The Sony LYTIA LYT-901 (model name: LYTIA 901, internally referred to as IMX09E during development) is Sony Semiconductor Solutions’ first 200-megapixel mobile image sensor under the LYTIA brand. Announced on November 27, 2025, it is a 1/1.12-type stacked CMOS sensor designed specifically for flagship smartphones. Mass-production shipments began in November 2025, positioning it for 2026 premium devices. Its primary goal is to deliver exceptionally high resolution and sensitivity while enabling high-quality digital zoom (up to 4x) from a single lens—potentially reducing the need for complex multi-camera arrays in future phones.

This sensor stands out as the world’s largest 200MP mobile sensor at launch (1/1.12-inch format with a 14.287 mm diagonal). It combines advanced pixel architecture, on-sensor AI processing, and premium HDR technologies to balance detail, low-light performance, and dynamic range—features that address real-world photography challenges like zoom cropping, mixed lighting, and video recording.

The Sony LYTIA LYT-901 (also referred to as LYT-901 or internally as IMX09E during development) is a stacked CMOS image sensor announced by Sony Semiconductor Solutions, with mass production and shipments starting in November 2025. As of now, it remains Sony’s flagship entry into the high-resolution mobile sensor category, featuring approximately 200 effective megapixels on a relatively large 1/1.12-type format. This makes it the largest 200MP mobile image sensor available at its launch, optimized for premium flagship smartphones launching in early 2026 (such as the Vivo X300 Ultra, which has officially adopted it as its primary main camera).

Below is a detailed breakdown of its core technical specifications, compiled directly from Sony’s official announcements and consistent reporting across reliable sources. These specs highlight its focus on balancing ultra-high resolution with practical usability in low light, high dynamic range, zoom performance, and video capabilities.

Basic Sensor Architecture

• Optical format / Image size: 1/1.12-type (diagonal approximately 14.287 mm) This is a large mobile sensor format—close to (but slightly smaller than) true 1-inch sensors—allowing significantly more total light capture than typical flagship sensors (e.g., 1/1.3-inch or smaller used in many 200MP competitors). The larger area contributes to better overall sensitivity and reduced noise despite the tiny individual pixels.
• Effective resolution: Approximately 200 megapixels This native high pixel count enables detailed stills and aggressive cropping for digital zoom without severe quality loss.
• Pixel size (unit cell size): 0.7 μm × 0.7 μm (horizontal × vertical) Standard for modern 200MP sensors; small pixels are compensated by advanced binning and light-gathering improvements in pixel structure and color filter design.
• Color filter array / Pixel arrangement: Quad-Quad Bayer Coding (QQBC) This proprietary 4×4 grid clusters 16 adjacent same-color pixels, enabling analog 16-in-1 binning for high-sensitivity modes while supporting on-sensor remosaicing (via integrated AI circuitry) to recover full-resolution detail for zoom or cropping.
• Sensor type: Stacked CMOS (two-layer design separating photodiodes from logic/processing circuits) Improves speed, power efficiency, and enables on-chip AI processing for features like remosaicing and advanced HDR.

Readout and Frame Rate Modes

The sensor supports multiple output resolutions via intelligent binning, with all-pixel autofocus available across modes:

• Full resolution: ~200 MP (4:3 aspect ratio) at up to 10 fps (Full RAW readout) Ideal for maximum detail in static scenes or heavy cropping.
• Binned modes:
  • ~50 MP (2×2 binning, 4:3) at up to 30 fps Balances detail and performance for everyday photography.
  • ~12.5 MP (4×4 binning / 16-in-1, 4:3) at up to 60 fps (including burst shooting) Default mode for superior low-light sensitivity and faster processing.
• Video-specific high-speed modes:
  • 8K (16:9) at 30 fps (via 2×2 binning)
  • 4K (16:9) at 120 fps (via 4×4 binning) These enable cinematic slow-motion and high-resolution video capture.
• Burst shooting examples: 60 fps at 12.5 MP, 30 fps at 50 MP, 10 fps at 200 MP.

Dynamic Range and HDR Technologies

• Dynamic range: Over 100 dB (equivalent to approximately 17 stops) in certain modes Achieved through a combination of:
  • Dual Conversion Gain HDR (DCG-HDR): Reads each frame at two different gain levels and composites them for smooth tonal gradients in a single exposure.
  • Fine 12-bit ADC (Analog-to-Digital Converter): Provides finer quantization (12-bit depth) than typical 10-bit mobile sensors, preserving subtle color and gradient details across the full tonal range—even at zoomed-in resolutions.
  • Hybrid Frame-HDR (HF-HDR): Merges short-exposure frames with DCG data (processed on-device), suppressing blown highlights and lifted shadows in high-contrast scenes.
• These features collectively deliver natural-looking results in challenging lighting (e.g., bright skies with detailed shadows), rivaling or exceeding dedicated camera sensors in some scenarios.

Interface and Power Specifications

• Output interfaces:
  • MIPI C-PHY: 2 or 3 trios, maximum 6.0 Gsps per trio
  • MIPI D-PHY: 2 or 4 lanes, maximum 2.5 Gbps per lane These high-speed serial interfaces ensure fast data transfer to the phone’s ISP (Image Signal Processor) without bottlenecks.
• Power supply voltages:
  • Analog: 2.8 V / 1.8 V
  • Digital: 0.82 V
  • Interface: 1.8 V or 1.2 V Low-voltage digital core helps reduce power consumption during high-resolution or high-frame-rate operation.

Additional Core Capabilities

• Autofocus: All-pixel autofocus (phase-detection across the entire array) in all supported modes.
• On-sensor AI processing: Integrated circuit for learning-based remosaicing, enabling “high-quality” 4x digital zoom (including in 4K/30 fps video) from a single lens without dedicated telephoto hardware.
• Other notes: Supports full-pixel readout for autofocus even in binned modes; designed for compatibility with flagship SoCs like Snapdragon 8 Elite series.

