Recently, Apple has launched its Deep Fusion Photography system in the latest developer beta of IOS 13.2. For processing medium-light images, Deep fusion will act as a new image pipeline. It is a revolutionary change in the field of cameras. As per the reports, it uses machine learning to create more detailed, sharper and more natural-looking photos. Indoor or medium- lighting would not be a problem anymore.
Company’s Senior VP Phil Schiller, at the launch event, described deep fusion as ‘computational photography mad science’. Adding on, Apple has rolled IOS 13.1.2. & iPad 05.13.1.2. up-date to fix bugs and upgrade performance. The IOS 13.1.2. up-date deals with flashlight, I cloud backup issue and majorly BlueTooth disconnect problem and addresses the shortcuts issue on the HomePod.
With Deep Fusion, the iPhone 11 and 11 Pro cameras are going to have 3 modes of operation which immediately kick in based on light levels and the lens you’re using:
- The regular wide-angle lens is going to use Apple’s enhanced Smart HDR for bright to medium-light scenes, with Deep Fusion kicking within for moderate to low light, and Night mode coming on for dark scenes.
- The tele lens will mainly use Deep Fusion, with Smart HDR just taking over for very bright scenes. (Night mode usually uses the conventional wide-angle lens, flat if the camera app shows “2x”.)
- The ultra-wide will always use Smart HDR, as it doesn’t support Deep Fusion or Night Mode.
Deep Fusion is doing a lot of work in the background and operating differently as compared to Smart HDR, Here’s how:
- By the time you press the shutter button, the camera has already grabbed 4 frames at a fast shutter speed to freeze activity in the shot and 4 standard frames. When you press the shutter button it grabs just one long exposure shot to capture detail.
- The long exposure shot and the three regular shots are merged into ‘Synthetic long’, this is how it differs from Smart HDR.
- Deep Fusion picks the short exposure picture with the most information and merges it with the artificial long exposure. Unlike Smart HDR, Deep Fusion merges these 2 frames, no more-although the synthetic long is currently made of 4 previously merged frames. All the components frame can also be processed for noise differently compared to Smart HDR, in a means that is much better for Deep Fusion.
- The pictures are run through 4 detail processing steps, pixel by pixel, each one customized to increasing quantities of details – the sky and walls space are probably the lowest band, while skin, fabric, hair, and so on are probably the highest level. This creates a sequence of weightings for how you can blend the 2 images – taking detail from a single and firmness, color, as well as luminance from the other.
- The final image is generated.