Both PPG and rPPG measure the same thing: tiny changes in blood volume that happen with every heartbeat. The difference is how they detect those changes. One touches you. The other doesn't.
That distinction sounds simple, but it changes everything about where and how vital signs can be measured. A pulse oximeter clipped to your finger works well in a hospital room. It doesn't work so well for screening a thousand people walking through an airport, or monitoring a sleeping infant without waking them, or checking a burn patient's heart rate when you can't attach anything to their skin.
Understanding what separates these two technologies — and where each one actually performs — matters if you're evaluating health monitoring systems, reading clinical research, or just trying to figure out what your smartwatch is doing when it flashes green light at your wrist.
"Remote photoplethysmography enables the contactless measurement of cardiovascular parameters by detecting blood volume pulse signals from video recordings of exposed skin." — Pirzada et al., IEEE Sensors Journal (2024)
How PPG works
Photoplethysmography has been around since the 1930s, though the modern version became widespread with the pulse oximeter in the 1980s. The setup is straightforward: an LED shines light into the skin, and a photodetector on the other side (transmission mode) or the same side (reflectance mode) measures how much light comes back.
Blood absorbs light. When the heart beats, arteries expand with incoming blood, and more light gets absorbed. Between beats, arteries contract, and less light is absorbed. That oscillating signal — the PPG waveform — tracks your pulse in real time.
The wavelength of light matters. Pulse oximeters use two wavelengths, red (around 660 nm) and infrared (around 940 nm), because oxygenated and deoxygenated hemoglobin absorb these wavelengths differently. By comparing absorption at both wavelengths, the device calculates oxygen saturation. Smartwatches typically use green light (around 530 nm) for heart rate, because green penetrates just deep enough for wrist-based reflectance readings and has a strong absorption contrast with blood (Tamura, 2019).
PPG works. It's been validated across decades of clinical use. The trade-off is contact: the sensor has to sit on your skin, pressing close enough to get a clean optical signal.
How rPPG works
Remote photoplethysmography detects the same blood volume changes, but from a distance. Instead of pressing an LED against your skin, rPPG uses a camera to record video of your face (or another exposed skin area) and extracts the pulse signal from pixel-level color changes that are invisible to the naked eye.
Every heartbeat pushes blood into the capillaries near the skin surface. That blood absorbs and reflects light slightly differently than the surrounding tissue. The changes are tiny — fractions of a percent in pixel intensity — but they're there, and signal processing algorithms can pull them out of video.
The first peer-reviewed demonstration came from Verkruysse et al. in 2008, who showed that heart rate could be extracted from standard webcam video under ambient light. Since then, the field has moved from basic color-channel analysis (early methods simply tracked the green channel over time) to sophisticated deep learning approaches that handle motion, lighting changes, and skin tone variation.
Where PPG uses its own light source (the LED), rPPG typically relies on ambient light — whatever illumination is already in the room. Some systems use controlled near-infrared illumination for consistency, but the core appeal is that a regular camera in ordinary lighting conditions can work.
The physics: same principle, different path
Both technologies exploit the same optical phenomenon: light interacts differently with tissue depending on how much blood is present. The Beer-Lambert law describes how light attenuates as it passes through an absorbing medium, and both PPG and rPPG are, at a basic level, measuring variations in that attenuation over time.
The differences are practical, not theoretical:
Signal strength. PPG sensors sit millimeters from the tissue and use focused LED light. The signal-to-noise ratio is high. rPPG cameras sit anywhere from 30 centimeters to several meters away, collecting reflected ambient light across thousands of pixels. The raw signal is much weaker, and the algorithms have to work harder to extract it.
Motion sensitivity. Move your finger while wearing a pulse oximeter, and the reading gets noisy. Move your head during an rPPG measurement, and the problem is worse — the camera has to track facial landmarks and compensate for pixel displacement on top of the motion artifacts in the blood volume signal.
Wavelength control. PPG devices choose their wavelengths precisely. Pulse oximeters use red and infrared for SpO2 calculation. rPPG systems working with standard RGB cameras are limited to the wavelengths captured by the camera sensor, which weren't designed for physiological measurement. This is a fundamental constraint for rPPG-based SpO2 estimation.
How they compare across vital signs
| Feature | PPG (contact) | rPPG (contactless) |
|---|---|---|
| Sensor | LED + photodetector on skin | RGB or NIR camera at distance |
| Light source | Device LED (red, IR, green) | Ambient light or controlled illumination |
| Contact required | Yes — fingertip, wrist, earlobe | No — camera-based, typically face |
| Heart rate accuracy | Gold standard (±1-2 bpm) | Near clinical grade in controlled settings (MAE ~2.96 bpm per 2025 study of 562 participants) |
| SpO2 measurement | Clinically validated, FDA-cleared | Experimental — 93.4% accuracy reported but not yet cleared for clinical use |
| Blood pressure | Emerging (PTT-based methods) | Emerging — one 2025 study achieved 7.52% mean absolute percentage error for diastolic BP |
| Respiratory rate | Derived from PPG waveform modulation | Derived from facial/chest video analysis — accuracy varies (84-95% reported) |
| Motion tolerance | Moderate — affected by movement | Lower — head motion and facial expressions introduce artifacts |
| Skin tone sensitivity | Minimal for SpO2 at clinical wavelengths | Higher — melanin affects signal; active area of algorithmic improvement |
| Ambient light dependency | None — uses own light source | High — performance varies with illumination |
| Multi-person capability | One device per person | Can potentially monitor multiple people simultaneously |
| Cost per measurement point | Hardware needed for each person | One camera can serve many; software-defined |
| Typical measurement time | Continuous real-time | 15-60 seconds of video |
| Regulatory status | Widely cleared (FDA, CE) | Limited — some CE-marked apps; no broad FDA clearance |
| Common form factors | Pulse oximeters, smartwatches, hospital monitors | Smartphones, webcams, surveillance cameras, kiosks |
Sources: Tamura (2019), Pirzada et al. (2024), PMC study of 562 participants (2025), JMIR Formative Research (2025).