These specifications position the LYT-901 as a versatile, high-end solution that prioritizes resolution without fully sacrificing low-light or HDR performance—addressing common criticisms of earlier 200MP sensors. Real-world results depend heavily on lens quality, OIS (optical image stabilization), and manufacturer tuning (e.g., Vivo’s implementation in the X300 Ultra emphasizes a 35mm equivalent focal length and upgraded CIPA 6.5 stabilization). As more devices launch and sample images emerge in 2026, these specs will be put to practical tests against competitors like Samsung’s ISOCELL HP series or OmniVision’s newer 200MP offerings.

Key technologies

Below is a detailed breakdown of its key technologies, based on Sony’s official announcements and technical descriptions.

1. Quad-Quad Bayer Coding (QQBC) Array

This is the foundational pixel arrangement that enables the sensor’s dual-personality behavior: high sensitivity in everyday shooting and full high-resolution detail when needed.

• How it works: The sensor uses a non-standard color filter array where 16 adjacent pixels (a 4×4 grid) of the same color (red, green, or blue) are grouped together. In normal (wide-angle or low-light) shooting modes, these 16 pixels are analog-binned (combined) into one effective larger pixel. This 16-in-1 binning significantly increases light-gathering capability, boosts the saturation signal level, reduces noise, and improves overall sensitivity—ideal for indoor, nighttime, or dimly lit scenes.
• Resolution trade-off and benefit: Binning delivers output equivalent to roughly 12.5 megapixels (strong default mode for most photos) with dramatically better low-light performance than a conventional Bayer sensor of similar physical size would achieve at full resolution.
• High-resolution mode activation: When maximum detail is required (e.g., cropping for zoom or capturing fine textures), the sensor performs remosaicing to convert the clustered QQBC data back into a standard Bayer-like pixel grid, unlocking the full ~200 effective megapixels.
• Advantages over traditional approaches: Compared to standard Quad Bayer (4-in-1) or non-binned sensors, QQBC provides a larger effective pixel area during binning while still preserving access to the native high pixel count. Advances in underlying pixel structure and color filter design further raise the saturation signal, contributing to better dynamic range even at full resolution.

This technology is central to the sensor’s claim of delivering both “high resolution and high sensitivity” in a single large-format (1/1.12-type, 14.287 mm diagonal) design with 0.7 μm individual pixels.

2. AI Learning-Based Remosaicing (On-Sensor AI Processing Circuit)

An industry-first for mobile image sensors: Sony integrates a dedicated AI-powered processing circuit directly into the LYT-901 die to handle the computationally intensive remosaicing step.

• Why it’s needed: Converting clustered QQBC data back to a full-resolution Bayer pattern requires extremely complex calculations to avoid artifacts (color moiré, aliasing, lost detail). Traditional software-based remosaicing on the phone’s application processor can be slow, power-hungry, and prone to quality loss—especially at high frame rates or during video.
• How the AI works: Sony developed a specialized AI learning-based algorithm trained to intelligently reconstruct fine details, textures, and edges from the binned/clustered data. This on-chip hardware acceleration enables fast, high-quality remosaicing without heavily burdening the smartphone SoC.
• Practical impact:
  • Enables up to 4x high-quality digital zoom from a single (monocular) lens with preserved sharpness, reduced noise, and minimal artifacts—even in 4K video at 30 fps.
  • Supports high-speed burst shooting and video modes (e.g., 30 fps at 50 MP binned, 10 fps at full 200 MP).
  • Reduces the dependency on dedicated telephoto lenses in multi-camera arrays, potentially simplifying smartphone camera modules in future designs.
• This on-sensor AI is one of the sensor’s most distinctive features, differentiating it from competing 200 MP sensors (e.g., Samsung ISOCELL HP series or OmniVision OVB0D) that rely more on traditional remosaic or off-chip processing.

3. Advanced HDR Technologies (Achieving >100 dB Dynamic Range ≈ 17 Stops)

The LYT-901 combines multiple HDR methods for smooth tonal reproduction across bright highlights, deep shadows, and everything in between—critical for real-world scenes with high contrast (e.g., backlit portraits, landscapes with sun and shade).

• Dual Conversion Gain HDR (DCG-HDR): Reads each frame at two different analog gain settings (high gain for shadows, low gain for highlights) and composites them in a single exposure. This single-frame approach minimizes motion artifacts compared to multi-frame merging.
• Fine 12-bit ADC: Upgrades the analog-to-digital conversion from typical 10-bit to 12-bit precision, providing finer gradation steps (4096 levels per channel vs. 1024). This preserves subtle color transitions and reduces banding/quantization noise, especially noticeable in HDR composites, gradients (skies, skin tones), and when cropping/zooming.
• Hybrid Frame-HDR (HF-HDR): Combines DCG data with additional short-exposure frames (very brief, microsecond-class exposures) processed partly on-sensor and finished on the phone’s ISP. This hybrid method pushes dynamic range beyond 100 dB (roughly 17 stops), delivering natural-looking results with suppressed blown-out highlights and lifted shadow detail without excessive noise.

Together, these HDR pipelines produce images closer to human visual perception, with smoother highlight roll-off and cleaner shadows than previous mobile sensors.

Additional Supporting Technologies

• Stacked CMOS Architecture (Exmor RS lineage): Separates photodiodes from logic circuits across layers, enabling faster readout, reduced rolling shutter in video, lower power consumption, and integration of on-sensor processing (AI remosaicing, HDR merging).
• All-Pixel Autofocus: Phase-detection autofocus points cover the entire sensor area in all modes/resolutions.
• High-Speed Video Modes: Native support for 8K at 30 fps (binned), 4K at 120 fps (binned), and high-quality 4K/30 fps at up to 4x zoom thanks to on-sensor AI.

