A while back I noted that the Fuji X-T1 is one of the few cameras that can use a high-speed UHS-II card. That article is here if you want to review it.
After it appeared, Ctein wrote to say:
I'm not so sure those UHS-II cards are a wise investment. I'd be very surprised if the Fuji can max out a 90 MB/sec. card. That would correspond to a sustained throughput of about six RAW images per second. Can it really process anywhere near that fast? I'm not saying the 250 MB/sec. card won't gain you something, but it's a lot like putting a really fast hard drive in a relatively slow computer. There's an incremental improvement but it's modest. Definitely, cards under 20–30 MB/sec are real bottleneck. But above that?
The way to find out is to do a time to comparison. Put in a 90 GB/sec card, set your camera to continuous burst, and mash down the shutter release. When it indicates the buffer is full (by refusing to make any more photographs or slowing way down), take your finger off of the release and time how long it takes to clear the buffer. Also, make note of how many photographs you are able to make before the buffer was full. Now repeat this with your 250 MB/sec card. How much of an improvement in buffer clearing time was there (and,if you could capture any more photographs in burst before running out of buffer)? Was the improvement in performance enough to be worth spending 4–10 times as much for the cards?
I performed the test he suggested using three cards: A 30 MB/sec. SanDisk Ultra, a 94 MB/sec. Sony, and a 280 MB/sec. SanDisk ExtremePro UHS-II card.
The slow Ultra did indeed prove a bottleneck: shooting large JPEGs plus RAW in continuous high (CH) mode (about 8 frames per second), the camera made 20 exposures before slowing to about one exposure every three or four seconds, and it took 1 min. 35 sec. to clear the buffer.
Moving up to the Sony, the camera made 22 exposures at high speed, then took 23 seconds to clear. The shooting rate with the buffer full improved to about one frame per second.
UHS-II card (right) is identifiable by its additional contacts
With the new ExtremePro UHS-II card, the camera took 24 exposures at high speed, and the buffer was cleared in 11 seconds. Moreover, the "buffer full" shooting rate only slowed to about 2 frames per second. I kept up the "buffer full" 2-FPS rate going for 1 min./120 shots, at which point the buffer required 9 seconds to clear.
So it appears that for me, under these shooting conditions, with these cards, the UHS-II card in the Fuji X-T1 records only a few more frames at the Continuous High shooting rate, but then clears about twice as fast as the 94 MB/s card, and also shoots twice as fast after the buffer's full.
You might have different UHS-I cards, and you might be interested in some other performance parameter, such as maximum number of frames in continuous low (CL) mode (which I didn't test). Don't extrapolate from my results to other conditions, and don't make the conceptual mistake of thinking one product is "better" unless it's actually better at the parameter you care about. For instance, if the number of frames captured at high speed were what I was interested in, it's pretty obvious that the increased performance of the ExtremePro card is marginal in that parameter. But if the FPS after the buffer is full is what you care about, then the UHS-II card is worthwhile.
Note also that only a few cameras can utilize UHS-II cards at the present time; the Fuji X-T1 is one of them. The Panasonic GH4 is the only other one I know about. There will doubtless be more in the future.
As Ctein notes, the real decision is whether the tested improvements in performance are worth the increase in cost. And that's up to you to decide.
Mike
(Thanks to Ctein)
Original contents copyright 2015 by Michael C. Johnston and/or the bylined author. All Rights Reserved. Links in this post may be to our affiliates; sales through affiliate links may benefit this site.
(To see all the comments, click on the "Comments" link below.)
Featured Comments from:
Peter Foiles: "You don't address another benefit of fast cards and that is how fast you get your images off the card and on to your computer. You need a USB-3.0-based UHS-II card reader to take full advantage."
Mike replies: As Peter says, you need a UHS-II card reader to see the improved speed of the cards while downloading. I didn't test downloading because I use the card reader on my MacBook Air.
Generally, I would think the benefit of high speed cards is more in getting images off the card quickly versus onto the card quickly. The write speeds would be more relevant for 4k high bitrate video.
