Filesize of modern DSLR killing?

I forgot to mention that 1/4 of the raw file size is just random noise that is filling up the last 2 bits. The sensors are not able to provide real data for those bits so you end up with non compressible random data instead.

The 14 bit mode is just an annoying marketing gimmick. I´d like to have a custom function to use 12 bit files instead. Nikon does offer this choice.
So with 12 bit files you end up with 4 bits, or TWICE, the non compressible random data!

Why would you possibly want that?
 
Most people these days have two to three 32 GB CF cards and upgraded their hard drives to a minimum of 1 Terabyte. Yes, I agree if you want better lenses and your 30D is working great....you should keep you camera but if you would like to upgrade your camera, the kit lens with the 7D is not a bad lens. It's your choice in the end.
 
I picked up an old manual film camera a while ago that I play around with once in a while. That is one of the biggest contributing factors to me improving my photography. I take alot less pictures with my digital and get more keepers because I learned to think about each shot alot more.
 
I have been thinking that, if one has a small laptop (netbook?), one could easily connect 2 USB drives of 1 Terabyte each, for less than $300...(one for primary storage, one for backup) that should go a long way... let me see 1000 Gigs / 25 Mb... not even sure how to calculate such a high number...LOL.

In today's world, one tends to think in Gigs, not Megs. So a 25 Meg RAW file is a pretty small file. Newer laptops come with 500 Gigs, for less than $600.
My colleague bought a new 7D 2 weeks ago and showed the new pictures. They look great. Even the high ISO"s he had of pictures of fishes in an aquarium. But...... every single JPG was approx 8 Mb so just one day in a Zoo had a result of 2.9Gb of pictures.....

That's a lot... From my last holiday through alaska and yukon, I had approx 1600 pictures from my 30D. Mainly in RAW format. With a 7D that would be:

Jpg: 1600 * approx 8Mb = 12800Mb
RAW 1600 * 25 Mb = 40000Mb!!

Wow.... that could be a problem and a bigger one of you want to use the movie option too. My old 60Gb imagetank would easily be filled.....

I think I still keep my 30D for a while and just buy better lenses
--
Jerome Boyer
http://www.pbase.com/jboyer
Equipment in profile
 
I picked up an old manual film camera a while ago that I play around with once in a while. That is one of the biggest contributing factors to me improving my photography. I take alot less pictures with my digital and get more keepers because I learned to think about each shot alot more.
Yes, the brain actually learns more when it's not in spray 'n' pray mode. I can sometimes go a couple of months on a 36 exposure roll of b+w film, because for me good b+w subjects are hard to come by. What the digital shooter needs to consider (not sports or kids in action) is "should I really take this photo or will I just end up deleting it later tonight on my computer?" That's sometimes I think about when I use my 5D instead of my film SLRs.

Sometimes it's fun to use a full manual film SLR, that operates on nothing but a watch battery where you are responsible for everything! :)
 
I forgot to mention that 1/4 of the raw file size is just random noise that is filling up the last 2 bits. The sensors are not able to provide real data for those bits so you end up with non compressible random data instead.

The 14 bit mode is just an annoying marketing gimmick. I´d like to have a custom function to use 12 bit files instead. Nikon does offer this choice.
So with 12 bit files you end up with 4 bits, or TWICE, the non compressible random data!

Why would you possibly want that?
I don´t understand what you want to say. If the sensor delivers 12 bit of image data the 12 bit raw format would be perfect to store them without having to fill out unneeded extra bits with random numbers.

I hope that the sensor offers enough data to fill the first 12 bits with actual image data. But I would not be terribly suprised if the 12th bit has just random noise too.

The point I was wanting to make is that the extra 2 bits just waste space on your raw buffer and flash card and hardisk and offer no quality advantage at all. And because random numbers are not compressible to store those to extra bits occupies more than their fair share of file space.

