The Sony IMX678, IMX385, IMX585 and IMX327 are high-performance STARVIS sensors widely used in industrial USB cameras and embedded AI systems. This guide compares their night-vision sensitivity, pixel structure, dynamic range and recommended use cases to help engineers choose the right sensor for low-light machine vision

Today, we dive deep into four popular Sony STARVIS sensors – the IMX678, IMX385, IMX585, and IMX327 – to see which one truly shines when the lights go out.

 

Why Night-Vision Performance Matters for Industrial Vision
Industrial cameras often operate in:
Dim warehouses
Outdoor checkpoints
Parking structures
Retail kiosks
AMRs navigating low-contrast environments
AI edge boxes with varying ambient light

 

In these environments, SNR, pixel size and HDR behavior determine whether objects are detected correctly.

Sony STARVIS series is specifically designed to push visibility into extremely low-light conditions, enabling:
Cleaner edges for detection
Lower inference errors
Better tracking performance
Reduced exposure instability

 

Understanding the Night Vision Game: Key Factors

Before we pit them against each other, remember what makes a sensor excel at night:

1. Pixel Size (µm): Larger pixels capture more light photons. This is HUGE for low light.
2. Sensor Size: A larger total sensor area generally allows for larger pixels or more of them.
3. Sensitivity: Measured in lux or quantum efficiency (QE), how well the sensor converts light into an electrical signal.
4. Technology: STARVIS (Back-Illuminated CMOS) is Sony's low-light champ. Look for STARVIS 1 or 2.
5. Signal-to-Noise Ratio (SNR): How much "grain" (noise) you see vs. the actual image signal. Higher SNR = cleaner image in low light.
6. Resolution vs. Light Gathering: Higher megapixels (MP) often mean smaller pixels, which can hurt low-light performance unless the sensor size increases significantly.

 

The Contenders: A Quick Intro

• Sony IMX678: The newcomer (2023). 8MP (3840x2160), 1/1.8" sensor size. Features Quad Bayer (pixel-binning) technology. STARVIS 2.
• Sony IMX385: The established low-light specialist. 2MP (1920x1080), 1/2.8" sensor size. Known for its large pixels. STARVIS (1).
• Sony IMX585: The high-resolution low-light contender. 8MP (3840x2160), 1/1.2" sensor size. Quad Bayer. STARVIS 2.
• Sony IMX327: The budget workhorse. 2MP (1920x1080), 1/2.8" sensor size. STARVIS (1).

 

Pixel Size Dominates Low-Light Performance

Sensor	Pixel Size	Effect
IMX385	3.75 µm	Best SNR, brightest image
IMX327	2.9 µm	Excellent low-light stability
IMX678	2.0 µm (STARVIS 2)	High sensitivity + high resolution
IMX585	2.9 µm	Strong sensitivity with 8MP detail

 

Round 1: Pure Low-Light Sensitivity & Cleanliness (Pitch Black Conditions)

• Winner: IMX385 & IMX585 (Tie, depending on implementation)
  • IMX385: Its crown jewel is its massive 2.9µm pixel size. Combined with its mature STARVIS design, this sensor captures an incredible amount of light per pixel. In near-total darkness, it often produces the cleanest, brightest image among these four, despite "only" being 2MP. Its native resolution avoids potential binning artifacts.
  •  
  • IMX585: Boasts a significantly larger sensor (1/1.2") than the others. While its pixels (2.0µm) are smaller than the IMX385's, the huge sensor area and advanced STARVIS 2 technology (higher QE) allow it to gather immense total light. Quad Bayer binning effectively creates large 4.0µm "super pixels" for low light, rivaling the IMX385's native sensitivity. Implementation is key: Good processing is essential for clean binned images.
  •  
• Contender: IMX678
  • Shares the same 8MP resolution as the IMX585 but on a smaller sensor (1/1.8"), leading to smaller native pixels (1.56µm). Quad Bayer binning creates 3.12µm super pixels. While STARVIS 2 helps, its smaller total light-gathering area compared to the IMX585 puts it a clear step behind in pure darkness performance. It relies heavily on effective binning and processing.
  •  
• Lagging: IMX327
  • Similar resolution and sensor size to the IMX385, but with smaller pixels (2.31µm) and likely older/less optimized STARVIS tech. It's significantly less sensitive than the IMX385 in very low light, requiring more IR illumination or showing more noise/grain.

