Originally posted by photoptimist That's an interesting observation. You are right that built-in memory can be faster than removable SD memory cartridges. And certainly Apple has made the architectural decision to not provide slots for "obsolete" removable memory cards. And as long as one does not take too many photos with a smartphone and has regular broadband access to the cloud, a fixed amount of image memory can work quite well.
But would photographers, especially pro photographers be happy with a camera with no removable memory cards? Even the K-1 with it's pokey 4 frames per second can fill a 128 GB internal memory bank in less than 11 minutes. Then what? You need a removable cartridge plus all the circuits and power to make a second transfer of the images.
But now it takes minutes to clear what has become a giant second buffer. If the ultimate destination for the image is a removable card, then the least cost, least complexity, and least power consuming option is to save the images directly from buffer to card. Adding a second layer of image transfer only makes things worse. So the "buffer -> cartridge" concept is actually the right one for high-end cameras.
But fast cards consume a lot of power. The UHS-II interface for high-speed SD cards can actually require about twice as much power as the Milbeaut processor consumes. There's a reason SD cards get so hot.
The deeper issue isn't that cameras haven't worked out their design yet. The real problem is that the underlying technology trends for data production and data transmission have never been favorable. It's a lot easier for Sony to make a new sensor that pumps out twice as much pixel data than it is for memory card makers to double the data rate of their interfaces.
I think the worst problem is that Japanese camera companies haven't even tried to come to a more mature solution, even if it doesn't fix the problem forever. Now, while the smartphones eat up their market, they appear to be even less likely to do so, which is strange. Unless an instinct of some sort wakes them up.
They have not set achievable good goals across the range of their cameras, so that the cost of implementing can be shared among models and minimised. However, n
ew specs would create new consumer appeal, and new demand.
Let's illustrate: 8 RAWs deep buffer obviously is a joke for any category today, but what about 50 RAWs /100 JPEGs in lower camera category, and 100 RAWs / 200 JPEGs in higher camera category, both as the new standard for a while? Then work out buffer → internal RAM → SD card best components.
Consumer cameras with one battery and without grip in this case may be built around
6 fps, 32 GB RAM and SDHC only, BUT
even modest 6fps and SDHC becomes formidable, because the choking is taken out of equation. RAM takes care of it. 6fps, 32 GB and SDHC consume less power than 11 fps, 64GB and SDXC/2 from a higher end camera. Hi-end camera have more powerful batteries and grips anyway. RAM storage takes care of sufficient enough writing speed on cards when writing is needed — not before. User notices less problems. Also, reduces the need for additional SD slot, or, even if card(s) are damaged, data is still in RAM. This scenario,
buffer → internal RAM (funnel) → card
is so versatile and better, that a user may be given
an option to press a button, and empty the RAM onto his card, when he wants it! Therefore cards can be in power save mode (of the sorts) until the transfer is requested, and not be powered up every time camera records something.