So you got a new GoPro HERO3 Black Edition! Luckily for you, GoPro has added about 137 different filming options so you’ll never really be sure if you’re using the right mode for the activity/film job. Hopefully I can break it down a bit and unshroud the mysteries of the modes. Keep in mind this is written for the slightly technologically inclined.
A few things need to be explained when talking about the technical aspects of video resolutions, framerates, and video quality. In digital video there are a lot of different variables that go into what makes up the overall IQ (image quality). 1080P on a GoPro is FAR different than 1080P on an Alexa – and it’s not just the image sensor that makes it different. Processors in cameras are a huge determinant to how video gets from the sensor to the SD card. There are tricks and shortcuts that can be made to ensure the camera pumps out all the different modes we want in the end.
Starting with the sensor, the camera is told to read a certain window of the overall sensor – the bigger the window, the more taxing it is on the processor. Reading the full sensor means you’re getting the full view coming from the wide angle lens of the GoPro. The sensor in the HERO3 Black edition is 12MP with a 4:3 aspect ratio. For 4K in the GoPro it’s reading the full width of the sensor, but not the full height – it’s capturing a 16:9 portion of the middle of the 4:3 full sensor. Since the final delivery is 4K, it doesn’t have to scale that initial capture at all. This capture window is identical for, say, 1080P Wide; however, it scales it down to 1080P for the final output. The final output size also has a say in processor taxation – we can get 4K at only 15FPS, while we can do that same capture window but scaled to 1080P at 60FPS.
When we talk about digital video sometimes we use the term native to the camera or not. In the HERO3 Black the camera can capture that full wide 16:9 window, scale it to 1080P and do that natively 60 times a second. 1080-60 and 2.7K are just about the most optimal modes in the camera that really show the true power of the sensor and processor combo. There are some modes in the camera that are read off of the sensor slightly differently which will end up showing some aliasing (jagged edges). Take 1440 for example – look at 1440-30 at full resolution and compare it against 48. The way 48 is read and processed is slightly different than 30FPS to get that large of a resolution combined with a high frame rate. You’ll see the same type of thing in a Canon 7D when looking at 1080P at full resolution compared against 720-60 at full resolution. Canon’s 720P will have jagged edges because of the way they process the image to get a high frame rate from such a large sensor (high pixel count).
Let’s talk about sensor window captures and fields of view (FOV). This is where the Medium & Narrow FOVs come from. 1080P Medium only captures a window in the very middle of the sensor, meaning we don’t see the full wide angle of the camera lens – only that middle slightly distorted part. The raw capture is still larger than 1080P, so it still scales down after the capture. 1080P Narrow is an even smaller capture window at the very very middle of the sensor so the distortion is even less. By this time, the capture size is nearly equivalent to the pixel size where we are reading nearly 1920 pixels by 1080 pixels (remember the full sensor is 4000 pixels wide by 3000 pixels tall). In this narrow mode it’ll appear fuzzier since the pixels are very small on the sensor and are accompanied by a decent amount of noise. When we are reading more pixels there are image processes that are done to clean up this noise – and scaling helps a lot. With no scaling, the narrow mode is extremely zoomed in from the original (imagine grabbing the middle 30% of a gopro photo – that’s what your full 1080P video will be). This is very different from digital zoom because the final readout is still native, we’re not digitally increasing the video capture size at all.