These technologies collectively position the LYT-901 as a highly versatile flagship sensor, optimized for premium 2026 smartphones (e.g., confirmed in Vivo X300 Ultra, expected in Oppo Find X9 Ultra and others). It emphasizes computational efficiency, single-lens versatility, and professional-grade image quality in a mobile form factor.

---

1) Quad-Quad Bayer Coding (QQBC) array

The Quad-Quad Bayer Coding (QQBC) array is the core pixel filter arrangement used in the Sony LYTIA LYT-901 (and potentially future high-resolution LYTIA sensors). It represents an evolution of Sony’s earlier Quad Bayer Coding technology, scaled up to handle the challenges of packing approximately 200 effective megapixels into a relatively large mobile sensor format (1/1.12-type, with tiny 0.7 μm individual pixels) while delivering both exceptional low-light sensitivity and true high-resolution detail when required.

QQBC is not just a minor tweak—it’s a deliberate design choice that enables the LYT-901’s signature dual-mode operation: high-sensitivity everyday photography (via heavy analog binning) and high-definition zoom/cropping (via intelligent reconstruction). This array directly supports the sensor’s ability to produce clean, detailed images up to 4x digital zoom from a single lens, often with quality approaching what dedicated telephoto modules provide.

1. Basic Structure of the QQBC Array

• Pixel grouping: In a standard Bayer filter (the most common color filter array in cameras), pixels are arranged in a repeating 2×2 pattern (one red, two green, one blue) to capture color information while approximating full RGB per “effective” site.
• Quad Bayer (Sony’s prior generation, seen in sensors like those in the Xperia 1 series or A7S III-derived designs): Groups 4 adjacent pixels (2×2) of the same color under one larger color filter patch. This allows analog 4-in-1 binning for better sensitivity.
• Quad-Quad Bayer Coding (QQBC): Extends this further by grouping 16 adjacent pixels (a full 4×4 grid) under filters of the same color (red, green, or blue). This creates much larger effective “super-pixels” during binning.

The 4×4 clusters are arranged in a tiled pattern across the sensor, repeating to cover the entire array. Visually, the raw color filter layout looks like large blocks of the same color (e.g., 4×4 green, then 4×4 red, etc.), rather than the fine checkerboard of a classic Bayer.

2. Dual-Mode Operation: How QQBC Enables High Sensitivity and High Resolution

The genius of QQBC lies in its flexibility—analog binning for normal shooting and digital remosaicing for detail-critical modes.

• Standard / Wide-Angle / Low-Light Mode (High Sensitivity):
  • The sensor performs analog addition (binning) of the 16 pixels within each 4×4 same-color cluster.
  • This combines their charge before readout, acting like one much larger effective pixel (~4× larger linear area, collecting ~16× more light per “site”).
  • Result: Output resolution drops to approximately 12.5 megapixels (200 MP ÷ 16), but with dramatically improved signal-to-noise ratio, higher saturation point, better dynamic range in shadows, and cleaner low-light performance.
  • This is the default mode for most photos, indoor scenes, night shots, or when maximum sensitivity matters more than pixel-level detail.
• High-Resolution / Zoom / Crop Mode:
  • Binning is disabled (or partially reduced).
  • The sensor reads individual (or smaller groups of) pixels.
  • To reconstruct a usable full-color image from the clustered same-color layout, the sensor applies remosaicing (pixel arrangement conversion).
  • This process interpolates and reconstructs a standard Bayer-like pattern, unlocking the full ~200 MP resolution (or intermediate modes like 50 MP with 2×2 binning).
  • Result: Fine details, sharp textures, readable text, and high-quality digital zoom (up to 4x claimed “high-definition” quality, even in 4K/30 fps video).

3. Technical Advantages and Trade-Offs

• Advantages:
  • Light gathering: The 16-in-1 analog binning provides superior low-light performance compared to non-binned or smaller-binned high-MP sensors (e.g., traditional 200 MP Bayer arrays without heavy clustering).
  • Dynamic range boost: Larger effective pixels during binning raise the saturation signal level, helping achieve the sensor’s claimed >100 dB (~17 stops) when combined with DCG-HDR and other techniques.
  • Zoom efficiency: On-sensor remosaicing (aided by dedicated AI hardware in the LYT-901) enables high-quality cropping/zoom without needing extra physical lenses, potentially simplifying future smartphone camera designs.
  • Versatility: Supports multiple output modes (12.5 MP binned at 60 fps, 50 MP at 30 fps, full 200 MP at 10 fps) without major compromises.
• Trade-offs:
  • Remosaicing is computationally intensive and can introduce minor interpolation artifacts if not handled perfectly (though Sony’s on-sensor AI circuit minimizes this).
  • In full-resolution mode, the tiny 0.7 μm pixels are inherently noisier without binning, so low-light shots at 200 MP would be impractical without stacking or aggressive denoising.
  • Compared to a pure Bayer 200 MP sensor (like some OmniVision alternatives), QQBC prioritizes sensitivity over native per-pixel color accuracy, relying on reconstruction for detail.

4. Context in Mobile Imaging Evolution

QQBC builds directly on Sony’s Quad Bayer legacy (used since ~2018–2019 in mobile sensors for better low-light via 4-in-1 binning). By quadrupling the cluster size to 16-in-1, Sony addresses the pixel-size shrink necessary for 200 MP on mobile-sized sensors without completely sacrificing usability. Competitors like Samsung (ISOCELL HP series) and OmniVision use different approaches—often traditional Bayer with off-sensor or simpler on-sensor remosaic—but Sony’s QQBC + on-chip AI aims to deliver a more seamless balance between sensitivity, resolution, and zoom quality.

