User Manual - Understanding camera settings
New to photography? Not sure about how ISO and F-stop affect the depth of field? Todd Vorenkamp has a great online learning photography tutorial.
Camera configuration overview
What’s all that stuff about buffers, triggers and saving?
To better understand these concepts, let’s first look at a regular digital video camera (camcorder, phone, professional camera etc.). These devices have 4 major components: A lens to focus light, an image sensor to convert the light into RAW pixel data, an image processor to convert the RAW data into compressed digital video files and a non-volatile storage device to store the digital video file(s).
In an example camera, the sensor produces RAW pixel date at 100MB/sec, the image processor performs around 100G ops/sec. and the storage device can write 20MB/sec. In this case, the stages in this pipeline are well matched and the camera is capable of continuously recording video.
Now let’s look at a high-speed camera where the where the RAW pixel data rate is much higher. The edgertronic SC1 pixel data rate is around 1GB/sec and the SC2+ is around 6GB/sec. This data rate is beyond what any practical image processor can handle, so a different approach is necessary.
How do you drink from a fire hose? Simple … blast the water into a 55 gallon drum and then drink it a glass at a time. High-speed cameras use a similar tactic by inserting a high-speed DRAM buffer between the sensor and the image processor. The pixel data is written into DRAM in realtime (blast) and then the image processing occurs in non-realtime (drink).
The DRAM buffer allows for pretriggering, an extremely useful feature not present in regular video cameras. Once a high-speed camera is powered up and configured, it starts capturing frames into the pretrigger buffer in DRAM. Once the pretrigger buffer is full, older frames are automatically discarded. When the user triggers the camera, there is already some prior vide history in the pretrigger buffer. This allows the camera to capture unpredictable events, such as lightning, by triggering the camera after the event has occurred.
Since high-speed cameras uniquely have this extra DRAM buffer stage, they need additional mechanisms for controlling it’s operation. That’s where duration, pretrigger and save mode come in. Duration is the time, in seconds, of the entire buffer (pretrigger + post trigger). Pretrigger is the percentage of the entire buffer that is pretrigger (pretrigger / ( pretrigger + post trigger)). Save mode controls when the image processor reads out the data stored in the buffers.
Duration and pretrigger are easy to explain. Let’s say you set an SC2+ to 1000fps, 1280x864 resolution, 1 sec duration, and 70% pretrigger. When you trigger the camera, provided that you allowed the pretrigger buffer to fill, you’ll have a total of 1000 frames in the buffer: 700 frames before the trigger and 300 after.
Save mode is a little more complex in that it controls how the buffers are managed. In the example above, 1000 frames only uses about 1/15th of the 16GB DRAM memory, so you have the option to segment the DRAM into multiple buffers. In addition, save mode allows you to specify how and when the image processing and save occur.
In Auto mode, you specify one or more buffers (all buffers use the same settings). You can then trigger the camera, with a minimal (~100ms) dead time between buffers. The data in the buffers will automatically be processed and saved when either (a) the set number of buffers are full, or (b) the user manually forces a save prior to filling all the buffers. Once all the buffers have been saved, the buffers are reset and the camera is ready to be triggered again. Auto mode is typically used in a research or industrial environment where a known number of events are to be captured in an experiment or test.
Review mode is similar to Auto, with one exception: prior to the image processing and saving steps, the user is able to select start and end frames in each of the buffers and save the video between those frames. Until the user manually releases and resets the buffers, the user is free to hop around and save one or more videos, from any region, of any buffer, in any order. This mode would be used in cases where the trigger timing is imprecise compared to the length of the event. For example, when capturing an explosion, you might set the duration to several seconds to make sure you don’t miss it, but the actual event might only last 1/10th of a second. You’d trigger the camera, and then using review, select just before and after the explosion.
In Software Release 2.4.1, we added Background save mode. In this mode, once a buffer is full, the camera will automatically begin processing and saving that buffer. When the save is complete, the buffer is reset and put back into the queue of available buffers. During the actual save, the UI displays sample frames from the save as opposed to the live preview images. This mode is useful for baseball where you want to record every pitch in a game. While the interval between pitches may be shorter than the save time for a single video, the camera has enough buffers to catch up during the pauses in the game play.