[Alcazar]

Hi everyone,
i have a K10D and one of two main reasons why i would upgrade to one of the current models is the speed of writing RAW files to its card.

During portrait shooting, i sometimes do 'manual bursts' (me hitting the shutter repeatedly) at about 2-3 pics/s (the K10D can't burst faster than that anyway but with a new camera i wouldn't increase my burst rate) for something like 2 to 4 seconds (again... pretty much the max for the K10D but i usually wouldn't burst for much longer if i could).

The most annoying thing is that i then have to wait up to 11 seconds until i can review those images on the screen. Which i find awkwardly long when working with a model.

Now, i was hoping i could change that by buying a K3 - the reviews spoke of snappy speed :-)
But looking at the numbers in the pentaxforums review makes me think i would actually increase that problem...?

Saving a single raw:
K10D: 1.4 sec (my measurement)
K3: 2.6 sec (from https://www.pentaxforums.com/reviews/pentax-k-3-review/performance.html)

wait time when buffer is full:
K10D: ~11s (my measurement)
K3: 15s (from https://www.pentaxforums.com/reviews/pentax-k-3-review/performance.html)

Of course i know that it's an unfair comparison because the files are much larger, but i care more for speed than for the added resolution.
And i know that the buffer is larger on the K3 (24 vs 12 pics) and that's an improvement, but if my thinking is correct that only has an effect on how long i can burst, not on how fast i'll be able to review the pics on the screen.

So... what do you think? Am i better off with my K10D? Am i missing something?
Thanks for your input
Micha

PS On a side note... i wonder why it takes 1.3 sec to delete a single image? isn't that just marking the file to be overwritable, ie ideally changing a single bit?
PPS The other reason i would upgrade is high iso... ok i admit the K3 has a slight advantage here

---

Last edited by Alcazar; 01-05-2014 at 05:45 AM. Reason: added 2 links to PF review

[Adam]

2.6s is normal for a raw file with a 95Mb/s card. However, the camera shouldn't lock up while it's writing to the card so you can continue to shoot. The fewer photos you take, the faster the buffer is flushed, and I don't think the speed will get in the way in the studio as long as you're doing really short bursts.

Even alongside the K-5, the K10 feels archaic to me. There's no reason to keep one if you have a K-3, IMO.

PS On a side note... i wonder why it takes 1.3 sec to delete a single image? isn't that just marking the file to be overwritable, ie ideally changing a single bit?

Probably because Pentax's programmers are lazy

[ArneTorp]

Have you tried turning off NR? I had an epiphany on my K10 when I turned it off for long exposures. Used to wait as long as the exposure it self for preview but just a second with the NR off. And hardly any difference in quality...

[striker_]

What sd card are you using?

[drugal]

The pentax usa site says that at continuous low (3fps) you can shoot 54 raw pictures before the buffer fills. With this in mind I don't think you should have any problem with the speed of the K-3.

[Alcazar]

However, the camera shouldn't lock up while it's writing to the card so you can continue to shoot.

Thanks Adam, but what was important to me was the time until i could review the images, not until i could continue to shoot... no problem there.

Probably because Pentax's programmers are lazy

Yes, i guess so?

Have you tried turning off NR? I had an epiphany on my K10 when I turned it off for long exposures. Used to wait as long as the exposure it self for preview but just a second with the NR off. And hardly any difference in quality...

Yes i know that feeling from when i found out :-D but i had it turned off for a long while now

[Alcazar]

What sd card are you using?

a Sandisk Ultra II 16GB, it says 15MB/s Class 4 on it. Why, should the numbers i'm getting be better? i only have a class 4 card because i assume the camera is the bottleneck here, not the card...?

The pentax usa site says that at continuous low (3fps) you can shoot 54 raw pictures before the buffer fills. With this in mind I don't think you should have any problem with the speed of the K-3.

Thank you, but... that doesn't mean i'll be able to review the images faster... just that i can shoot more images?

[Ex Finn.]

Can you rent one to try it out, instead of relying on non-user replies.

[VisualDarkness]

Class 4 is a low-end card and could be the problem.

[crussellsprout]

I just upgraded from a K10D to a K3 myself a week ago, and the new camera seems so much faster in every respect. I don't normally burst RAW shots as in your application, but I just did a couple quick tests firing off 12 RAWs in a few seconds and then waited to see how long the delay was to preview the shots after pressing the view button -- with the buffer half full like this it definitely seemed less than a 10s wait, though I didn't time carefully. Not sure if the Class 4 card is limiting writing the K10D 10MP RAWs, but I certainly wouldn't recommend using it shooting RAWs on the K3. I just picked up a 64GB Lexar Class 10 UHS-I card on sale for $40, no excuse not to upgrade to a faster card these days!

P.S. Selecting all 12 shots to delete (using the +/- button as I saw another poster point out) is a great feature on the K3 to speed up deleting, it probably took 1.3s to delete all of them

[Na Horuk]

Well, I guess the only thing that can be done to persuade the OP is if someone has both of these cameras and would make a video of doing what the OP wants to do, with a visible timer
Anyway, the K-3 should perform much faster in every respect. Please note that previews can be significantly slowed down if you have things like distortion correction or CA correction enabled (even if you shoot raw, because it still corrects jpeg previews embedded in the raw container)
Also, while its true that cameras are often the bottleneck, 15MB/s is still very slow. Go for at least 45MB/s with modern cameras. It does make a difference. A fast, reliable card is a relatively cheap investment (compared to lenses and cameras and tripods), but it is an important cornerstone. Like SanDisk, Lexar.. some people like Transcend..