In real-world devices (e.g., early 2026 flagships like the Vivo X300 Ultra), QQBC allows photographers to capture clean, detailed shots in tough lighting while still cropping deeply or zooming digitally with confidence—pushing mobile photography closer to the flexibility of dedicated cameras.

---

2) AI Learning-Based Remosaicing (On-Sensor AI Processing Circuit)

AI Learning-Based Remosaicing (On-Sensor AI Processing Circuit) is one of the most innovative and distinctive features of the Sony LYTIA LYT-901 (also known as LYT-901), marking an industry-first in mobile CMOS image sensors. This technology integrates a dedicated AI-powered processing circuit directly into the sensor die itself. Its primary purpose is to handle the complex remosaicing (array conversion) required by the sensor’s Quad-Quad Bayer Coding (QQBC) pixel layout—converting clustered pixel data back into a high-resolution, standard-like pixel arrangement with superior detail preservation, especially during zoom or high-detail capture.

This on-sensor approach dramatically improves processing speed, power efficiency, and image quality compared to traditional software-based remosaicing performed on the smartphone’s main application processor or ISP (Image Signal Processor). It enables practical, high-quality use of the sensor’s full ~200-megapixel resolution for digital zoom up to 4x, even in demanding scenarios like 4K video recording at 30 fps.

1. Why Remosaicing Is Necessary in the LYT-901

The LYT-901 employs the QQBC array, where 16 adjacent pixels (a 4×4 grid) share the same color filter (red, green, or blue). This design delivers two key benefits:

• In normal/wide-angle/low-light shooting → The 16 pixels are analog-binned into one effective larger pixel → Output is ~12.5 MP with excellent sensitivity, low noise, and strong dynamic range.
• In high-resolution/zoom/crop modes → Binning is reduced or disabled → The sensor must read closer to individual pixels to access the full resolution.

However, because the raw data comes from large same-color clusters (not the alternating RGB pattern of a classic Bayer filter), the output cannot be directly processed as a standard color image. Remosaicing is the required conversion step: it reconstructs the data into an approximation of a conventional Bayer pattern (or equivalent full-color-per-pixel grid) so that downstream algorithms (demosaicing, sharpening, noise reduction, etc.) can work normally.

Traditional remosaicing is computationally intensive—it involves sophisticated interpolation to guess missing color information at each pixel site while avoiding artifacts like moiré patterns, aliasing, color fringing, or loss of fine detail (especially high-frequency elements like text, fabrics, hair, or distant patterns). On older sensors or competing designs, this step is often handled off-sensor by the phone’s ISP, which can lead to:

• Slower processing
• Higher power consumption
• Visible quality degradation at high zoom or in video
• Limitations on frame rates

Sony solves these issues by moving the entire remosaicing engine inside the sensor.

2. How the On-Sensor AI Processing Circuit Works

Sony developed a specialized AI learning-based remosaicing algorithm tailored specifically for the QQBC array and embedded its dedicated hardware acceleration circuit directly into the stacked CMOS architecture of the LYT-901 (building on the company’s Exmor RS stacked sensor lineage, where photodiodes and logic layers are separated).

• AI Learning Aspect → The algorithm is trained (via machine learning) on vast datasets of real-world scenes, patterns, textures, and edge cases. This training allows it to intelligently predict and reconstruct high-frequency details (fine lines, small text, intricate fabrics, distant objects) that are notoriously difficult for conventional interpolation methods. Sony explicitly highlights “superior reproduction of details such as fine patterns and letters.”
• On-Chip Hardware → Unlike software running on the phone’s CPU/GPU/NPU, this is a purpose-built circuit integrated into the sensor die. It performs the remosaicing in real time during readout, before data is sent off-sensor via MIPI interface.
• Key Operational Benefits:
  • Speed — Enables high frame-rate output even in high-resolution or zoom modes (e.g., full 200 MP bursts at 10 fps, 50 MP at 30 fps, or 4K/30 fps video with up to 4x zoom).
  • Efficiency — Offloads heavy computation from the phone’s main processor, reducing power draw and heat during prolonged shooting or video.
  • Quality — The learned model excels at preserving sharpness and reducing artifacts in zoomed/cropped regions, making single-lens 4x digital zoom viable with results approaching dedicated telephoto lenses in many scenarios.

This on-sensor AI remosaicing is what allows the LYT-901 to claim “high-definition image quality” at up to 4x zoom from a monocular (single-lens) setup, including smooth, detailed 4K/30 fps video at those zoom levels.

3. Practical Impact and Real-World Advantages

• Digital Zoom Revolution → In flagship phones using the LYT-901 (e.g., confirmed in devices like the Vivo X300 Ultra and expected in others), users can crop deeply into the frame for 2x–4x zoom without needing a physical telephoto module. The AI ensures zoomed shots retain fine detail, low noise, and accurate colors—critical for portraits, street photography, product shots, or reading distant signs/text.
• Video Performance → Supports real-time remosaicing during video, enabling high-quality 4K/30 fps at 4x zoom with minimal latency or quality drop—ideal for vlogging, stabilization-heavy recording, or cinematic zoom pulls.
• Simplified Camera Modules → By delivering strong zoom performance from one large, high-res sensor, manufacturers can potentially reduce the number of rear lenses (e.g., main + ultrawide only), lowering cost, thickness, and complexity while maintaining flagship-tier versatility.
• Comparison to Competitors → Samsung’s ISOCELL HP-series 200 MP sensors and OmniVision’s OVB0D also use remosaicing, but typically rely on off-sensor or simpler on-sensor methods without dedicated AI hardware acceleration. Sony’s integrated AI approach gives the LYT-901 an edge in zoom quality and processing efficiency.

