions on paper. Its fixed measurement fields are not dedicated video-review or consultation fields; use supplementary notes rather than relabelling another quantity. The optional local passport and archive workflows have their own access, contents and limitations. A session-only rescue JSON excludes separate landmark and raw-video files, so it is not a complete recording backup or a professional report to import elsewhere.
Compare privacy, cost and responsibility before sharing or paying
Free and local describe different things. A free tool can send data, while a browser-based workflow can still keep a sensitive copy on a device or in a synced download folder. Check what is entered, recorded, uploaded or retained, who receives it, and how you can remove or export it. A video, measurement sheet or result link may identify a person. Do not call it anonymous merely because it lacks a name field or because processing is performed locally.
All public OpenBikeFit tools remain free. Reading this guide creates no fit record or booking. Deliberately opening another page makes normal web requests; an interactive workflow has its existing adult/access and data notices. The guide links carry no measurements or record IDs. External provider links have their own policies and terms. Check current prices and actual inclusions directly before purchasing any service; no universal price or free-trial entitlement is promised here. Custom Balance IT business work is separately agreed and is not a rider fitting consultation.
Take the next step without confusing an example with your result
If you have not chosen a method, inspect the public fictional analysis first. It lets you see the kinds of quality checks and repeated observations involved without recording or creating a rider result. If your question is about body-based calculation, read the input method and open the separate calculator with your own measurements when ready. If your task is preserving the existing bicycle, start with the blank worksheet and exact references instead. None of those reading links performs a calculation, starts a camera or accepts a gate for you.
Keep the conclusion proportional to the evidence you actually collected. A formula was evaluated, a bicycle point was measured, a clip was analysed, or an identified person reviewed particular material: say which happened. A complete educational workflow still does not establish ideal fit, component permission, health, safety or a guaranteed riding outcome. When the necessary observation or expertise is missing, the useful next result is a clearly framed question for the appropriate competent person, not a more authoritative-looking number.
Practical questions
Frequently asked questions
Can I do an online bike fit with only height and inseam?
You can inspect a body-based calculator estimate if its required inputs are available. Those inputs do not record your pedalling, installed bicycle or physical compatibility. Keep the output labelled as calculated, inspect the actual method and do not treat it as a completed fitting or an adjustment instruction.
Do I need a camera or trainer to use OpenBikeFit?
Not for the calculator, public fictional analysis, reading guides or blank worksheet. The separate recording workflow needs suitable stationary support, framing and device readiness. Reading or opening a guide never grants camera permission or accepts the interactive workflow's access notices.
Is an AI bike-fit report the same as a human fitter reviewing me?
No. An automated report is produced by software. Human review is a separate service whose reviewer, source material, deliverable and follow-up should be identified. A report badge, account or payment does not establish that a person reviewed the original material.
Which is more accurate: a calculator or video analysis?
They may estimate different quantities, so there is no universal numerical winner. Compare the intended quantity, assumptions, recording conditions and independent validation for the actual method. A precise-looking value, tracked point or repeated result does not alone prove physical accuracy or a correct position.
Can an online calculator choose the exact frame size I should buy?
A body-based range does not establish an actual frame model, completed contact coordinates, component limits or a permitted build. Identify the manufacturer's chart and exact endpoints separately, then obtain any necessary physical or professional checks. The guide does not choose a bicycle or approve a purchase.
Why does a video angle disagree with my calculator result?
A projected angle and a calculated length are not the same observation. First identify the input choices, units, saddle reference, recording view, angle convention and event. Preserve the disagreement; do not average different quantities or move a component to force agreement.
Does a free or local-first result mean my measurements are anonymous?
No. A result link, video or technical record can identify a person, and local files or browser history may be synced. Read the actual privacy and export boundaries. OpenBikeFit's session-only rescue file excludes separate pose and raw-video files and is not a complete recording backup.
Does OpenBikeFit provide a personal human fitting consultation?
Its public tools are free educational software, not an included human fitting appointment, medical assessment or personal review service. A separately contracted Balance IT business implementation is not a rider consultation. Ask a prospective fitter directly about their scope, qualifications, deliverable and follow-up.