Where each technology makes sense
PPG isn't going away. For continuous bedside monitoring in hospitals, wearable fitness tracking, and any situation where a device is already on the body, contact-based PPG is proven and reliable. Pulse oximeters are cheap, well-understood, and FDA-cleared. Smartwatches with PPG sensors have become a mainstream health tool, tracking heart rate and rhythm for millions of people daily.
rPPG fills gaps that PPG can't. Some situations where contactless measurement has clear advantages:
Burn and wound care. Patients with burns or skin conditions may not have intact skin available for sensor attachment. A camera pointed at an unaffected area can still capture vital signs.
Neonatal and pediatric monitoring. Babies and young children often can't tolerate adhesive sensors. Contactless monitoring avoids the distress and skin damage associated with prolonged sensor contact on fragile skin.
Screening at scale. Airports, office buildings, and public health checkpoints need to assess many people quickly. Contact-based measurement requires individual devices and time. A camera system can potentially screen multiple people passing through a space.
Telemedicine. During a video consultation, the clinician can see you but can't measure your pulse through the screen — unless rPPG is running on your camera feed. The patient's own smartphone or laptop becomes the measurement device.
Mental health and behavioral research. Researchers studying stress, cognitive load, or emotional response can capture physiological data from participants without attaching sensors that might alter the very states being studied.
The accuracy question
For heart rate, rPPG in controlled conditions is approaching PPG-level accuracy. A 2025 clinical validation study enrolling 562 participants reported a mean absolute error of 2.96 bpm and heart rate accuracy of 99.1% when compared to reference devices. Shoushan et al. demonstrated 99.4% correlation between rPPG-derived heart rate from smartphone and laptop cameras and ECG reference measurements.
Blood pressure is harder. A 2025 study published in JMIR Formative Research developed an rPPG model that performed well for diastolic blood pressure prediction, with a mean difference of just 0.16 mmHg (SD 3.22 mmHg) against reference measurements in a preoperative clinic setting. Systolic blood pressure prediction was less consistent. Another smartphone-based study reported 93.94% accuracy for systolic and 92.95% for diastolic blood pressure.
SpO2 remains the biggest gap. Pulse oximeters use precisely chosen wavelengths to differentiate oxygenated and deoxygenated hemoglobin. Standard RGB cameras don't capture those exact wavelengths, so rPPG-based SpO2 estimation relies on indirect inference. The 562-participant study reported 93.4% SpO2 accuracy, but noted that no participants with SpO2 below 95% were present — meaning the algorithm wasn't tested in the clinical range where SpO2 measurement matters most.
These numbers come from controlled environments with cooperative subjects and good lighting. Real-world accuracy drops with movement, poor lighting, distance from camera, and compressed video. Closing that gap between lab and life is where most current rPPG research is focused.
Where rPPG is heading
The trajectory is toward ubiquity. Every smartphone, laptop, tablet, and security camera already has the hardware rPPG needs. The barrier is software maturity and regulatory acceptance, not hardware deployment.
Several trends are pushing the field forward. Deep learning models trained on larger, more diverse datasets are improving accuracy across skin tones and lighting conditions. Edge computing allows rPPG processing to happen on-device rather than in the cloud, addressing privacy concerns. And the pandemic-era spike in telemedicine created real demand for remote vital sign measurement that hasn't gone away.
The regulatory path is still forming. A few CE-marked rPPG applications exist in Europe, and the FDA has been engaging with camera-based measurement companies, but there's no standardized validation framework yet. That will come — the clinical evidence base is growing — but it means rPPG sits in an intermediate space: technically capable of a lot, but not yet cleared for most clinical decision-making.
PPG and rPPG aren't really competitors. They're the same measurement principle applied at different scales. PPG is the stethoscope on the chest. rPPG is the thermal camera scanning the crowd. One gives you depth and precision up close. The other gives you reach. Both have a role, and as rPPG matures, those roles are becoming more clearly defined.
Frequently asked questions
What is the main difference between PPG and rPPG?
PPG requires physical contact — a sensor with an LED and photodetector placed on the skin (typically a fingertip or wrist). rPPG works at a distance using a standard camera and ambient or controlled light to detect the same blood volume changes through subtle skin color shifts. The underlying physics are similar, but the measurement method is fundamentally different.
How accurate is rPPG compared to traditional PPG?
For heart rate, rPPG has reached clinical-grade accuracy in controlled settings. A 2025 study of 562 participants found a mean absolute error of 2.96 bpm and 99.1% accuracy compared to reference devices. Blood pressure and SpO2 measurements via rPPG are less mature, with accuracy varying by study and conditions.
Can rPPG replace pulse oximeters?
Not yet for clinical use. Pulse oximeters remain the gold standard for SpO2 measurement, and rPPG-based oxygen saturation estimation is still being validated. For heart rate monitoring in non-clinical settings, rPPG is approaching the point where it could serve as a viable alternative.
What do you need to use rPPG?
A camera — that's essentially it. Most rPPG systems work with standard RGB cameras, including smartphone front-facing cameras and laptop webcams. Good lighting helps accuracy, and the subject needs to be relatively still during measurement, typically for 15 to 60 seconds.
Does skin tone affect rPPG accuracy?
It can. Melanin absorbs light differently, which affects the strength of the rPPG signal. Earlier algorithms showed reduced accuracy on darker skin tones, though recent research has focused on building more diverse training datasets and developing algorithms that perform consistently across the Fitzpatrick skin tone scale.