Posted by: Jandrewyang | Friday, 16 January 2015 at 06:21 PM
I note that this analysis misses the post-shoot part of the workflow: the amount of time it takes to offload images from the card for selection, processing, etc. That can be a pretty substantial time suck for anyone doing high-volume photography. I've known photogs who tended to covet high-speed cards (and card readers capable of keeping up) for this reason alone.
Posted by: Johnwhitley | Friday, 16 January 2015 at 06:22 PM
Mike/Ctein,
Would a superfast card like this be likely to help in the other direction? I'm thinking of the excruciatingly slow processing rate of Sigma Photo Pro software.
--Charlie
Posted by: Charlie Ewers | Friday, 16 January 2015 at 06:51 PM
if you have a card reader that can handle UHS-II speeds, these cards will also download to your computer more quickly
Posted by: sporobolus | Friday, 16 January 2015 at 07:19 PM
We're starting to see video codecs producing streams above 25MB/s (I've suddenly seen multiple discussions of 50MB/s video, plus of course the excitement that 4K video can bring you). That might be another reason why higher write throughput could be important.
Or, just to make uploading the data faster after a long day of shooting (that's based on read rate of course). For this, the camera doesn't even have to support UHS-II, just the card reader you use to upload.
But yeah, I don't own any UHS-II cards myself and am not particularly faunching after them.
Your points about optimizing things you actually care about are extremely cogent. You might care about max burst speed. You might in addition care about how soon you can perform a second max-speed burst after the last one, or you might not (How fast do the cars come around the race track? Maybe your buffer will be empty in any case, or maybe not. Or it won't matter until the next play. Whatever your action environment needs.) You might care about max long-term burst rate (that is, beyond buffer capacity. You might be fine shooting but have a need to get uploads into your fileserver very fast (especially if one reader is serving multiple photographers at a tournament shoot, say). Or, no doubt, other things I haven't thought of. You probably can't optimize them all at once, you certainly can't optimize them all cheaply. You can pretty likely spend a lot of money optimizing things that don't actually matter to you.
Posted by: David Dyer-Bennet | Friday, 16 January 2015 at 08:29 PM
I hardly ever overflow the buffer in the camera, but when I'm downloading from a 64gb card to the computer and there's waiting time to go make a sandwich, but I don't want a sandwich, well then I'd like a faster card.
Posted by: hugh crawford | Friday, 16 January 2015 at 09:17 PM
This is a real concern for modern video cameras; the Blackmagic can shoot 4K video in DNG RAW at 30FPS; that means one minute of recording lays down 1800 4K DNG files.
Posted by: Bob Blakley | Friday, 16 January 2015 at 10:23 PM
Another advantage of the faster cards is that with a good card reader, getting the images to a computer can be a good deal quicker. Not important for small batches, but if you tend to shoot a lot at once or don't import very often, it can be useful.
Posted by: Bernard Scharp | Saturday, 17 January 2015 at 02:35 AM
Image capture is seldom my bottleneck, it is in post production when i must download hundreds of images from a project or thousands after travel. But much of that is after the heat of battle has cooled.
Cards should be a matter of balance, fast enough to keep up with one's style of shooting versus capacity, perhaps large enough to shoot a job or for a day, versus speed of downloading at the end of the day.
The last seems the most important to me and that's another bottle neck of the system between Firewire and favors of USB with assorted card readers, not to mention what is under the hood of the system that I use to edit the photos. I simply want to clear out the cards so I can edit and move on.
Take some of the card speeds with a grain of salt since the words "up to..." May be present. The best way to figure out what works best for you is to test just like Ctein. YRMV!
Posted by: Larry Angier | Saturday, 17 January 2015 at 02:43 AM
There is another factor to consider when testing cards: electronic noise. Certain SD cards generate electronic noise during write operations that come up in the image of my Leica M9 (due to just questionable shielding in the implementation of the electronics). As a result, I use only Sandisk 8Gb Ultra (or the equivalent to their old generation Sandisk 8Gb Extreme cards) -- this particular card doesn't seem to generate banding noise in the camera. And yes, different sized cards generate different noise profiles, so here size does matter!
To test this in your camera: shoot a predominantly black subject at a range of ISOs in RAW, with continuous shooting (i.e. taking a photo whilst the card is writing). In your raw processor, boost exposure and shadow tones and look for banding -- tell tale signs of electronic noise.