[crussellsprout]

Please note that previews can be significantly slowed down if you have things like distortion correction or CA correction enabled (even if you shoot raw, because it still corrects jpeg previews embedded in the raw container)

Interesting, I didn't even think about this affecting RAW saving, but I actually had all lens corrections enabled during my test above (had the old 18-55 kit lens on, only one I use corrections with, incidentally) -- without those corrections the test should have been even more satisfactory for the OP.

Go ahead and make the upgrade, I think you will be just as impressed as I have been! Jumping to a flagship model 5 generations newer is quite the upgrade

[dosdan]

a Sandisk Ultra II 16GB, it says 15MB/s Class 4 on it. Why, should the numbers i'm getting be better? i only have a class 4 card because i assume the camera is the bottleneck here, not the card...?

My fastest SD cards were Sandisk Ultra 32GB & 16GB. These were more than enough for the K-5, which won't make use of the extra speed. (It does help a bit though when I transfer the files afterwards in a USB2 ext card reader to my PC.)

But I was appalled to see how slow the buffer clearing was in the K-3 with the Ultra.

First off, look at the buffer performance figures here:

Pentax K-3 Review - Performance and Burst Mode - PentaxForums.com

A point of confusion is the Sandisk Extreme vs Extreme Pro. There is a 3rd version, Extreme Plus. In some sales regions an Extreme is actually an Extreme Plus. (If so, it will show "80MB/s" as the read speed on the label.)

The claimed speeds, in MB/s Read/Write, are:

Extreme: 45R/45W
Extreme Plus: 80R/60W
Extreme Pro: 95R/90W

The Extreme version that PentaxForums used in their K-3 burst test seems to be the Extreme 45R/45W version. So the difference between this and the Extreme Pro was more significant.

At a number of athletics meets with K-5 raw files, I've filled the 32GB card and had to put in the 16GB card. I knew the K-3 files were significantly bigger, and that the quicker buffer clearance would encourage me to shoot more bursts, so I decided I had to go 64GB.

And very frequent K-5 buffer filling in many athletics events made me decide I wanted the very best buffer clearance rate I could get in the K-3. So I chose the Extreme Pro.

Note: the Samsung Pro 64GB is much cheaper. But it is 80R/40W, and while this would be good for general use, for sports you really need the best write speed possible. Normally I'm a cheapskate with SD cards, but not this time.

Here is the performance of the Extreme Pro 64GB in my USB3 ext. card reader. (Completely filled, and then verified, so this is whole-card performance. Some cards start off quick and then slow down a bit, but not this card):

Writing speed: 81.6 MByte/s
Reading speed: 83.7 MByte/s
H2testw v1.4

Note: these are true MB (1,048,576 bytes), not 10^6 bytes.

When I put the card in the K-3 and used its USB3 port, instead of an ext. card reader, to copy the test files to the PC, I only got 27.5 MB/S. So, at least in my case, I'll be sticking with using an ext. card reader.


Dan.

[MadMathMind]

Also, while its true that cameras are often the bottleneck, 15MB/s is still very slow. Go for at least 45MB/s with modern cameras. It does make a difference. A fast, reliable card is a relatively cheap investment (compared to lenses and cameras and tripods), but it is an important cornerstone. Like SanDisk, Lexar.. some people like Transcend..

And 15MB/s is the burst rate, not sustained transfer. I have some Sony Class 10 cards that are rated for 40 MB/s that sustain throughput of about 16-20 MB/s through a card reader. Some is lost to the overhead of the card reader and USB interface, but 40 MB/s is well under the maximum of the USB2.0 spec (480 Mb/s -> 60 MB/s, of which about 60-80%, or 36-48 MB/s can be expected). In short, the speed claims of the card are to be taken with a grain of salt.

So, I'd try a faster card to see what happens. The lead-in and out times always eat a lot of time, but if you got 100% of the rated speed immediately upon the start of writing, the raw files would still take just over 2 seconds to write. That is exactly what you are seeing, no?

The K10D's raws are about 40% or so (right?) of the K-3, so actually, the K-3 is performing better because it's taking only about 1.85 as much time to write over double the amount of data to the same card.

PS On a side note... i wonder why it takes 1.3 sec to delete a single image? isn't that just marking the file to be overwritable, ie ideally changing a single bit?

Depends on the implementation of delete. If the camera is deleting by writing 0's to the original file's location, that would explain it. The other explanation is that the camera uses the file system poorly. When the card on my K-5 IIs gets full, it takes seconds to fire up the viewer. Even when empty, it takes longer to delete than I like.

[Alcazar]

the K-3 is performing better because it's taking only about 1.85 as much time to write over double the amount of data to the same card.

Yes, technically it is. But for me, it means it's performing worse since ... well... i wait 1.85 times as long