Why Image Stabilization Matters
Smartphone camera megapixel counts keep climbing year after year, but what truly determines video quality lies elsewhere: Image Stabilization.
In video recording, frame-to-frame shake accumulates, making footage uncomfortable to watch no matter how high the resolution. For photos, faster shutter speeds can mitigate the effects of hand shake in bright conditions, but in low light, slower shutter speeds make motion blur an unavoidable problem.
There are three main approaches to solving this: OIS (Optical), EIS (Electronic), and HIS (Hybrid), which combines the two. Let's break down the principles, pros and cons, and real-world differences in how each actually works.
OIS — Optical Image Stabilization
How It Works
OIS physically moves the lens or image sensor to counteract hand shake. When the gyro sensor inside the camera module detects movement, an actuator shifts the lens assembly or sensor in the opposite direction to correct the optical axis.
Implementation Methods
OIS is broadly divided into two approaches:
- Lens-shift: Moves a portion of the lens group along the X and Y axes. This is the traditional approach with a long history in the camera industry.
- Sensor-shift: Moves the image sensor itself. Apple introduced this with the iPhone 12 Pro Max and has since expanded it across the entire iPhone lineup. Samsung also adopted sensor-shift OIS with the Galaxy S26 Ultra. Chinese manufacturers like Huawei (Pura 70 Ultra, Pura 80 Ultra) and OPPO have also implemented it in their flagships. Since the sensor is lighter than the lens, it offers faster response times and a wider correction range, and adoption is expanding across the flagship market.
Actuator types include VCM (Voice Coil Motor), SMA (Shape Memory Alloy), and ball-guide mechanisms, selected based on tradeoffs between module size, power consumption, and correction range.
Pros
- Optical correction means no image quality loss
- Stabilizes the light reaching the sensor, making it strong in low-light conditions
- Effective for both still photos and video
Cons
- Additional components (actuators, gyros, driver ICs) increase module cost and size
- Correction range has physical limits (typically around ±1 degree)
- Durability concerns from drops and impacts (OIS lock mechanisms required)
- Additional power consumption
EIS — Electronic Image Stabilization
How It Works
EIS corrects hand shake through software rather than hardware. It calculates inter-frame shake based on gyro sensor data and applies cropping and warping (geometrically transforming and aligning the image) to each frame to produce stabilized video output.
In simple terms, the actual output uses a narrower area than the full sensor capture, and the position of this output region is adjusted frame by frame according to the detected shake.
Pipeline
A typical EIS processing pipeline works as follows:
- Gyro data acquisition — Angular velocity data is read from the IMU (Inertial Measurement Unit)
- Motion estimation — Gyro data is combined with (in some cases) image-based motion vectors to estimate camera movement
- Target path generation — Intentional panning is separated from unintentional shake. A smoothing filter creates a stabilized target path
- Frame transformation — Affine (transformation preserving parallelism, including rotation, scaling, and translation) or homography (transformation that also accounts for perspective) transforms are applied to each frame for correction
- Crop and output — The final frame is cropped to remove black borders caused by the transformation
Pros
- No additional hardware required, favorable for cost reduction
- No impact on module size
- Algorithm improvements possible through software updates
- No physical correction range limits, enabling compensation for large movements (practically effective up to around ±3 degrees)
Cons
- FOV (field of view) loss due to cropping (typically 10–20%)
- Resolution degradation from the crop + warp process
- Difficult to apply to still photos (requires inter-frame comparison)
- Weak at correcting high-frequency vibrations
- Cannot reduce motion blur itself in low light (since it's not optical correction)
- Combined with rolling shutter distortion, correction results can appear unnatural
HIS — Hybrid Image Stabilization
Concept
HIS combines OIS and EIS. The term "HIS" itself is more of a marketing and technical convenience label used by manufacturers rather than an industry standard, but the concept of using OIS and EIS together is adopted by most flagship smartphones today.
Actual Operating Structure: OIS First-Pass Correction → EIS Post-Processing
It's easy to imagine HIS as OIS and EIS equally sharing the workload, but the actual implementation is closer to a serial pipeline.
- OIS corrects first, physically — When the gyro sensor detects shake, the actuator moves the lens or sensor to optically counteract the movement.
- EIS reads the OIS correction results and post-processes — Residual micro-shake, rolling shutter distortion, and motion blur remaining after OIS correction are refined through software to produce smooth final video.
The key point is that EIS knows what OIS corrected and by how much. By passing OIS actuator motion data to the EIS pipeline, over-compensation is avoided and the strengths of each system can be leveraged.
