In the realm of high-end aerial cinematography and remote sensing, “ghosts” are not supernatural entities but are instead physical and digital phenomena that can degrade image quality or provide misleading data. For drone pilots and imaging professionals, understanding what these ghosts are made of is essential for achieving crystal-clear 4K footage, accurate thermal maps, and precise photogrammetry. These artifacts are primarily composed of stray photons, misaligned data packets, and internal reflections within the complex architecture of modern drone camera systems.
The Physics of Light: The Origins of Optical Ghosting
To understand what optical ghosts are made of, one must look at the physical construction of the lens and the behavior of light as it travels through glass elements. In an ideal scenario, light from a subject travels through the lens and hits the sensor directly. However, in the high-contrast environments often encountered during aerial flight—such as flying toward a sunset or over reflective water—light behaves in more complex ways.
The Air-Glass Interface and Internal Reflection
Optical ghosts are fundamentally made of reflected light. Every time light passes through an air-to-glass interface, a small percentage of that light is reflected rather than transmitted. In a multi-element drone lens, which may contain upwards of ten individual glass elements to correct for aberrations, these tiny reflections can bounce back and forth between lens surfaces.
When a bright light source is within or just outside the frame, these internal reflections eventually reach the sensor in a position different from the primary image. These secondary images are the “ghosts.” They are essentially faint, inverted, or displaced versions of the light source, made of the same photons as the original image but diverted from their intended path.
The Role of Lens Coatings and Geometric Shapes
The appearance of these ghosts—their color and shape—is dictated by the physics of the lens. The distinctive greenish or reddish hues of lens ghosts are made of the specific wavelengths of light that were not absorbed by the anti-reflective coatings. If a drone lens has a multi-layer coating designed to minimize flare, the “ghost” is the residual light that the coating failed to suppress. Furthermore, the geometric shape of a ghost often mimics the shape of the camera’s aperture. If you see a series of hexagonal or circular spots across your aerial footage, you are seeing the physical silhouette of the diaphragm blades, rendered in stray light.
Total Internal Reflection in Protective Domes
For drones utilizing waterproof housings or FPV (First Person View) systems with protective domes, ghosting becomes even more prevalent. The curved surface of a dome can act as a mirror, reflecting the light from the camera’s own lens back into the sensor. In these cases, the ghost is made of the camera’s own reflection, a phenomenon that can be particularly distracting in cinematic aerial shots where the sun hits the dome at an acute angle.
Computational Ghosts: Artifacts of Processing and Motion
As drone technology has shifted toward heavy reliance on computational photography, a new type of ghost has emerged. These are not made of stray light, but rather of “stray data”—errors that occur when the drone’s onboard processor attempts to merge multiple frames or interpret rapid movement.
High Dynamic Range (HDR) Ghosting
Modern drone cameras frequently use HDR imaging to capture details in both dark shadows and bright highlights. This process involves taking multiple exposures in rapid succession and blending them into a single image. However, if the drone is moving quickly or if there is a moving object in the frame (such as a car or a swaying tree), the object will be in a different position in each exposure.
When the internal software merges these frames, it creates a “ghosting” effect where the moving object appears translucent or repeated across the frame. These ghosts are made of temporal discrepancies. They are the digital manifestation of a subject existing in two places at once within the same processing cycle. To combat this, advanced drone AI now uses “de-ghosting” algorithms that attempt to identify which version of the object is the “real” one and mask out the others.
Rolling Shutter and Temporal Artifacts
Most drone sensors use a rolling shutter, which captures the image row by row from top to bottom. When the drone’s propellers enter the frame or when the aircraft undergoes high-frequency vibration (known as “jello”), the sensor records the movement at different stages of the shutter’s transit. This creates “ghost-like” distortions where straight lines appear bent or propellers appear as detached, floating arcs. These artifacts are made of time-lagged data, representing a failure of the sensor to capture a single moment in time across the entire imaging plane.
