Sun Dec 13, 2015 10:34 am
It is probably better to take small steps. Most of my suggestions are control related, and easier, and very very useful and productive. As for external controls, in the pre BM, Red and SI days, guys where aiming to use USB keyboards and keypads to control, and there are other such control panels that use USB (depends on if camera uses it). Same as WiFi or bluetooth. So apart from reconfiguring the onboard buttons there are little options on the BM cameras. What we really need is assignable buttons and controls that can be stuck on where you wish, and a way to put on a near field controller. I thought I found such buttons, but it turns out different unfortunately.
Now going back to non control issues, not much needs to be done that can't be done in software. Auto balance color and exposure adjustment on button push, that is software level stuff. If there was a standard format that recorded adjustments in meta data alongside the raw compressed data, that would be good as playback and transcoding can automatically apply to the footage, if you hand it off to the news room and such forth. But that has to be programmed in, likely a lot of hardware fpga programming.
The one other thing is to get 1080p50/60. As I understood the sensor in the pocket used the 1080p100 technology. Now I don't know what the heat profile of the sensor is at p50, but let's just assume it isn't a problem at the moment, but running the compression engine at double the data rate might generate twice the heat or more, unless you redesign the FPGA circuit to use a more easy to compress codec, which can be too complicated as people are saying. But instead, why not push 1080p50 (or whatever maximum frame rate that can be done without overheating) at the maximum data rate in prores that is reliable. It might be doable. Sure it might not be as high quality as you want, but now it becomes an issue of reconfiguring how the circuits are used, which is easier than programming a new codec engine.
But what about 1080p50/60 raw. The problem has been also the data rate the SD cards would do. Now these cards are going much faster, but are prone to heat issues affecting performance. Micron and Intel saves the day though. They have developed an alternative to flash that is cheaper, many times faster and lower powered. Write speeds are much better, which was an issue under flash due to all the unreliability in speed and patching up bad parts of the memory array, in the background then shoved onto market. It can be used as a primary memory. This is not the only new memory too.
Now, let's get realistic, pulling 1080p50 in and shoving it out is not very processor intensive,.it is not such an issue. Compressing it and using flash to record is a much bigger issue processing and flash wise. So recording raw uncompressed with the new memory hopefully should not be such an issue, as long as the data path (and SD card interface chip) supports the data rate. I don't see BM not using something that does not have a fast controlling processor? Frankly, we are at a stage of not needing FPGA in a 4k camera and making it cheaper. When the AMD arm based APU, and arm based server chips from different groups, become available it will make sense to use that, realising you might get long term supply of the chip off a server part and have chip family level upgradability in the supply chain. At the moment the latest Nvidia chip is probably enough to make an 8k camera main board real cheap, I mean really cheap. The embedded board is less than $500 and has camera interfaces that can pull in over 40fps accumulative 8k worth of data. A descent deal can see that done a lot cheaper. The shield Android TV unit goes for $199US retail, and was recently discounted to $149US. I can get a Android 8 core (I think) media player that could maybe be made to process 4k data for around $50US. But you use these today and next year who knows if they will be available, or changed with different varying performance requiring a re-optimisation of the software, external hardware changes. The vast consumer Android market pushes a lot of short term product, butna manufacturer like BM needs consistent long term supply of compatible parts that don't vary in various electrical timing performance characteristics (things can run faster, but interface and timing has to be reliable). Often you will spend a lot of time optimising for a chip to get the most out of it and reliable timing. Often the chips are stretched so you have to do this more. Getting an 8k capable chip and running it on a 4k camera will require less optimisation. Now a finale thing is that chip technology is running out of advancement space under current technology, we are within a generation or two where decreases in chip feature sizes will yield no more increase in performance per watt (unlike the past) due to higher and higher power leakage producing heat. We are facing a temporary future were decreases in size will yield hotter chips unsuitable for smaller devices. We can hop over to less leaky GA instead of silicon and get maybe a double performance, until we get some other alternative future tech (give it till 2020 or so). So if you get an 8k pocket camera, it might pay to look at if to hold onto it, it might be a little while until something better comes. The ingress of data is not the issue, the compression and processing is.
So doing a little that achieves a lot should be possible, doing more might not be worth it considering we might see a 4k pocket like camera (just seen two more cameras, one sub $1000, and there is a several lens like camera that allows you to do 3D and change focus in post) and lots of people will start wanting things like that instead. So concentrating on control, software features, gui and adjustments might be a better target for a BM lantern. I however, think it is morally reprehensible to not do things like 1080p50/60, controls and color correction options, a bit better latitude or frame rate if newer sensors allow it, if you can easily, the customer can get a fast enough SD card if they want to use it.
aIf you are not truthfully progressive, maybe you shouldn't say anything
bTruthful side topics in-line with or related to, the discussion accepted
cOften people deceive themselves so much they do not understand, even when the truth is explained to them