[stevebrot]

My conclusions so far are:
- i should try out a better card for my current camera

Indeed. I upgraded to a class-10 card for my K10D and the improvement in write speed was palpable.


Steve

[Reltih]

no, the camera is not blocked, only the play mode is... you can take another shot right away but you have to wait for the image to be displayed.

if you have some time on your hands, would you try out for me on your K3 how much times passes between
- taking just one shot (in Raw, with NR & lens corrections disabled)
- and the image showing up on the back screen?

Today I just checked again. Yes there is a short delay, approximately less than 1.5 seconds (RAW, no corrections, no lens cap, memory card 95/60). I have never been bothered by this delay and I don't know how to measure it as accurately as Adam and Dan.

[dosdan]

I have never been bothered by this delay and I don't know how to measure it as accurately as Adam and Dan.

I used a stopwatch to determine the delay between taking a shot and having it appear on the back LCD screen. But this will include a human reaction-time component. You can improve on this if you have a video cam that can shoot between 25p-60p FPS. You put it reasonably close to the back of the camera, raise the K-3's flash, have a white wall in front of the camera, start filming, and then take a shot.

Then load the video into a video editor and step through each frame. You will have 40ms (25p), 33ms (30p), 20ms (50p) or 17ms (60p) timing granularity depending on the frame rate used. The flash is very short and you might miss it within a frame, but you should be able to see the shutter click on the audio track as a secondary means of determining when the shot was taken. ( I'd suggest a 1/100s or faster K-3 shutter speed to reduce the delay from this part of it.) You stop counting video frames when the review image appears on the back LCD.

BTW, recording the sound of a camera's burst e.g. with an MP3 recorder, and then loading it into an audio editor, is an easy and accurate (at 44Khz sampling rate, the granularity is 23μS) way to determine burst FPS and buffer depth. See:
https://www.pentaxforums.com/forums/115-pentax-k-5/160077-k-5-continuous-shoo...ter-noise.html

An even easier, but less accurate way is to use the K-3 to take a continuous burst of a timer program running on a PC screen:
https://www.pentaxforums.com/forums/115-pentax-k-5/160077-k-5-continuous-shoo...ml#post1689644

Dan.

[dosdan]

I've done a timing test using video-frame timing, as I described earlier in this thread.

K-3, raw, Sandisk Extreme Pro (95MB/s) 64GB, flash shot of a part of my garage, no corrections (unsupported lens). I used a K-01 shooting 30p video (33ms/frame) for timing.

Time from flash to image appearing on back LCD screen: 0.76s (23 frames)

Time from flash to write light going out: 2.15s (65 frames)

Max. possible timing error (due to frame granularity) here is 2 frames (start + finish frames) = +/- 0.067s, with the typical error being half that, +/- 0.033s

Dan.

---

Last edited by dosdan; 01-17-2014 at 04:04 AM.

[Alcazar]

Time from flash to image appearing on back LCD screen: 0.76s (23 frames)

Time from flash to write light going out: 2.15s (65 frames)

Dan, you're awesome!
Your post made me realise i have made a big mistake: Assuming that it takes the same time for both the automatic preview to appear on the screen and for the play mode to be available.

Because i have the automatic preview disabled! (don't know if that's what it's called - i mean the function where after taking shots the camera shows the last image for a few seconds)
Instead, i wait for the play mode to become available (mainly because i thought it would save battery not having the screen avtivate after every shot regardless of if i want it to)

I guess the automatic preview comes from the buffer and is therefore much faster, while the play mode needs the files to be written to the card to give access to them.

That also explains the discrepancy between what people have been saying here and the numbers from the PF review of the K3.

So... the K3 seems to be a fast camera but i can be much better off with either camera if i reactivate the automatic preview (even if that means i can only see the last shot).
Well, another thing learned. Thanks Dan!