Most modern sensor and sensor integration implementations should handle this without any issues -- but it is something I do check. My Pentax on the other hand passed this test no issues at all.
Pak
Posted by: Pak-Ming Wan | Saturday, 17 January 2015 at 03:11 AM
A fast card is also interesting when you upload the photos to the computer.
Posted by: Allan | Saturday, 17 January 2015 at 03:50 AM
Completely off topic. considering the number of uses of the letters, t, x, and so on, how long before we end up with a Fuji Tri-x 1? presumably with a mono sensor.....
Posted by: Ger Lawlor | Saturday, 17 January 2015 at 04:32 AM
Right now, it's of no use to me at all - I am a boring old plod who takes a good 20s just to take a single frame and chimp the hell out of it.
However I can well understand its general relevance to many others, with cameras like the new NX1 shooting 20 fps with 28MP files, not to mention 4K video at 60fps.
With stacked BSI APSC sensors I suspect the global shutter is closer to reality as well. We could have stills frame rates in the 100s and ultra fast HD video frame rates. Great for slow motion video...
[Except I don't think the NX1 supports UHS-II cards. Its specs say "SD, SDHC, SDXC (up to 64 GB)." --Mike]
Posted by: Steve Jacob | Saturday, 17 January 2015 at 08:02 AM
Mike or Ctein: something I have been curious about but have not seen mentioned anywhere is if various SD cards offer better download speeds? That is, when the card is slotted in the MacBook will the images transfer at different rates for different cards? Then there is the iPad which seems to take forever to just load the preview from a 32gig SD! I haven't really noticed a camera slow down since the Kodak DCS760 would jam up when auto bracketing was turned on and it would pause after 5 frames. Just curious. Oh ... and nice testing!
Posted by: Santasimage | Saturday, 17 January 2015 at 08:02 AM
I bought the Sandisk UHS II card and tried it with my Nikon knowing that it is not supported but wanting to see if there was any performance gain at all. There wasn't. It actually was much slower than the 90MB Sandisk. Using Blackmagic Speed Test with a UHS I Lexar reader It only made it to 40MB write speed (I didn't test it with the UHS II Reader). All of this was expected, but I had to see it for my self.
Posted by: Michael Steinbach | Saturday, 17 January 2015 at 09:42 AM
Why are we still debating cards? By now high end cameras should come with ample amounts of fast built-in memory and the latest USB and wi-fi specs like any decent smartphone, but the camera world seems to be perpetually stuck in the 90s and 00s.
Posted by: Øyvind Hansen | Saturday, 17 January 2015 at 10:20 AM
Conversely, as taking a shot a minute is pretty fast going for me, is there any point in my using anything but the slowest cards?
Posted by: Guy Batey | Saturday, 17 January 2015 at 11:20 AM
One thing that takes much of the pain out of this kind of many-shot testing is to set the camera up focused on an online stopwatch (www.online-stopwatch.com , for example). Start the watch running, blast away, and when you stop to watch the buffer drain, press one last time to mark the time when the write-light stops blinking. Then you can analyze the timing of each phase as finely as you like.
This sort of thing has been de rigeur for years with the somewhat rough firmware in the Leica M8, M9 and the like, for which using a faster card did not always produce faster shooting or a more quickly drained buffer.
scott
Posted by: scott kirkpatrick | Saturday, 17 January 2015 at 04:10 PM
[Except I don't think the NX1 supports UHS-II cards. Its specs say "SD, SDHC, SDXC (up to 64 GB)." --Mike]
Was surprised by that so I checked.
According to Samsung themselves the specs say SD, SDHC, SDXC(up to 64GB), UHS-I, UHS-II
Not sure of the reason for the 64GB limit but I don't think that relates to the speed.
Posted by: Steve Jacob | Sunday, 18 January 2015 at 08:23 AM
I would not buy a camera for fast card support alone, so if my camera has it, it's by accident. But I will eventually find good use for a faster card. Then I will buy it.
Whether it's roads or data, never bet against greater bandwidth.
Posted by: Bob Sacamano | Sunday, 18 January 2015 at 03:16 PM