The best example of this structure is Google Pixel 2's (2017) Fused Video Stabilization. Google extracts gyro signals and OIS motion data together to precisely estimate camera movement, then synthesizes frames through a 3-stage pipeline with machine learning-based motion filtering. Sony Xperia's Optical SteadyShot also officially employs a structure that drives OIS and EIS simultaneously.
Angle-Based Role Division
As an approach to OIS-EIS collaboration, dividing roles based on correction angle is also discussed.
- OIS's physical correction limit is approximately ±1 degree
- EIS can provide practical correction up to approximately ±3 degrees
Using this difference, shake within ±1 degree is handled optically by OIS without quality loss, while larger movements exceeding ±1 degree are handled by EIS through software. This minimizes EIS crop amount while securing a wide correction range.
However, the extent to which this approach is implemented in actual commercial products varies by manufacturer, and most do not disclose their detailed internal structures.
Differences by Camera and Mode
One important point to note is that even on the same smartphone, the stabilization method changes depending on the camera and shooting mode.
- Still photos → OIS only (EIS cannot be applied as it requires inter-frame comparison)
- Video recording (main camera) → OIS + EIS combination often operates
- Ultra-wide camera → EIS only (no OIS hardware)
- Telephoto camera → Primarily OIS
Even modes marketed as providing powerful video stabilization, such as Samsung's "Super Steady" and Apple's "Action Mode," actually work by switching to the ultra-wide camera and maximizing EIS crop. Rather than sophisticated OIS+EIS collaboration, they are closer to modes that aggressively apply EIS.
Pros
- Adding EIS software post-processing on top of OIS optical correction produces smoother video than either alone
- Since OIS handles first-pass correction, EIS crop amount can be reduced
- Most flagship smartphones adopt this combination, making it effectively the standard configuration for video stabilization
Cons
- Synchronization between OIS motion data and the EIS pipeline is challenging to implement
- OIS hardware costs are still included
- Behavior varies by camera and mode, making it difficult for users to experience consistency
- Implementation levels vary significantly between manufacturers, and most do not disclose their detailed structures
Comparison Summary
| Attribute | OIS | EIS | HIS |
|---|---|---|---|
| Correction Method | Hardware (Physical) | Software (Digital) | Hardware + Software |
| Image Quality Impact | No loss | FOV loss, resolution degradation | OIS-level maintained + minimized EIS crop |
| Low-Light Performance | Excellent | Limited | Excellent |
| Still Photos | Effective | Ineffective | Same as OIS |
| Video Stabilization | Effective (range limited) | Effective (quality tradeoff) | Most effective |
| Module Cost | High | No additional cost | High |
| Module Size | Increases | No impact | Increases |
| Implementation Difficulty | Medium | Medium | High |
| Correction Range | ±1° (physical limit) | ~±3° | OIS ±1° + extended by EIS |
Future Trends
The Rise of AI-Based EIS
While traditional EIS has been based on gyro data and conventional computer vision, deep learning-based motion estimation and frame synthesis technologies have been advancing rapidly in recent years. Google's Pixel series leads this field, evolving toward reducing crop loss by combining inter-frame interpolation with super-resolution.
Expansion of Sensor-Shift OIS
Sensor-shift was pioneered by Apple, followed by Chinese manufacturers like Huawei and OPPO, and Samsung finally adopted it with the Galaxy S26 Ultra. As major flagship manufacturers adopt it one after another, sensor-shift is becoming the new standard in the flagship market. Sensor-shift increases lens design flexibility, creating strong synergy with lens module miniaturization.
Software-Defined Camera
In the long term, the weight of hardware correction will likely decrease, with software correction combining sensor data and AI processing becoming more important. However, the limits of physics — particularly insufficient light in low-light conditions — cannot be fully overcome by software alone, so OIS is unlikely to disappear entirely.
Conclusion
Image stabilization is not a simple feature but a system design challenge where hardware, software, physics, and cost are complexly intertwined. There is no single "best" answer among OIS, EIS, and HIS. The optimal combination varies depending on product positioning, target price, and primary shooting scenarios.
In the next article, I plan to dive deeper into the internal structure of the EIS pipeline, covering gyro data processing, frame transformation, and smoothing algorithms.
This article was written based on hands-on experience in camera module development. What type of image stabilization does your smartphone use? If you find any errors in this article, please point them out in the comments. Feedback and opinions are always welcome.
Top comments (0)