I did experiment to convert raw files with less bits in them. A 9 bit raw file came very close to the actual image while an 8 bit raw file produced useless junk because the raw format does not use the gamma curve that lets the JPG format and computer monitor get away with using just 8 bits.

I am pretty sure that not the Canon engineers but the marketing department introduced the 14 bit raw fornat of the latest cameras.

The engineers should have made the actual number of bits a custom function instead of only offering 14 bit storage.
 
no text
 
Not a JBOD array, just a bunch of disks, and make backups (a good backup strategy trumps RAID almost all the time. Servers should use both.)

Linux software RAID is pretty good from what I've read, but let's face it, most of us aren't willing to go there.
My experience with inexpensive RAID cards (
Never had an issue with 'real' RAID cards.
--
Primary equipment:

7d, EF-S 10-22, 17-55 f/2.8 IS, EF 50 f/1.4, 70-200 f/4.0L IS, 70-200 f/2.8L, 580EX, and an HP B9180
 
Not a JBOD array, just a bunch of disks, and make backups (a good backup strategy trumps RAID almost all the time. Servers should use both.)
Heh, right. My RAID1 is occasionally rsync -ed to USB drive of the same size. That's a good strategy for a data paranoic ;-)
Linux software RAID is pretty good from what I've read,
It is. Currently, I have one RAID1 array and one RAID5 array (less redundancy than mirror, but more economical use of the drives).
but let's face it, most of us aren't willing to go there.
It's all question whether there's someone who could set-up a "headless" Linux server with RAID and Samba on the network to act as a storage space for a Windows machine...

Personally, I work on Linux directly, as I have ideological problem with Windows. ;-) The only Windows I ever use is Vista that came with my notebook, and only when I need to set up Personal functions in my 1D.
My experience with inexpensive RAID cards (
They're called "fakeraid", because all they are is just a special SATA controller with some small additional circuitry/BIOS and windows driver which creates the RAID device.
At work, I've had 1 machine have severe data corruption on it's RAID1 array, and another have frequent blue-screen errors. The blue-screen machine was able to be used again by breaking the array, so no loss there, but not exactly confidence inspiring. The other machine took about 8 man-hours to get back up and running correctly. Both were SiL cards. Take that for what you will (I've had little experience with other low-end RAID systems.) I've only seen a few hard drives fail out of a few hundred (and most if not all were within the first month or after several years.)
Never had an issue with 'real' RAID cards.
--
Primary equipment:

7d, EF-S 10-22, 17-55 f/2.8 IS, EF 50 f/1.4, 70-200 f/4.0L IS, 70-200 f/2.8L, 580EX, and an HP B9180
--
Cheers,
Martin

 
Years ago I used to fix SMD's (removable pack disc drives) and it took two of us to lift a 40mb one. Later on we got the first "Winchester" drives in. I think they were 10mb and about half that had to be bad tracked out so they probably ended up as 5mb. Still later one of my customers ran a multi million £ business on an Amstrad pc with a 20mb drive. Yesterday I saw an add for a 32gb CF card.

How times have changed...
 
I don´t understand what you want to say. If the sensor delivers 12 bit of image data the 12 bit raw format would be perfect to store them without having to fill out unneeded extra bits with random numbers.

I hope that the sensor offers enough data to fill the first 12 bits with actual image data. But I would not be terribly suprised if the 12th bit has just random noise too.
So, correct me if I'm misunderstanding you, but are you saying that you suspect Canon's sensor only delivers 11 bits of data per site?
The point I was wanting to make is that the extra 2 bits just waste space on your raw buffer and flash card and hardisk and offer no quality advantage at all. And because random numbers are not compressible to store those to extra bits occupies more than their fair share of file space.

