If you have ever photographed under flickering lights, such as the sodium vapor lamps especially common at sporting venues, you know the types of problems that lighting can cause. One image is bright and the next is significantly underexposed with a completely different color cast, or worse, significant brightness and color balance variances can occur in individual images. The biggest problem occurs when using fast/short action-stopping shutter speeds under these types of lights.
Here are five images captured in a Canon EOS-1D X Mark II 14 fps burst at an outdoor soccer game shot under the lights on a Division II university field with the Anti-flicker mode disabled.
The exposures for each image are the same (1/1600, f/2.8, ISO 8000), but the brightness of the images is obviously different. Those captured during the dim cycle of the lights will need to be brightened in post processing and increased noise levels will be noticed. These results are actually quite good compared to some I've come home with from other venues.
In the example below, the top 8 images are consecutive frames captured in a 10 fps burst from the Canon EOS 7D Mark II with a 1/1000 second shutter speed. The subject is a white wall and the lights are fluorescent tubes (I had to go all the way to my basement to find these two sets of four 4' fluorescent tube lights). All images were identically custom white balanced from the center of an optimally-timed image. What you see is the frame capture frequency synching with the light flicker's frequency to cause a different result in almost every frame. The camera's frame rate will make a frame-to-frame difference, but the effect remains similar.
The killer problem with these images is that the entire frame is not evenly affected, making correction a post processing nightmare. The cause of this problem is that, at fast/short shutter speeds, the flicker (dim cycle) happens while the shutter curtain is not fully open.
Because the shutter opens and closes only in the up and down directions (with camera horizontally oriented), the area affected runs through the frame in the long direction regardless of the camera's orientation during capture. When the flicker-effected area is fully contained within the frame, the amount of area affected is narrower at faster shutter speeds and wider with longer shutter speeds.
At significantly longer shutter speeds, the effect from the flickering lights is better averaged in the exposures. At 1/25 second, a reference image I captured during the same test looks very nice.
In this light flicker test, I shot at 1/500, 1/1000 (as shown) and 1/2000 seconds. The 1/500 second test showed approximately 2/3 of the frame severely affected at most, but the 10 frames captured around the most-affected frame had various amounts of one frame edge strongly affected. As you would expect, the 1/2000 second test showed an even narrower band of the flicker's effect running through the image (a smaller slit of fast-moving shutter opening being used), but ... I'm guessing that there are not many venues with flickering-type lighting bright enough to allow use of this shutter speed at a reasonable ISO setting. The 1/500 and 1/1000 settings are more real world settings.
The bottom set of results show off the Canon's awesome Anti-flicker mode. The only difference in the capture of the second set of images was that Anti-flicker mode was enabled. These were a random selection of 8 consecutive frames, but the results from all Anti-flicker mode enabled frames were identical regardless of shutter speed tested. These results are dramatically better than the normal captures and you will not see the slightly-less-than-perfectly-even lighting in most real world photos without a solid, light-colored background running through the frame. Tests from the Canon EOS 5Ds were similar to the 7D II results.
When enabled (the default is disabled), Flicker Mode adjusts the shutter release timing very slightly so that the dim cycle of the lighting is avoided. In single shot mode, the shutter release lag time is matched to the light flicker cycle's maximum output. In continuous shooting mode, the shutter lag and the frame rate are both altered for peak light output capture. Shutter lag can be affected, making the camera feel slightly less responsive and the max high speed burst rate is reduced slightly.
Canon's anti-flicker mode has been designed to work with light flicker occurring at 100Hz and 120Hz frequencies. When such flicker is detected but flicker mode is not enabled, a flashing Flicker! warning will typically show in the viewfinder. The FLICKER warning shows solid when a flicker is detected and the camera’s setting is enabled.
Canon's Anti-flicker mode really is a game changer – it will save the day for some events. This feature alone is worth the price of a camera upgrade for some photographers. The technology works very well.