Why buffer slower in single mode?

[pelagia]

Advertisement (gone after registration)

When I shoot in continuous mode (CM), I get 7 to 8 shots in a row before camera starts to write and slows down. But when I shoot in single mode (SM), I hardly can shoot more than 4 frames in a row. After that it slows down considerably (displays the warning "Attention! Data transfer"). I tried to cheat the camera ans shoot in CM by 1 or 2 frames, but it did not give out 8 frames in a row as I hoped. It slowed down after 4 frames.

 

What is the reason? Does the camera start to write to the memory card, as soon as I release the button? Is it so that it writes to buffer as long as the button is pressed and starts writing to the card when the button is released (not when the buffer is full)? Can there be any improvement in new firmware?

 

Any suggestions for improving shooting speed SM, as fast as in CM?

 

Tom

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[chris_tribble]

One suggestion is not to worry? My own feeling is that it's an M and burst mode isn't really what it's all about...

 

Otherwise check the card you're using against the Leica FAQ and choose only cards with the fastest data transfer...

[jaapv]

Well the obvious solution is to switch to C when you go machine-gunning I cannot see any reason to shoot that way when in single mode. Whatever happened to the decisive moment?

[wparsonsgisnet]

Tom, I get 7-8 shots in the buffer before the full message -- using dng only.

 

You musts be shooting with dng + one of the jpeg settings. The jpeg takes some time to compute and slows the camera down.

[pelagia]

Bill, thanks a lot for the essential answer. Yes it worked: pure DNG is faster! And I also noticed that DNG+18MP JPG is faster than DNG+1MP JPG (?!). This has probably to do with processing time of JPG.

 

Thank you others, too. But please notice I did not talk about machine gunning and about burst shooting. I wanted the M9 it to be "the M", fit for the moment in single mode. However manually cocked M6 is faster than M9 in single mode.

 

Tom

[wparsonsgisnet]

Tom, I have been complaining about the too-small buffer size since I got my M8 at the beginning of time.

 

I shoot a lot of dance and *constantly* fill the buffer. After the requisite 7-8 pix, it takes a long time for the buffer to clear. I can, of course, shoot ONE MORE pic every 4 seconds, but that's no way to run a railroad, as they say.

 

To gain max speed with the M8/M9, I suggest you also turn off the review.

 

Of course, the images are so wonderful that a little pain must be what the doctor ordered.

[35mmSummicron]

Advertisement (gone after registration)

I am a fashion photographer and uses an M9 to shoot, and run into the buffer all the time.

 

It is frustrating, but I knew what I was getting into as I've shot with the M8 for 3 years and tested the M9 before acquiring my own.

 

I only shoot uncompressed DNG @ 160 ISO on continuous which generally yields 7-8 frames. You will also find that as you increase ISO, it takes longer to write (due to noise reduction time).

At 160 ISO it takes 26-27sec to completely clear a full buffer of uncompressed DNG's only.

 

shooting compressed DNG only will take about 18 sec to completely clear.

 

The M8 will shoot 12 frames DNG only clearing in about 25sec.

 

Hope this helps.

 

andy

[wparsonsgisnet]

Andy, nice summary, I see the same thing. In the M9 we need to use a Sandisk Ultra II to get these speeds.

 

By my calculation, the buffer in the M9 is the same size as that in the M8. That is, 12 pix at 10mb is approx = to 7 at 18mb.

 

Gee, they listened to me when I asked for a Fast Wide Prime lens -- in face they made 2 of 'em! -- but they weren't listening with regard to the buffer.

 

And memory is so cheap.

[ho_co]

This doesn't seem like rocket science to me.

 

I think Stefan seemed apologetic or even embarrassed in the Reichmann video when he said the M9 actually had a bigger buffer than the M8, but that it didn't seem to offer any advantage because the files were bigger.

 

Didn't Canon offer a modification for one of their cameras that increased buffer size? Something along the lines of, those made after a certain date had a larger buffer, and those made before that date could have it retrofitted at the customer's expense?

 

Couldn't Leica simply offer a bigger memory for those of us who want it? Is there something hard-wired in the camera's processing that precludes a larger buffer?

[zlatkob]

My own feeling is that it's an M and burst mode isn't really what it's all about...

 

Whatever happened to the decisive moment?

 

Excuses ... excuses. If it's an M, it should shoot like an M. An M film camera could shoot for more than 4 or 8 frames without getting stuck. And with the Motor Winder M, it could shoot 3 frames per second for 36 frames:

14408 Leica Motor-Winder M for M7, M6, M4P & M4-2

So let's please not re-define M photography as a slow photography just because that's what we have with digital.

 

And one thing I've noted from seeing contact sheets of some decisive moment photographers is that when the moment called for it, they would shoot a lot and quickly.

[35mmSummicron]

The buffer on the M9 compared to M8/M8.2 is in fact smaller (7-8 frames vs 12 respectively)

 

However on M9, 7 frames * 36.4mb per uncompressed DNG = 254.8mb

on M8, 12 frames * 10mb per DNG = 120mb

 

The M9 is pushing more than twice the amount of data at roughly the same speed....