4. Limitations and Context

While groundbreaking, the technology is not magic:

• At extreme zoom (>4x) or in very low light without binning, the tiny 0.7 μm pixels still face physical limits (noise, diffraction).
• Remosaicing inherently involves some interpolation, so pixel-peeping at 200%+ may show minor reconstruction artifacts compared to a native Bayer sensor (though Sony’s AI minimizes this significantly).
• Full benefit depends on the phone maker’s tuning of the downstream ISP pipeline.

---

3) Advanced HDR Technologies

Advanced HDR Technologies in the Sony LYTIA LYT-901 (also referred to as LYT-901) form a sophisticated, multi-layered system designed to achieve exceptionally wide dynamic range, smooth tonal transitions, rich color gradation, and natural-looking results in high-contrast scenes. These HDR capabilities push mobile imaging closer to professional camera performance, particularly in challenging lighting conditions such as backlit portraits, landscapes with bright skies and deep shadows, indoor scenes with windows, or nighttime cityscapes with neon lights and dark alleys.

The LYT-901 combines three core HDR-related innovations—Dual Conversion Gain HDR (DCG-HDR), Fine 12-bit ADC, and Hybrid Frame-HDR (HF-HDR)—to deliver a claimed dynamic range of over 100 dB (approximately 16–17 photographic stops, depending on exact measurement conditions and mode). This is significantly higher than typical mobile sensors (often 10–13 stops in practice) and approaches levels seen in some high-end cinema or mirrorless cameras.

1. Dynamic Range Basics and Why >100 dB Matters

Dynamic range measures a sensor’s ability to capture detail simultaneously in the brightest highlights and darkest shadows without clipping (pure white blowout) or crushing (pure black loss of detail). It is expressed in decibels (dB) or photographic stops (each stop ≈ 6 dB; 100 dB ≈ 16.67 stops).

• Human eye (adapted) perceives roughly 20 stops in real-world scenes.
• Most previous flagship smartphone sensors achieved 11–14 stops in real-world output after processing.
• Sony claims the LYT-901 exceeds 100 dB in its 16-pixel additive (binned) QQBC mode, delivering images with:
  • Preserved cloud detail in bright skies without blown-out whites.
  • Visible textures and colors in deep shadows without excessive noise.
  • Natural highlight roll-off (gradual transition to white instead of harsh clipping).
  • Reduced motion artifacts in HDR composites.

This performance is especially valuable when cropping or zooming (up to 4x digital), as dynamic range is maintained across the frame.

2. Dual Conversion Gain HDR (DCG-HDR)

DCG-HDR is a single-frame HDR technique that reads each exposure twice at different analog gain settings within the same capture.

• How it works:
  • High-gain readout → Amplifies shadow areas for better detail and lower read noise.
  • Low-gain readout → Captures highlight areas without saturation (clipping).
  • The two readouts are composited (merged) on-sensor or early in the pipeline to create one HDR frame with extended range.
• Advantages:
  • Eliminates ghosting/motion blur common in multi-frame HDR (no need to align separate long- and short-exposure shots).
  • Faster capture → Ideal for video and burst shooting.
  • Works across all resolutions, including full 200 MP and binned modes.
• Contribution to overall HDR:
  • Provides a strong base dynamic range in a single exposure.
  • Maintains tonal quality even when digitally zooming (up to 4x), as the HDR data is available before cropping.

This technology has been refined in Sony’s previous stacked sensors but is optimized here for the high pixel density and QQBC array.

3. Fine 12-bit ADC (Analog-to-Digital Converter)

The LYT-901 upgrades the on-sensor ADC from the typical 10-bit precision found in most mobile sensors to 12-bit.

• How it works:
  • ADC converts analog voltage (charge from photodiodes) into digital values.
  • 10-bit → 1,024 levels per channel (R/G/B).
  • 12-bit → 4,096 levels per channel (4× more gradations).
• Benefits:
  • Finer steps in brightness and color → Smoother gradients (e.g., blue skies, skin tones, sunsets) with less banding or posterization.
  • Better preservation of subtle details in shadows and highlights during HDR merging.
  • Improved post-processing latitude (e.g., pulling up shadows or recovering highlights in editing apps without artifacts).
• Synergy with other HDR tech:
  • The extra bit depth ensures that the composite from DCG-HDR and HF-HDR retains maximum tonal information, especially noticeable in zoomed crops or video grading.

This upgrade is crucial for a high-resolution sensor where small tonal errors become visible when viewing or cropping large files.

4. Hybrid Frame-HDR (HF-HDR)

HF-HDR is the flagship HDR feature unique to the LYT-901 among Sony’s mobile lineup at launch. It builds on DCG-HDR by adding a hybrid multi-frame element.

• How it works:
  • Starts with DCG-HDR single-frame data (high/low gain from one exposure).
  • Captures one or more very short additional exposures (microsecond-scale short frames) to grab extra highlight information.
  • Composites the short-exposure frame(s) with the DCG data on the smartphone’s application processor (post-processing ISP).
  • The hybrid approach combines the motion-artifact resistance of single-frame DCG with the extreme highlight recovery of multi-frame merging.
• Performance claim:
  • Achieves >100 dB dynamic range (Sony’s official figure; industry analyses often equate this to roughly 16–17 stops).
  • Particularly effective in 16-pixel additive (binned QQBC) mode, where larger effective pixels boost signal-to-noise ratio.
• Real-world impact:
  • Significantly reduces highlight blowout (e.g., bright windows/sun) and shadow blackout.
  • Produces images closer to human vision in mixed lighting.
  • Supports smooth HDR in video modes (e.g., 4K/30 fps at zoom) with minimal artifacts.

Sony provides comparison visuals showing superior highlight/shadow retention compared to conventional HDR methods.

Overall System Integration and Practical Benefits

These technologies work together synergistically:

• DCG-HDR + Fine 12-bit ADC → Excellent single-frame base HDR and tonal precision across the zoom range.
• HF-HDR → Pushes total dynamic range to >100 dB by intelligently adding short-exposure data.