Video Compression and Inter-frame Prediction
In long-range FPV systems or digital downlinks, ghosts can appear as “macroblocking” or “smearing.” These digital ghosts are made of prediction errors. Most video compression formats (like H.264 or H.265) do not record every single frame in its entirety. Instead, they record a “keyframe” and then only record the changes in subsequent frames. If the signal drops or the processor is overwhelmed, the system may “predict” motion incorrectly, leaving a ghostly trail of pixels from a previous frame lingering on the screen.
Thermal Ghosts: Residual Heat and Sensor Calibration
In industrial drone applications, such as search and rescue or infrastructure inspection, thermal imaging cameras (bolometers) are the primary tool. These sensors operate in the long-wave infrared spectrum, and they have their own unique “ghosts” that are made of something entirely different: residual thermal energy.
Thermal Latency and Image Persistence
Thermal ghosts occur due to a phenomenon called “burn-in” or thermal latency. When a drone’s thermal camera is pointed at a very hot object—such as a power line transformer or a solar panel—the pixels on the microbolometer sensor can become saturated. Even after the camera moves away, the “heat signature” of that object may remain on the screen for several seconds.
This ghost is made of the physical heat retained by the sensor material itself. It takes a moment for the sensor to return to its baseline temperature, creating a ghostly afterimage of the hot object. Professional drone pilots mitigate this by performing a Non-Uniformity Correction (NUC), which involves a mechanical shutter closing briefly to recalibrate the sensor and “clear the ghosts.”
Reflection of the Operator
Because thermal cameras detect heat, any highly reflective surface (like glass windows or polished metal) acts like a mirror for infrared radiation. A common “ghost” in thermal drone inspections is the reflection of the drone itself or the thermal energy of the ground below. These ghosts are made of indirect infrared radiation, and they can lead to false positives in inspections if the pilot does not understand the emissivity and reflectivity of the materials being surveyed.
Engineering the Solution: What Eliminates the Ghosts?
If ghosts are made of stray light, temporal errors, and residual heat, then the evolution of drone technology is essentially a quest to eliminate these components through better engineering and smarter software.
Advanced Nano-Coatings
To eliminate optical ghosts, manufacturers are developing sophisticated nano-coatings. These coatings are made of layers of material only a few atoms thick, with varying refractive indices. They work by creating destructive interference for reflected light—effectively making the light waves cancel each other out before they can bounce back to the sensor. By changing what the lens surface is “made of” at a microscopic level, engineers can virtually eliminate the stray photons that create ghosts.
Global Shutter Sensors
The transition from rolling shutters to global shutters in high-end drones like the DJI Phantom 4 RTK or specialized FPV cameras is a direct response to digital ghosting. A global shutter captures every pixel on the sensor at the exact same microsecond. By eliminating the time-lag between the top and bottom of the frame, the “ghosts” created by motion and vibration are physically impossible to generate.
AI-Driven Post-Processing
In the world of aerial filmmaking, “ghosts” are often removed in post-production using AI-driven temporal filters. These algorithms analyze the frames before and after a ghost appears. Since a lens flare or an HDR artifact follows different mathematical patterns than the rest of the scene, the software can isolate the ghost and replace it with clean data from adjacent frames. In this context, the solution to a ghost made of bad data is a replacement made of “predicted” good data.
Conclusion
In the context of drone technology and aerial imaging, ghosts are the unwanted byproducts of the interaction between light, time, and heat. Whether they are made of internal reflections within a lens, misaligned frames in an HDR stack, or residual heat on a thermal sensor, they represent the limits of current hardware. As we continue to refine optical coatings, sensor readout speeds, and computational algorithms, the “ghosts” that haunt our footage are slowly being exorcised, leaving behind nothing but the high-fidelity reality that modern drone technology aims to capture. Understanding what these ghosts are made of is the first step for any pilot or technician in mastering the art and science of the aerial perspective.