 

I think the biggest problem is that the uncompressed DNG's don't have any sort of data/lossless compression. 36.4mb DNG is comparable in size to a data/lossless Phaseone P30 IIQRL back raw file (M9 18mp vs P30 30mp).

[ho_co]

Thanks for running the numbers, Andy!

 

If I understand correctly, the M9's buffer is larger, right?

 

...M9, 7 frames * 36.4mb per uncompressed DNG = 254.8mb

on M8, 12 frames * 10mb per DNG = 120mb...

 

 

If I'm not mistaken, that would mean that the M9's buffer is more than twice the size of the M8's, but since the files are larger, it takes fewer files to fill it. That would also explain what I take for embarrassment on Stefan's part in the video.

 

My question still stands: What stands in the way of installing a larger buffer?

[35mmSummicron]

"buffer" usually refers to the number of images you can take in rapid sequence without delay. So no the M9's buffer is up to 8 frames before the camera bogs down. The M8 in comparison can take 12 frames.

 

I brought up the numbers in my previous post to point out that the M9 has to process A LOT of information.

[ho_co]

Okay, I stand corrected. Actually, I've never heard 'buffer' used that way. Apple/Oxford dictionary just says

a temporary memory area or queue used when transferring data between devices or programs operating at different speeds.

 

Seems illogical to me to think of a buffer changing its size depending on the data it's holding.

 

So in your terms, Andy, the camera designers would say, "Let's give this camera a 12-frame DNG buffer, a 48-frame compressed JPG buffer" instead of just "Let's give the camera a 250mb buffer"? Of course, if you frame it that way, you get the benefit of specifying your intended JPG compression at the same time.

 

I haven't re-checked the Reichmann interview, but as I said, I don't think that's the way Stefan uses the term.

 

Thanks for the heads-up!

[k-hawinkler]

Didn't Canon offer a modification for one of their cameras that increased buffer size?

 

Howard,

 

I remember Nikon did for the D3, cost is $500.00 plus your state tax plus shipping cost.

Here are the details: D3 Digital-SLR Camera Buffer Memory Expansion Service

 

K-H.

[ho_co]

Thanks, K-H. I was sure someone had done that.

 

I just wonder if there's something that prevents an upgrade to the M9. Is a bigger memory circuit physically too big for the M body? Does it require a different set of connections? Would it require much more energy, or a great many more process cycles, or a different addressing scheme?

 

Maybe they'll offer an M9.2 upgrade path just before the M10 comes out.

 

I'm sure there's a good reason that Leica didn't enlarge the storage space of the M9 buffer more than they did. Stefan seemed aware that it might be seen as a shortcoming, so it's clear that they considered the alternatives.

 

And there's no one like Leica for exhausting all alternative paths before committing to one.

[pelagia]

Thank you everybody for your comments, but the buffer size was not exactly my point in the initial question. Please let me re-iterate.

 

In continuous mode I get 7-8 frames before the buffer fills. But in single mode I get only about 4 frames! Do you experience the same? Looks like after releasing the button, writing to buffer stops and (slower) writing to the memory card starts.

 

Tom

 

P.S. And I liked this by Zlatko:

...seeing contact sheets of some decisive moment photographers is that when the moment called for it, they would shoot a lot and quickly.

[mjh]

In continuous mode I get 7-8 frames before the buffer fills. But in single mode I get only about 4 frames! Do you experience the same? Looks like after releasing the button, writing to buffer stops and (slower) writing to the memory card starts.

Yes, C isn’t just a continuous fire option; internal processing is different as well. This is true for most digital cameras.

 

I think that in C mode the M9 stores raw data in the buffer, postponing JPEG creation (if JPEG or DNG+JPEG is selected) until the time when either the button is released or the buffer is full. In S mode, JPEGs are created immediately which takes more RAM.

 

Edit: So much for theory; in practice it turned out that I could shoot 7 frames in either mode (DNG+JPEG fine).

[Advertisement]

Hallo,

Look here: Leica M

[Shootist]

I say check your settings. It really sounds like the camera has been switched to DNG+JPG which will slow the camera down and fill the buffer quicker.

[wattsy]

I just wonder if there's something that prevents an upgrade to the M9. Is a bigger memory circuit physically too big for the M body? Does it require a different set of connections? Would it require much more energy, or a great many more process cycles, or a different addressing scheme?

 

It might be penny pinching on Leica's part - trying to keep the M9 under a certain price point (a bit like with the lack of sapphire glass) - but I'd like to give them the benefit of the doubt that the buffer is limited for a hardware reason (size of chips, etc.). Either way, the buffer is small when you actually need it and, as others have said, you don't need to be machine gunning to fill the buffer. It is very frustrating to find yourself in a situation of intense 'action' where you have filled the buffer and something important is suddenly happening. This is not the first time that this problem has been quietly raised in this forum and it is annoying for the usual suspects to make quips about the decisive moment or that a "burst mode" is not what the M is about.