Key Takeaways

AI auto-tracking cameras cannot track multiple targets simultaneously. In multi-person scenarios, you need to manually assign the target you want to track.

When the target is covered or faces away from the camera for more than 3 seconds, the AI tracking may become unstable.

When the target moves rapidly (e.g., getting on or off stage, or running), it may exceed the tracking capabilities of the auto-tracking camera's PTZ mechanism.

When stage light changes colors drastically, strobe, or when there is smoke, strong light beams or laser, it can cause exposure or focus jitter and the AI auto-tracking frame may be lost or misjudged.

Dispelling the Myth: AI Auto-Tracking Cameras Cannot “Read Minds”

Many people think that once an AI auto-tracking camera is set up, they can leave it alone and let it do the filming automatically. However, this is not the case. Upon arriving at a filming location, we often find that the camera lens loses its target or starts panning randomly. In fact, while AI auto-tracking cameras are convenient, they are not human camera operators; they have their own visual logic and inherent limitations. As long as we truly understand the “temperament” of AI auto-tracking cameras, we can safely navigate various live broadcast scenarios.

Getting Lost in Crowds: AI Cannot “Multitask”

In actual shooting scenarios, the first problem we often encounter is that AI cannot multitask. Current AI auto-tracking cameras only have the ability to track one target at a time and cannot simultaneously track multiple targets or judge their priority. When speakers, performers, or walking audience members appear on stage at the same time, the AI can only determine the main target based on factors like the proportion of the screen they occupy. At this time, the lens often mistakes a passerby for the main target, causing the frame to shift. Therefore, in multi-person scenes, we need to manually specify the desired target using a controller or the camera’s Web UI. If the camera lens tracks the wrong target, you must switch to manual control, re-specify the desired target, and then restart the auto-tracking function.

The Fear of Sudden Disappearance: The Crisis of Losing Track Due to Occlusion and Turning Around

Besides the issue of multiple people leading to mis-tracking, AI is also very afraid of “being blocked.” This is because the AI algorithm relies heavily on the contours of the target’s head, shoulders, or upper body when locking onto the desired tracking target. If the target happens to be blocked by an object or faces away from the camera for more than three seconds, the AI will hesitate because it cannot find the target’s features.

At this time, the footage of the AI auto-tracking camera might freeze or directly return to the preset position. However, when the speaker or target reappears or resurfaces, the AI will usually automatically re-detect and resume tracking. If you feel the camera is reacting too slowly, you can also manually click on the Web UI to re-lock the target.

The Trial of Deadly Backlight: Better a Blown-Out Background Than Losing the Main Character

In addition to the “people” issue, lighting is also the biggest nemesis affecting AI auto-tracking cameras. During actual filming, a common situation we encounter is the speaker standing with their back to a row of bright windows, or standing in front of a high-brightness LED video wall. In such strong backlit environments, the camera, to protect the background from overexposure, will lower the overall exposure value, which turns the speaker into a black “silhouette.”

When reduced to a silhouette, the various features on the speaker—including clothing color and facial features—disappear. As a result, the AI auto-tracking camera cannot recognize the desired target. To solve this problem, we must go to the camera’s OSD menu to select Backlight Compensation, or go to the Web UI to select Wide Dynamic Range (WDR), to forcefully increase the brightness of the target person in the center of the frame. If it is still not bright enough, please switch to manual mode to increase the Exposure Value (EV) or Gain. At this point, we must make a trade-off: we would rather let the background overexpose into a pure white field to ensure the brightness of your main tracking target is normal. Please remember, as long as the facial features and contours of your target person light up, the AI auto-tracking camera can closely follow the target.

Colorful Illusions: How to Deal with Flashing Stage Lights?

Compared to backlight, colorful stage lights are another nightmare. Because AI can easily misjudge flickering lights and shadows as human movement, it causes screen jitter or tracking errors. This happens because when the scene features drastic color-changing spotlights, strobe lights, lasers, or thick smoke, the frequent changes in lighting interfere with your AI auto-tracking camera’s human detection system. To handle such high-contrast and chaotic lighting scenes, the best approach is to set the white balance and exposure of your AI auto-tracking camera entirely to manual. This prevents the exposure and white balance values from jumping erratically when lighting and brightness change rapidly, which can cause facial overexposure.

Additionally, you need to turn on the WDR function to counter high contrast, and adjust the tracking speed to an appropriate level. Furthermore, use the camera’s Web UI or remote control to lock onto the target person; this will reduce the hesitation of your AI auto-tracking camera lens.

Why is the Lens Always a Beat Too Slow? Cracking the Settings for Camera Movement, Framing, and Shutter Speed

The final key point is regarding “speed and framing” settings. Sometimes, when the speaker turns and walks, the camera might seem a beat too slow or unresponsive. When you set the tracking parameters too softly, your AI auto-tracking camera might not be able to keep up with the speaker’s sudden fast movements or turns. Additionally, if you set the framing of the AI auto-tracking camera to a close-up, it is very likely that due to the tight framing, the speaker will disappear from the edge of the screen with a slight movement, causing the camera to frequently lose the target. It is recommended that you widen the framing to a half-body or full-body shot. This can increase the framing’s fault tolerance, and the visual sense of delay will also be significantly reduced.

Besides this, when the ambient lighting of the shooting scene is insufficient, it will cause your AI auto-tracking camera to lower its shutter speed. When your camera’s shutter speed drops below 1/60th of a second, noticeable motion blur will occur on the footage. These blurry afterimages make it difficult for your AI auto-tracking camera to identify the target it wants to track, which may lead to tracking failure. Therefore, please ensure your camera’s shutter speed is at 1/100 or higher, because only by ensuring that moving images are clear and sharp can your AI auto-tracking camera unleash its strongest tracking capabilities.

Conclusion and Summary

An AI auto-tracking camera can be a very powerful assistant for you, but what you should do in advance is that you need to clear the visual obstacle for the camera to keep the good lighting condition. Additionally, set the suitable framing and shutter speed for your camera. With proper manual target locking, your AI camera will deliver professional and stable footage.

5 Most Common Failure Scenarios and How to Recover

The following lists 5 common failure scenarios and recovery techniques. Please see the table below for details.