 

Round 2: Detail in Low Light (Moderate Illumination / IR)

• Winner: IMX585
  • When there's some ambient light or strong IR illumination, the IMX585's massive sensor and high native 8MP resolution shine. It captures significantly finer detail than the 2MP sensors while maintaining excellent low-light sensitivity thanks to binning. This is the sweet spot for detailed night identification.
  •  
• Strong Contender: IMX678
  • Also provides native 4K detail when light allows. While slightly noisier than the IMX585 in low light due to its smaller sensor, it offers a compelling high-res option. Binning provides a decent low-light mode.
  •  
• Trade-off: IMX385
  • While exceptionally sensitive and clean, the 2MP resolution limits fine detail capture compared to the 8MP sensors. Great for seeing "something is there," less ideal for reading small details at distance without ample light/IR.
  •  
• Lagging: IMX327
  • Suffers from both lower resolution and lower sensitivity than the IMX385. Detail capture in low light is the weakest among the four.

 

Round 3: Handling Challenging Light (WDR - Wide Dynamic Range)

• Critical: WDR combats scenes with both very bright and very dark areas (e.g., a doorway at night).
• Winner: IMX385
  • Traditionally known for excellent true WDR performance (not just digital tone mapping) due to its design and larger pixel charge capacity. Handles headlights, bright windows against dark rooms exceptionally well.
  •  
• Strong: IMX585 & IMX678
  • STARVIS 2 sensors incorporate advanced WDR techniques. The IMX585's large sensor helps. Performance is generally very good, but the IMX385 often still holds a slight edge in the most extreme scenarios. Quad Bayer can also aid HDR processing.
  •  
• Adequate: IMX327
  • Offers WDR, but typically less effective than the others, especially the IMX385, in very high-contrast scenes.

 

The Verdict: Which Sensor for Your Night?

• Ultimate Pure Low-Light King (Cleanest Image in Darkness): Sony IMX385. If your priority is seeing anything at all in near-total darkness with minimal noise, this 2MP veteran is hard to beat. Ideal for very low-light perimeter monitoring, astronomy guides.
• Best Balance of Resolution & Low-Light Performance: Sony IMX585. The large 1/1.2" sensor is a game-changer. It delivers stunning 4K detail when light allows and rivals the IMX385 in sensitivity when binned, making it the top choice for demanding applications where detail matters day and night (premium security, critical dashcams). The overall night vision champion for most high-end use cases.
• High-Resolution Value Contender: Sony IMX678. Offers 4K detail on a more common (and potentially cheaper) 1/1.8" sensor size. Good low-light performance via binning, but demonstrably behind the IMX585. A solid choice if budget constraints exist but 4K is desired.
• Budget-Conscious Basic Low Light: Sony IMX327. Gets the job done for basic 1080p night vision at a lower cost, especially with good IR illumination. However, expect more noise and less detail than the IMX385 in true darkness.

Summary Table

Feature	IMX678	IMX385	IMX585	IMX327
Resolution	8MP (4K)	2MP (1080p)	8MP (4K)	2MP (1080p)
Sensor Size	1/1.8"	1/2.8"	1/1.2"	1/2.8"
Pixel Size	1.56µm (3.12µm binned)	2.9µm	2.0µm (4.0µm binned)	2.31µm
Tech	STARVIS 2 (Quad Bayer)	STARVIS 1	STARVIS 2 (Quad Bayer)	STARVIS 1
Pure Darkness	Good (binned)	Excellent	Excellent (binned)	Fair
Low-Light Detail	Very Good	Fair	Excellent	Poor
WDR	Very Good	Excellent	Very Good	Good
Best For	Cost-effective 4K	Ultimate sensitivity	Best balance detail/sensitivity	Budget 1080p

 

Real-World Night Vision Performance
IMX385 — Best Overall Starlight Performance
Largest pixels
Brightest output under <0.01 lux
Excellent for security cameras and night-vision monitoring
Ideal for parking lots, warehouses and checkpoints

 

IMX327 — Stability + Cost Efficiency
Similar low-light behavior to IMX385 but lower performance ceiling
Best for cost-sensitive AMR fleets, entry-level AI vision, and smart city nodes

 