[Alcazar]

Sorry to bother you guys, but can someone tell me how the K5II(s) is performing in this regard? ie
- Time from shot to image appearing on back screen
- Time from shot to write light going out

Thanks,
Micha

[dosdan]

Sorry to bother you guys, but can someone tell me how the K5II(s) is performing in this regard? ie

I don't have a K5II, only a K5. But I don't think the Mk II version was improved in this area.

K-5, raw, flash shot of a part of my garage, no corrections (unsupported lens). I used a K-01 shooting 30p video (33ms/frame) for timing.

3 Sandisk SD cards ( card speed MBytes/s measured in h2testw in a USB3 ext card reader) :

Card	Write	Read
Ultra 16GB	9.04	43.4
Ultra 32GB	10.6	43.6
Extreme Pro 64GB	81.3	83.8

Timings (secs after the flash fired)

Card	Review image apears	Write light goes out
U16	1.63s	5.20s
U32	1.67s	5.83s
EP64	1.63s	3.27s

I can't explain why the U32, which has a faster writing speed than the U16, took longer to finish writing. I repeated the full test and confirmed that this was indeed the case.

Based on the difference in write speed of the the 2 Ultras and the Extreme Pro, I would think that the "write light goes out" value is limited by the K5 card writing hardware to a h2testw-measured maximum of about 14-18 MBytes/sec. So any faster SD writing speed in a K5 is probably wasted, as regards any further improvement in shot buffer clearance performance.

Dan.

---

Last edited by dosdan; 01-22-2014 at 05:23 AM.

[rawr]

A somewhat related K-3 continuous performance speed-test article:

Digital camera long-term real-time reports to be worried about: PENTAX K-3 [4th] - Digital Camera Watch

[Alcazar]

Card	Review image apears	Write light goes out
U16	1.63s	5.20s
U32	1.67s	5.83s
EP64	1.63s	3.27s

Thank you so much, again, Dan!
I was thinking maybe the K5 is the better camera for me, since i don't really need 24MP. If the results from your test would have been similar to the K3, that would have been a good choice, but thanks to you i know that when i upgrade, it will be the K3.

Alright, i'm off to make some Caipirinhas. Cheers :-)
Micha

[jiimaa]

3 Sandisk SD cards ( card speed MBytes/s measured in h2testw in a USB3 ext card reader) :

Card	Write	Read
Ultra 16GB	9.04	43.4
Ultra 32GB	10.6	43.6
Extreme Pro 64GB	81.3	83.8

[...]

Based on the difference in write speed of the the 2 Ultras and the Extreme Pro, I would think that the "write light goes out" value is limited by the K5 card writing hardware to a h2testw-measured maximum of about 14-18 MBytes/sec. So any faster SD writing speed in a K5 is probably wasted, as regards any further improvement in shot buffer clearance performance.

I have an USB 3.0 Lexar reader, on an USB 3.0 card, on a Mac Pro, and my Extreme Pro 16 GB seems to perform worse than your 64.



I tested with diglloyd disktester fill-volume.

Still, I have the K-5II and I think your conclusion is right - the K-5 is such a snail in writing to card, I see no difference between these two cards when shooting. As both cards are very close to each other in read speed when the images are dumped to the computer, in practice there is no difference at all between the Ultimate and Extreme Pro cards when using them with K-5, although the Extreme Pro performs noticeably better in write, and slightly better in read, when tested on a setup fast enough for the card to be the slowest component. The K-5's write is way, way less than the worst performer of these two cards. The Transcend used here is not the gray ultra, but the red ultimate (rated 90 MB/s / 45 MB/s) version, and the SanDisk is the gold top extreme pro (95/90) version.

[derekkite]

Get an extreme pro. The K3 will work much better. The K5 became tolerable with it. A friend who has a D7100 put one in and the buffer issue became tolerable.

I bought a 32 g last january and it still works fine. A year of heavy use. They are not inexpensive, but I've ceased buying one cheap one after another, putting up with dodgy performance.