Visual artefacts on an LED screen are like dirt on a window – they break the illusion and make you aware there is something between you and the scene. When viewing through a camera rather than directly with the eye, there are a whole new set of potential artefacts to worry about. We’ll try to give an overview here, and go into more detail in future blogs.
First, you need to ensure each frame seen by the camera contains only one frame of content on the LED screen. If you have rolling bars, then the refresh of the LED screen and the refresh of the camera are not synchronised. If you have tearing or thick static bars, then you may be synchronised but out of phase. Genlock and phase adjustment are essential tools for any on-camera set up.
If you are seeing thin, dark or light static lines on the screen, then it is likely that the time that the camera shutter is open is not an integer multiple of the PWM (pulse width modulation) cycle of the LED screen. Lighter lines are where the camera is seeing an extra partial PWM refresh and dark lines are where the camera is missing out on part of a PWM refresh. In the past, this was only solvable by choosing a camera shutter angle that didn’t show the effect – which meant some creative options were not available to the camera operator. However, Brompton Technology has developed a patented solution to this called ShutterSync, which allows the LED refresh timing to be tuned to suit the shutter angle of the camera rather than the other way around.
Banding in gradients within the image is another form of artefact. Quite often in a virtual production setting we are running LED screens at a tiny fraction of their brightness capability – but that means lots of our bit depth is being wasted, which means there may not be enough left to get a smooth gradient. Choosing a panel with higher bit depth drivers is a good start – but run it at low enough brightness and the same issue is there. Brompton’s Operating Modes allow for different panel configurations that prioritise different performance characteristics – and one option is to have a mode (Studio Mode) for low brightness operation that turns down the driver current so you can utilise the full bit depth. In situations where you need more dynamic range without turning down the maximum brightness, then Brompton has two complementary tools available: Extended Bit Depth, which makes use of any spare performance of the driver chips to multiply up the PWM refresh, creating additional virtual bits, and Dark Magic, which applies spatial and temporal dithering. These tools are designed to be used together and that will usually yield best results.
Other commonly encountered visual issues are moire and scan line artefacts. The LED processing does not have the answers here. Moire is interference patterns due to the grid of pixels on the camera sensor interacting with the grid of pixels on the LED screen, and is dealt with by selecting an appropriate pitch of LED for the distances in your shoot and/or keeping the LED out of sharp focus. Scan line artefacts are due to the electrical design of the panel – where one output from a driver chip is being reused to drive multiple LEDs. This keeps down the number of driver chips, which is more cost effective and necessary to make very fine pitches possible – but be wary of higher scan mux ratios (relatively low is 1:8 and very high would be 1:32) as it will have some performance trade-offs. Scan line artefacts typically show up when panning or tilting the camera across the LED screen, and so it is important to try any camera moves in a camera test with the exact panels you want to use.
In the next few blogs in this series, we will go into more detail on how to recognise and eliminate common visual artefacts to ensure nothing breaks the illusion when recreating reality.