I did experiment to convert raw files with less bits in them. A 9 bit raw file came very close to the actual image while an 8 bit raw file produced useless junk because the raw format does not use the gamma curve that lets the JPG format and computer monitor get away with using just 8 bits.
You say a 9bit raw 'came close' to a 12bit/14bit raw file. What method did you use to determine that? I'm left wondering if you understand what advantages a greater bit depth offers.
I am pretty sure that not the Canon engineers but the marketing department introduced the 14 bit raw fornat of the latest cameras.
I'm pretty sure this is a wild-arsed guess your making here. :-)
The engineers should have made the actual number of bits a custom function instead of only offering 14 bit storage.
Perhaps, though there are other things that I would prefer ahead of a CF I'd personally never use ;-)
 
DSLR dynamic range measurements end up with about 9 or 10 stops of dynamic range. In a raw file you need one bit to encode each stop of dynamic range. Having 2 bits extra is probably a good idea. But having 4 bits extra is just a waste of storage and raw buffer space. Some people made a hobby out of analyzing DSLR sensors and RAW files. One of them wrote a program that allows to shift the bits in a raw file. So you can discard the least significant bits. I used this program to turn a 14 bit raw file into a 12 9 and 8 bit raw file. I could see no difference between the 12 and 14 bit version. The 9 bit version looked different but came very close to the look of the 14 bit version. The 8 bit version looked totally wrong and was clearly insufficient to encode the actual raw data.
I don´t understand what you want to say. If the sensor delivers 12 bit of image data the 12 bit raw format would be perfect to store them without having to fill out unneeded extra bits with random numbers.

I hope that the sensor offers enough data to fill the first 12 bits with actual image data. But I would not be terribly suprised if the 12th bit has just random noise too.
So, correct me if I'm misunderstanding you, but are you saying that you suspect Canon's sensor only delivers 11 bits of data per site?
The point I was wanting to make is that the extra 2 bits just waste space on your raw buffer and flash card and hardisk and offer no quality advantage at all. And because random numbers are not compressible to store those to extra bits occupies more than their fair share of file space.

I did experiment to convert raw files with less bits in them. A 9 bit raw file came very close to the actual image while an 8 bit raw file produced useless junk because the raw format does not use the gamma curve that lets the JPG format and computer monitor get away with using just 8 bits.
You say a 9bit raw 'came close' to a 12bit/14bit raw file. What method did you use to determine that? I'm left wondering if you understand what advantages a greater bit depth offers.
I am pretty sure that not the Canon engineers but the marketing department introduced the 14 bit raw fornat of the latest cameras.
I'm pretty sure this is a wild-arsed guess your making here. :-)
The engineers should have made the actual number of bits a custom function instead of only offering 14 bit storage.
Perhaps, though there are other things that I would prefer ahead of a CF I'd personally never use ;-)
 
DSLR dynamic range measurements end up with about 9 or 10 stops of dynamic range. In a raw file you need one bit to encode each stop of dynamic range.

Having 2 bits extra is probably a good idea. But having 4 bits extra is just a waste of storage and raw buffer space. Some people made a hobby out of analyzing DSLR sensors and RAW files. One of them wrote a program that allows to shift the bits in a raw file. So you can discard the least significant bits. I used this program to turn a 14 bit raw file into a 12 9 and 8 bit raw file. I could see no difference between the 12 and 14 bit version. The 9 bit version looked different but came very close to the look of the 14 bit version. The 8 bit version looked totally wrong and was clearly insufficient to encode the actual raw data.
I think your missing the point of what 14bit brings to the table. It's more about increasing flexibility in post and when moving between colour spaces.

Have a read of this :-
http://www.earthboundlight.com/phototips/nikon-d300-d3-14-bit-versus-12-bit.html

It explains the advantages of 14bit over 12bit. They aren't huge, but they are there.

I'm quite happy to provide the extra space it takes to record in 14bits. Storage space is very cheap and I find this whole 'file-size is too big' argument ridiculous. It's staggering to me that someone would spend so much money on a 7D and good glass and then complain about the (comparatively insignificant) cost of storage!
 

Keyboard shortcuts

Back
Top