IMX585 — Balanced Sensitivity + 8MP Resolution
Large sensor + 8MP detail
Excellent for facial recognition, LPR/ANPR, AI retail analytics
Better HDR than IMX327

 

IMX678 — STARVIS 2 = High Sensitivity + High Resolution
8MP detail + high quantum efficiency
Best night-vision clarity among high-resolution sensors
Ideal for AI detection pipelines needing clean edges for CNN models
Perfect for industrial USB cameras, kiosks, robotics and inspection systems
 

Why Choose Mature Sensors in 2025? (The ROI Advantage)

• Software Maturity: "Native support in almost all Linux Kernels, Android SDKs, and Windows UVC drivers. No spending months debugging beta drivers."

• Cost Efficiency: "Production yields are maximized, offering the best price-per-pixel ratio in the market."

• Supply Stability: "Massive global stock ensures your production line never stops."

 

Sensor Model	Resolution	Key Strength	Best for Mass Production Projects:	Driver Maturity
IMX291	1080p	Extreme Sensitivity	ATM Cameras, Parking Lot Gate, Face ID Kiosks	⭐⭐⭐⭐⭐ (Universal)
IMX335	5MP	Best Value (Cost)	Industrial Microscopes, Document Scanners, Standard CCTV	⭐⭐⭐⭐⭐ (Universal)
IMX415	4K	Entry 4K Resolution	Video Conference Bars, Smart Retail Shelves, OCR	⭐⭐⭐⭐⭐ (Universal)

 

FAQ

Q1. Which Sony STARVIS sensor is best for ultra-low-light scenes?

IMX385 delivers the best starlight sensitivity thanks to its 3.75µm pixels and strong SNR.

 

Q2. Which sensor is best for AI edge devices requiring high detail?

IMX678, because its STARVIS 2 architecture offers high sensitivity with 8MP detail.

 

Q3. IMX585 vs IMX678: which is better for night-vision analytics?

IMX678 offers better HDR and more detail, while IMX585 provides a larger pixel and improved brightness. Choose IMX585 for brightness, IMX678 for clarity.

 

Q4. Is IMX327 still relevant for robotics in 2025?

Yes. It is stable, sensitive and cost-efficient, widely used in AMRs and entry-level AI vision modules.

 

Q5. Which sensor provides the cleanest edges for CNN/AI inference?

IMX678, due to superior detail, HDR control and STARVIS 2’s noise reduction pipeline.

 

Q6. Which is best for industrial USB3 cameras?

IMX678 → high-resolution industrial vision
IMX585 → strong night-vision + 8MP
IMX385 → extreme low-light
IMX327 → cost-optimized vision

 

Final Thoughts:

There's no single "best" sensor. The IMX585 stands out as the overall leader for combining exceptional low-light sensitivity (thanks to its huge sensor and STARVIS 2 binning) with high 4K resolution, making it ideal for premium applications. The IMX385 remains the pure sensitivity king for the darkest scenarios where resolution is secondary. The IMX678 is a compelling 4K option where the IMX585 might be overkill or too expensive. The IMX327 is a reliable budget choice for basic needs with adequate IR support.When choosing, prioritize your specific needs: Is seeing anything in near-total blackness critical (IMX385) Or do you need to identify faces/license plates clearly at night with good resolution (IMX585) Understanding these sensor strengths will ensure your camera truly owns the night.

 

Ready to Scale Your Production? Stop worrying about driver integration and supply shortages. Goobuy stocks these proven modules for immediate deployment.

 

Relative useful product application and Starvis camera model

1,  STARVIS IMX291 vs IMX335 vs IMX415: Night Vision Camera Guide (2)

 

2,  IMX678 STARVIS USB Camera: Single-Exposure HDR for Industry

 

3,  Custom STARVIS USB Cameras: Complete Project Guide

 

4,  1/2 Inch IMX385 STARVIS Starlight USB Camera with Metal Case  ( NOVEl products)

 

5,  4K Ultra HD IMX678 USB Camera Module with STARVIS 2  ( NOVEl Products)

 

6,  Sony starvis IMX385 vs IMX462/485/585: Night Vision Benchmark

 

7,  Sony STARVIS 2020–2027: Evolution & Future Trends

 

8,  Novel imx335 imx385 Sony Starvis usb camera review