In flagship devices (e.g., Vivo X300 Ultra and others launching in 2026), this results in:

• Cleaner, more natural photos in high-contrast everyday scenarios.
• Better video HDR (less flickering, smoother gradients).
• Greater creative flexibility when editing RAW or cropping/zooming.

---

4) Stacked CMOS Architecture

Stacked CMOS Architecture in the Sony LYTIA LYT-901 (also known as LYT-901) is the foundational physical and electrical design that enables many of the sensor’s advanced capabilities, including on-sensor AI processing, high-speed readout, reduced rolling shutter distortion in video, lower power consumption during complex operations, and seamless integration of sophisticated features like remosaicing and HDR merging. This architecture builds directly on Sony’s long-standing Exmor RS lineage of stacked sensors, which Sony has refined over more than a decade for both mobile and professional imaging applications.

The LYT-901 is explicitly described as a stacked CMOS image sensor in Sony’s official announcement, continuing the evolution from earlier back-illuminated (BSI) designs to fully separated, multi-layer stacked structures. This separation of key functional layers allows Sony to pack high pixel density (~200 effective megapixels at 0.7 μm pitch on a 1/1.12-type format), advanced logic circuitry, and specialized processing directly onto the chip without compromising light sensitivity or introducing excessive heat/noise.

1. Basic Principle of Stacked CMOS vs. Conventional CMOS

Traditional CMOS sensors (including early back-illuminated types) place photodiodes (light-sensitive elements) and supporting circuitry (signal amplification, analog-to-digital conversion, noise reduction, etc.) on the same silicon substrate layer. This shared space creates trade-offs:

• Wiring and transistors block some light → Reduced sensitivity (especially in small pixels).
• Limited space for advanced logic → Slower readout speeds, higher rolling shutter in video, and offloading of complex processing to the phone’s ISP (increasing power use and latency).

Stacked CMOS (also called laminated or 3D-stacked) solves this by physically separating the layers using advanced wafer bonding and through-silicon via (TSV) interconnects:

• Pixel layer (top): Contains only the photodiodes, microlenses, color filters, and minimal pixel-level transistors. This is typically back-illuminated (light enters from the rear) to maximize photon capture without obstruction.
• Circuit/logic layer (bottom): Houses high-density analog/digital circuitry, ADCs, memory buffers, signal processing, and—in the LYT-901’s case—dedicated hardware for AI-based remosaicing, HDR compositing, and other on-sensor computations.

These layers are bonded together (often via copper-to-copper direct bonding or hybrid bonding), with vertical interconnects (TSVs or similar) providing high-bandwidth, low-latency communication between them. The result is a thicker but far more capable chip.

2. Specific Advantages in the LYT-901

The stacked design is critical for the LYT-901’s flagship features:

• Integration of On-Sensor AI Processing Circuit — The dedicated AI remosaicing engine (for converting QQBC-clustered data to high-resolution Bayer-like output) requires significant logic resources. Stacking provides ample space in the lower layer without crowding the photodiodes, enabling real-time, high-quality remosaicing for 4x digital zoom, 200 MP bursts (10 fps), and 4K/30 fps zoomed video.
• Faster Readout Speeds — Separating logic allows parallel, high-speed data paths. This supports modes like full 200 MP at 10 fps, 50 MP at 30 fps, 8K/30 fps (binned), and 4K/120 fps, with reduced rolling shutter (jello effect) in video—vital for smooth panning or stabilization.
• Improved Power Efficiency and Heat Management — Complex operations (AI remosaicing, DCG-HDR readout, fine 12-bit ADC) stay on-sensor, offloading the smartphone SoC. The stacked structure minimizes interconnect capacitance and allows optimized power domains, reducing overall consumption during demanding tasks.
• Enhanced Dynamic Range and Noise Performance — The pixel layer focuses purely on light capture (with advances in pixel structure and color filters raising saturation signal). The logic layer handles sophisticated HDR merging (DCG + Hybrid Frame-HDR) early in the pipeline, achieving >100 dB dynamic range with cleaner shadows and smoother highlight roll-off.
• All-Pixel Autofocus Across Modes — Phase-detection autofocus points span the entire array, even in full-resolution or high-frame-rate modes, thanks to fast readout from the stacked logic.

3. Evolution and Context in Sony’s Lineage

Sony pioneered stacked CMOS for mobile with Exmor RS (around 2012–2013), initially using through-silicon vias for bonding. Later advancements included 2-Layer Transistor Pixel technology (seen in some LYTIA models), where photodiodes and pixel transistors are on separate substrates to further boost full-well capacity and dynamic range. While the LYT-901 announcement emphasizes the stacked nature and QQBC + AI integration, it aligns with this progression—prioritizing logic density for computational features over pure pixel-level innovations like 2-layer transistors (though some elements may be incorporated).

Compared to non-stacked competitors (e.g., some Samsung ISOCELL HP-series 200 MP sensors), the LYT-901’s stacked architecture gives it an edge in on-chip intelligence, speed, and efficiency—key for single-lens zoom versatility and video prowess in 2026 flagships (e.g., Vivo X300 Ultra, Oppo Find X9 Ultra).

In summary, the stacked CMOS architecture is not just a structural detail—it’s the enabling technology that allows the LYT-901 to deliver professional-grade computational photography in a mobile form factor, balancing ultra-high resolution with practical usability, low-light performance, and advanced processing directly at the sensor level.

---

5) All-Pixel Autofocus

All-Pixel Autofocus (also referred to as “all-pixel autofocus,” “full-pixel autofocus,” or “all readout modes retain all-pixel autofocus capability”) is a key autofocus feature integrated into the Sony LYTIA LYT-901 (LYT-901) mobile image sensor. Sony highlights this capability prominently in its official announcement and supporting materials, emphasizing that it remains fully available across all supported readout modes and resolutions—a significant advantage for a high-resolution ~200-megapixel sensor with tiny 0.7 μm pixels.

