Canon 7D firmware, and the 650D

[BrianSA]

Well, however they have done it, this firmware update, including the magical buffer size growth, is now officially released. It's on DP Review today.

[Alvin]

verry interesting.....
they say it will be available from August only though

[Grant Atkinson]

Brian, thanks for the link...and I did say in an earlier post, the news release that was on Canon Rumours looked real to me...
Alvin, just a bit more patience :-)...not too long to go! I can't seem to post actual yellow 'smiley' faces, when I select them my post just goes white and hangs..but good news for sure :-)
cheers
Grant

[jab2]

What can I say?

Would like to know how they increased the burst depth, unless they were choking the buffer previously. There were rumors back in 2004 that the 20D's buffer were chocked, and a similar improvement were on the cards. That improvement never materialized, and Canon denied at the time that they underspec hardware in some models.

Canon, how about doing the same trick for the 1Ds Mk III?

[Grant Atkinson]

Hi Cobus
I am pretty sure that the camera was designed intentionally with the possibility of such an upgrade in its product life...so they just limited it initially. I really think we may start seeing slightly longer product life cycles with dslr bodies and this is probably part of planning for that. Not just Canon, but even Nikon, their D300s has been going for a while, and its sensor even longer...the D700 ran a good while..with sensor tech advances slowing slightly, it makes sense to keep bodies current for longer, specially at the higher and high mid-range of the market. Just my thoughts...and if it is the case, I for one would be quite happy, having to fork out every two years is getting harder and harder :-).
cheers
Grant

[Deon RacePics]

Cant wait to get my hands on that firmware version.... Im an early adopter, hope it aint gonna bit me. I have no camera body upgrades planned yet, am glad neither does Canon.

I thought the 350D was a long running model.

[BrianSA]

I was always surprised at the small buffer size in the 7D, only 2 secs with RAW. I would have expected at least 3 secs, preferably shooting RAW plus large JPG. I suspect it was designed with a large neough buffer for 3 secs shooting RAW, but there was a problem getting the code right initially.

It does make a 2nd hand 7D to replace at least one of my 40Ds even more attractive now.

[Grant Atkinson]

I don't shoot sport too much, but with wildlife I can maybe only recall hitting the buffer with my 7D once, and it clears pretty quickly. I only shoot raw, and have all the extra space and time wasting things turned off, like in-camera NR, auto-lighting, etc..to max out speed. If you are used to shooting with the 40D, then the 7D buffer as it currently is seems harder to hit..it does benefit from fast UDMA cards though. I used a 550D for a few days recently, nice camera, but hit the buffer almost immediately with jumping dolphins..
I used hit the buffer now and again with the 40D, and the 50D, and more often with the 1Dmk3...never with the 1Dmk4.
cheers
Grant

[BrianSA]

Thanks Grant, that's interesting. I have wanted to upgrade to 7Ds since they were launched and they are starting to appear at prices I can actually think about now. I've sent an email to ODP about their latest 2nd hand ad. Maybe I can upgrade one of my 40Ds.

[Deon RacePics]

All my issues regarding file naming with the 7D machine is resolved. On the 1D it aint a problem, but using three 7D bodies dumping images into the same folder provided its own set of problems... It is the end of that, and I cannot wait ...

So who is going to ask Roger to explain the deeper buffer now? I dont shoot in RAW, but with DLO option open to all the users of DPP, I have been thinking about it.

[Mike Fraser]

Some more info and some video clips from Canon:

http://www.usa.canon.com/cusa/consum...EOS7D_firmware

[Roger Machin]

Yowza ! - sheesh you guys really overthink things - don't you ?
The memory used for the buffer in the EOS cameras is brutally expensive - it is most certainly NOT the same as memory for PC's - that kind of memory is widely available and mass produced. The memory used in DSLR's is predominantly high capacity , small in size , very fast and low on power consumption and heat generation - therefore not easy or cheap to increase the capacity on a camera. It would also be very difficult to change the actual capacity of the memory somewhere in the life of a camera in the market - so many other factors have to be taken into consideration - most certainly not as easy as taking a 4GB Ram module out of a PC and plugging in an 8Gb module !.
Now - without revealing anything at all - the simple answer can be found in the press release for the firmware regarding the increase in burst :
"Thanks to the new firmware which adds powerful memory management algorithms taken from Canon’s flagship EOS-1 series, photographers will now be able to enjoy greater flexibility in continuous shooting, with the maximum RAW buffer now extended to a new high of 25 RAW files or 130 JPEGs. "
"Memory management algorithms" indeed. Most manufacturers throw in lovely terms like this to tell you what we've done without telling you anything at all ! ;-)

[Deon RacePics]

Not overthink, over analyse

[jab2]

Interesting comparison.

Chris, do you mean empty the buffer faster, as writing it faster to the buffer would not fit more data in it? If so, yes, it is a possibility, but hardware architecture has a much greater influence on data transfer rates than the transfer management code itself. OK, taken as fact that the code is efficient in the first place. Can't however see that Canon will use inefficient code at a time that they are under tremendous pressure form Nikon.

I would like to post a question. Why do we have a low limit on burst depth in the first place? Why does Canon, and for that matter Nikon, or all the other DSLR manufacturers not install enough memory to make it possible to have a burst depth of say 50 shots? I mean memory is relatively cheap, and looking at the size of memory sticks, space should not be a problem.

Well ,I think the memory we are familiar with in our PCs, and our memory cards are not the type that is suited for use in a DSLR as buffer memory. I suspect we are looking at something which still commands a relatively high price premium, and that physical size also plays a role. Just look at the design handstands the manufactures had to do to process HD video in a DSLR, processors many times faster, and up to three that works on the data. If the specs we desire were anyway reachable I'm pretty sure we would have seen it on the shelves already. It makes no sense for a manufacturer to release a product and within 2 1/2 years find out it can increase burst depth with 70% just by writing the code a bit different.

Also if this code is now ready to use in consumer products, why do we not see it's use in the new EOS-1DX, as there is no astronomical increase in burst depth between it and the 1DMk IV (30 frames RAW)? Taken that the EOS-1D MkIV had two DIGIC 4 processors and the 1DX have two DIGIC 5+ (17 faster that the DIGIC 4) plus a DIGIC 4 dedicated to autofocus. It is increase to 50 frames RAW (60%increase), as determined on a pre-production model tested by Ron Martinsen, but looking at the extra hardware in processors, this does seem such a small improvement that Canon do not even mentioned burst depth in their various press releases. I really had to dig deep to get the number above.

So, yes, there is a need for higher numbers, but I still cannot see this being increase after production by changing the firmware code.

Edit: I just find a number of 35 (38 with UDMA) for the EOS-1DX in an article on burst rate by Eduardo Angel on learn.usa.canon.com. That is only a 17% increase.

The memory used for the buffer in the EOS cameras is brutally expensive - it is most certainly NOT the same as memory for PC's - that kind of memory is widely available and mass produced. The memory used in DSLR's is predominantly high capacity , small in size , very fast and low on power consumption and heat generation - therefore not easy or cheap to increase the capacity on a camera. It would also be very difficult to change the actual capacity of the memory somewhere in the life of a camera in the market - so many other factors have to be taken into consideration - most certainly not as easy as taking a 4GB Ram module out of a PC and plugging in an 8Gb module !.
Now - without revealing anything at all - the simple answer can be found in the press release for the firmware regarding the increase in burst :
"Thanks to the new firmware which adds powerful memory management algorithms taken from Canon’s flagship EOS-1 series, photographers will now be able to enjoy greater flexibility in continuous shooting, with the maximum RAW buffer now extended to a new high of 25 RAW files or 130 JPEGs. "
"Memory management algorithms" indeed. Most manufacturers throw in lovely terms like this to tell you what we've done without telling you anything at all ! ;-)

Roger, did Canon really improve their code that much in 2 1/2 years to gain a 70% improve in burst depth, or did I overestimate their marketing ethics?

Is this improvement in memory management potentially be used in all bodies using DIGIC IV processors?

Can this improvement in code, with some modifications, potentially be used on the other DIGICs too?

[Nqina Dlamini]

Used 7D are suddenly an attractive proposition.......

[Roger Machin]

Roger, did Canon really improve their code that much in 2 1/2 years to gain a 70% improve in burst depth, or did I overestimate their marketing ethics? - it looks like it......

Is this improvement in memory management potentially be used in all bodies using DIGIC IV processors? - theoretically, yes........

Can this improvement in code, with some modifications, potentially be used on the other DIGICs too? - hmmm not sure......

[Mike Fraser]

Revised manual for the 7D incorporating the firmware 2 update available for download:

http://www.usa.canon.com/cusa/consum...uresAndManuals

[Phillip.Grobler]

It would be very interesting to now how they did it - my guess would be

The sensor is 18 MP x 14 bits = 252Mbits / 8 / 1024 / 1204 = 30.04 MB uncompressed x 15 shots = 450MB
So it the 7D probably has 512MB storage with 13% space left for safety or file overhead.

They would probably be very conservative - one lost frame in 1000 would be far worse for the brand than having a smaller buffer.

The output RAW files of the 7D are 17-27MB but this include a 1-2MB JPG image. So we can guess the compressed RAW data is 25MB
If they are able to compress the raw data on the fly instead of after buffering it you are left with 512 / 25 = 20.48 Shot buffer.
All that is left is that you need to be able to move 2.25 frames a second from the buffer to the CF and you have 25 frames

An easier way would be to just use 12 bits instead of 14 and then the uncompressed RAW files would 14% smaller - and who will notice

It might be all fiction but at least I can sleep tonight "knowing" how it works

[BrianSA]

I don't shoot sport too much, but with wildlife I can maybe only recall hitting the buffer with my 7D once, and it clears pretty quickly. I only shoot raw, and have all the extra space and time wasting things turned off, like in-camera NR, auto-lighting, etc..to max out speed. If you are used to shooting with the 40D, then the 7D buffer as it currently is seems harder to hit..it does benefit from fast UDMA cards though. I used a 550D for a few days recently, nice camera, but hit the buffer almost immediately with jumping dolphins..
I used hit the buffer now and again with the 40D, and the 50D, and more often with the 1Dmk3...never with the 1Dmk4.
cheers
Grant

Grant, you convinced me and ODP offered a good deal on a trade in, so I now have a 7D and have one 40D less. Picked it up this morning, met Sean too and sampled their excellent coffee, whilst ogling all the girls

Now plodding through the manual.

[Forgiss - Sean Nel]

HAHAHahahaha... good to meet you this morning Brian! Sorry that I couldn't be of more help!