This technology ensures fast, accurate, and reliable focusing no matter the shooting scenario: full-resolution stills, high-frame-rate bursts, binned low-light modes, zoomed digital crops, or high-speed video recording. It contributes to the sensor’s overall versatility, making it suitable for flagship smartphones where quick subject tracking (e.g., portraits, moving subjects, or video) is essential.

1. What “All-Pixel Autofocus” Means in Practice

In modern mobile CMOS sensors, autofocus is typically achieved using phase-detection autofocus (PDAF) pixels embedded within the image array. These special pixels detect phase differences in incoming light to calculate distance and drive lens focus motors rapidly.

• Traditional PDAF → Only a subset of pixels (e.g., dedicated rows/columns or sparse on-sensor phase-detect points) are used for focusing. Coverage is good but not uniform—focus performance can degrade toward edges/corners or in certain modes.
• All-Pixel Autofocus → Virtually every pixel (or a very high density covering the entire sensor area) incorporates phase-detection functionality. This provides phase-detect coverage across 100% (or near-100%) of the frame, enabling:
  • Extremely precise focus point selection anywhere in the scene.
  • Superior subject detection and tracking (e.g., eyes, faces, animals, vehicles) even off-center.
  • Faster acquisition and re-locking in continuous AF (especially useful for burst shooting or video).
  • Consistent performance in low light, high contrast, or when digitally zooming/cropping (since focus data spans the full array before cropping).

Sony’s official phrasing—”all-pixel autofocus” and “all readout modes retain all-pixel autofocus capability”—indicates the LYT-901 maintains this full-frame PDAF coverage whether reading out:

• Full ~200 MP at 10 fps
• 50 MP (2×2 binning) at 30 fps
• 12.5 MP (16-in-1 QQBC binning) at 60 fps
• 8K/30 fps or 4K/120 fps video modes

This consistency is notable because high-resolution readout often reduces available focus data or slows AF in competing sensors.

2. Likely Implementation: Dual-Pixel or Similar On-Pixel Phase Detection

While Sony does not explicitly detail the exact pixel-level structure in public materials (e.g., no direct mention of “Dual Pixel” branding like Canon or Samsung’s “Super Quad PDAF”), industry analysis and contextual clues strongly suggest the LYT-901 uses a dual-pixel phase-detection approach or an equivalent high-density on-chip PDAF method:

• Each photosite is split into two (or more) sub-pixels that receive light from slightly different angles → Phase difference between left/right sub-pixels determines defocus direction and amount.
• This enables on-sensor phase detection without sacrificing light-gathering area significantly.
• Sony has long experience with dual-PD architectures in previous Exmor RS / LYTIA sensors and professional cameras.
• Community discussions (e.g., on forums like GSMArena) note that Sony’s description aligns with dual-PD behavior, though some debate exists about feasibility at 0.7 μm pixel pitch (the world’s smallest claimed dual-PD pixels are around 0.8 μm in recent Sony research papers, but advances may have closed the gap by 2025–2026).

Alternative possibilities include Quad PDAF or advanced on-chip microlens designs that split light for phase detection across more pixels, but the end result is the same: phase-detect data from effectively the entire sensor surface.

3. Advantages in Real-World Use

• Faster and more reliable AF → Especially in challenging conditions: low light (down to very dim levels), backlit subjects, or when the main subject is small/off-center.
• Better tracking for moving subjects → Continuous AF (CAF) benefits from dense phase-detect points, reducing hunting or missed focus during bursts or video.
• Zoom-friendly focusing → When cropping digitally up to 4x (via on-sensor AI remosaicing), the full-frame PDAF data ensures the zoomed region remains sharply focused without refocusing delays.
• Video performance → Smooth, continuous focus pulls in 4K/120 fps or 8K/30 fps modes, with minimal focus breathing or lag—critical for cinematic or stabilized footage.
• Reduced dependency on laser AF or ToF modules → Many flagships pair sensors with auxiliary focus aids; strong all-pixel PDAF can minimize or eliminate the need for them, simplifying camera module design.

4. Comparison to Competitors

• Samsung’s ISOCELL HP-series 200 MP sensors (e.g., HP2) use “Super Quad Phase Detection” or similar, but coverage is often described as high-density rather than strictly “all-pixel,” and full-res readout may limit AF performance more noticeably.
• The LYT-901’s explicit retention of all-pixel AF across modes gives it an edge for consistent, high-speed operation in premium 2026 devices (e.g., Vivo X300 Ultra, expected Oppo Find X9 Ultra).

5. Limitations and Context

• At 0.7 μm pixels, light per sub-pixel for PDAF is limited → Low-light AF performance still benefits from the sensor’s overall high sensitivity (via 16-in-1 binning) and advanced processing.
• Ultimate real-world speed/accuracy depends on phone maker’s ISP tuning, lens quality, and software algorithms (e.g., AI subject recognition enhancing PDAF data).
• No public sample images or detailed benchmarks isolate AF performance yet (as of March 18, 2026, with initial devices just launching), but Sony’s claims position it as a flagship-grade feature.

In summary, all-pixel autofocus on the LYT-901 leverages dense on-sensor phase-detection to deliver uniform, high-performance focusing across the sensor’s massive resolution and varied readout modes. It supports the sensor’s goal of single-lens versatility—combining high detail, zoom quality, and reliable operation in a compact mobile package—making it a cornerstone for next-generation smartphone photography and videography.

---

6) High-Speed Video Modes

High-Speed Video Modes in the Sony LYTIA LYT-901 (LYT-901) refer to the sensor’s supported video recording capabilities at elevated frame rates and resolutions, leveraging its stacked CMOS architecture, fast readout speeds, intelligent pixel binning (via the Quad-Quad Bayer Coding array), on-sensor AI processing, and high-bandwidth MIPI interfaces. These modes enable smooth, high-quality motion capture suitable for action footage, slow-motion effects, stabilized video, and professional-grade mobile filmmaking—while maintaining features like all-pixel autofocus, advanced HDR, and (in select cases) high-quality digital zoom.

The LYT-901’s video performance stands out for a 200-megapixel-class mobile sensor, as it balances extreme resolution potential with practical high-frame-rate output through aggressive binning and on-chip processing. All video modes retain all-pixel autofocus capability, ensuring consistent, fast focusing during recording.

Official Video Mode Specifications

The sensor supports the following key video-oriented modes in 16:9 aspect ratio (optimized for video workflows), plus hybrid still/video-relevant modes:

• 8K (8K4K, 7680 × 4320 pixels, 16:9): 30 fps
  • Achieved via 2×2 binning (reducing effective resolution while combining 4 pixels per color group for better sensitivity and faster readout).
  • This delivers native 8K video at a smooth 30 frames per second—suitable for high-detail cinematic capture, future-proof content creation, or heavy cropping in post-production.
  • Benefits from the sensor’s stacked design for reduced rolling shutter distortion (jello/wobble effect) during panning or fast movement.
  • HDR support (DCG-HDR and/or HF-HDR) is available, helping maintain >100 dB dynamic range even at this high resolution.
• 4K (4K2K, 3840 × 2160 pixels, 16:9): 120 fps
  • Enabled by 4×4 binning (16-in-1 effective combination via QQBC array), which significantly boosts light sensitivity, lowers noise, and accelerates data readout to support this high frame rate.
  • Ideal for smooth slow-motion playback (e.g., 120 fps footage played back at 24/30/60 fps yields 4–5× slowdown with natural motion).
  • Excellent for action sports, wildlife, vlogging with stabilization, or creative effects like bullet-time-style sequences.
  • The high frame rate helps minimize motion blur in fast subjects and pairs well with electronic image stabilization (EIS) in flagship phones.
• Special Zoom-Enhanced 4K Mode: 4K at 30 fps with up to 4x digital zoom
  • This is a flagship highlight: the on-sensor AI learning-based remosaicing processes QQBC-clustered data in real time to deliver high-definition zoomed video without a dedicated telephoto lens.
  • Maintains sharpness, detail (fine patterns, text, textures), and low noise even when cropping deeply into the frame for 2x–4x equivalent zoom.
  • Enables “monocular” (single-lens) zoom versatility in video, reducing reliance on multi-camera arrays and simplifying device design.
  • Frame rate remains fluid at 30 fps, with HDR and autofocus fully functional—making it practical for content creators shooting cinematic zoom pulls, interviews, or dynamic scenes.

Additional context for video-adjacent modes (often used in hybrid still/video workflows):

• 12.5 MP (roughly 4:3, binned): Up to 60 fps (via 4×4 binning or 2×2 crop with HDR options like AD12 split-HDR, DCG-HDR, or low-bitrate modes).
  • This provides smooth 60 fps video at a resolution suitable for 4K output after downscaling, with excellent low-light performance.
• 50 MP (4:3, 2×2 binning): 30 fps — Useful for high-detail 4K video sources or intermediate zoom.
• Full 200 MP readout is limited to 10 fps (still-focused), so it’s not practical for video but shows the sensor’s raw capability.

How High-Speed Modes Are Achieved

• Pixel Binning — Heavy analog binning (2×2 for 8K/30, 4×4 for 4K/120) combines charge from multiple pixels before readout, increasing effective pixel size, boosting signal-to-noise ratio, and allowing faster data throughput without overwhelming the MIPI C-PHY/D-PHY interface (up to 6.0 Gsps per trio).
• Stacked Architecture — Separates photodiodes from logic/processing layers for parallel, high-speed readout with minimal latency and power draw.
• On-Sensor AI Remosaicing — Critical for the 4x zoom 4K/30 mode; handles complex reconstruction in real time, offloading the phone’s ISP.
• Advanced HDR Integration — DCG-HDR (single-frame dual-gain) and HF-HDR (hybrid multi-frame) ensure wide dynamic range (>100 dB ≈ 17 stops) in video, with smooth highlight roll-off and clean shadows—even at high frame rates.
• Fine 12-bit ADC — Provides richer tonal gradation for video grading and post-production.

Practical Benefits and Real-World Impact

In 2026 flagship devices (e.g., Vivo X300 Ultra and others adopting the LYT-901), these modes translate to:

• Cinematic 8K/30 fps footage with massive cropping latitude.
• Ultra-smooth 4K/120 fps slow-motion without sacrificing too much low-light quality (thanks to binning).
• Professional zoom video from a single main lens—ideal for vloggers, filmmakers, or users wanting simpler camera bumps.
• Reduced rolling shutter and motion artifacts compared to non-stacked high-MP sensors.
• Power-efficient operation for longer recording sessions.

Limitations

• Actual phone implementation depends on the manufacturer’s ISP tuning, thermal management, stabilization algorithms, and codec choices (e.g., H.265/HEVC or higher-bitrate options for 8K).
• 8K/30 and 4K/120 are demanding; battery life, heat, and storage requirements may limit prolonged use.
• Full benefit shines in well-lit or moderately lit conditions; extreme low light favors heavier binning (e.g., 12.5 MP/60 fps modes).
• No public consumer video samples widely available yet , but early device reviews confirm the sensor delivers noticeably smooth, detailed high-frame-rate video with strong zoom retention.

Overall, the LYT-901’s high-speed video modes position it as a versatile tool for next-generation mobile videography—bridging high resolution, computational zoom, and fluid motion capture in ways that challenge traditional multi-lens setups.

---