I mean it is in traditional math, but it isn't in binary. In binary math a 13 bit number can be twice what the 12 bit is, and a 14 bit can be 4 times what the 12 bit is.
14 bit/12 bit = 4
This is only true when you look at the numeric range that can be encoded into N bits.
However, when we are packing bits, 14 bits takes 16.7% more storage space than 12 bits.
DynoMoHum wrote:
A sample RAW image I have from the 1DM3 has a file size of 13.8MB. This includes a 1.4MB JPEG that is embedded in that file. So, minus the embedded JPEG, that's 12.4MB
Does the 1DM3 really embed the JPEG into the RAW file, in addition to writing a separate file 350D does this but 30D does not. I would assume the 1DM3 would not be crippled like the 350D.
astrolucida wrote:
Does the 1DM3 really embed the JPEG into the RAW file, in addition to writing a separate file 350D does this but 30D does not. I would assume the 1DM3 would not be crippled like the 350D.
It is the preview image DynoMoHum is talking about. Every RAW file has a preview embedded that has a bit less resolution than the original (I'm guessing it's about half the W x H, don't have a RAW image on my laptop I'm writing this from). So this is different from selecting RAW+Jpg shooting.
Emile Gregoire wrote:
It is the preview image DynoMoHum is talking about. Every RAW file has a preview embedded that has a bit less resolution than the original (I'm guessing it's about half the W x H, don't have a RAW image on my laptop I'm writing this from). So this is different from selecting RAW+Jpg shooting.
I don't believe that the embedded preview JPEG inside the RAW takes that much space. In 30D, when I take a Small Normal JPEG, it is typically 200-400 kB. However, when I zoom in using the back LCD, the resolution is worse for RAW-only images than RAW+S normal JPEG. Hence, I would assume that the preview image takes even less space than 200-400 kB (in 30D, with 1DM3 it might be a bit more, but not 1.4 MB).
I don't own a 1DM3... or a Xti(anymore) I own a 300D... so what I know about the 1DM3 and even the Xti... comes from files I have on hand..
I have a 1DM3 image that I got off the internet somewhere, it was put up to demonstrate the ISO 3200 performance. All I have is the CR2 file... The record mode was CR2 only for this image (based on info that was in the CR2 file) There is indeed a JPEG embedded in this CR2 file.
Anyway... using a program called exiftool, you can see the size of the embeded JPEG in a RAW file... and CR2 files do indeed have JPEG data embedded in them. For the CR2 file, I have from a 1DM3, the embedded JPEG is 1936x1288, the size of a small JPEG for that camera.
When I looked at some Xti CR2 files I had from when I owned one... the embedded JPEG image is 3888 x 2592 pixels... the exact same size as a full sized JPEG for that camera. I was shooting RAW+JPEG with that camera, and I also know that the camera also wrote a seperate JPEG image along side the CR2 file... So, at least in the case of the Xti... when shooting RAW+JPEG, it seems to record a CR2 file and a JPEG file... and the CR2 file has a embedded JPEG that is pretty much exactly the same sized as the JPEG that got recorded separately. I assume if you shoot RAW only, it will only embed the smallest size JPEG in the CR2 file, but I can not confirm this, unless someone were to send me a photo that was shot as RAW only with a Xti.
astrolucida wrote:
Does the 1DM3 really embed the JPEG into the RAW file, in addition to writing a separate file 350D does this but 30D does not. I would assume the 1DM3 would not be crippled like the 350D.
DynoMoHum wrote:
Anyway... using a program called exiftool, you can see the size of the embeded JPEG in a RAW file... and CR2 files do indeed have JPEG data embedded in them. For the CR2 file, I have from a 1DM3, the embedded JPEG is 1936x1288, the size of a small JPEG for that camera.
So, at least in the case of the Xti... when shooting RAW+JPEG, it seems to record a CR2 file and a JPEG file... and the CR2 file has a embedded JPEG that is pretty much exactly the same sized as the JPEG that got recorded separately. I assume if you shoot RAW only, it will only embed the smallest size JPEG in the CR2 file, but I can not confirm this, unless someone were to send me a photo that was shot as RAW only with a Xti. ...Show more →
Just downloaded exitftool and made the following observations:
30D: the embedded JPEG size is 1728x1152, no matter what JPEG size is selected along the RAW image, including the situation where no RAW is produced.
350D/XT: when no separate JPEG is written, the embedded JPEG is 1536x1024 but when L JPEG is written (the only option available), the embedded JPEG is 3456x2304.
400D/XTi: the same as 350D/XTi, except that the dimensions are 1936x1288 and 3888x2592.
So it seems that the 30D can create upto two differently compressed and sized JPEGs, while the 350D/400D produces only one at the maximum. The RAW needs a preview JPEG, so if no separate JPEG is required, the camera can produce a small one but when a large one is required, the camera does not produce a small one separately.
This shows how the 350D/400D has been intentionally crippled - it has the software and capacity for producing a small JPEG along with the RAW but no menu operation to select that mode!
Interestingly enough, the exiftool seems to be able to extract the small JPEG directly from the RAW. This would help my workflow a lot, so I need to study how to use it in a mass operation.
Thanks for the information... it helps me considerably, as I have been trying to understand the behavior of Canon cameras with regard to RAW files and the size of the embedded JPEG that gets saved.
To make sure I'm clear... The 350D uses CR2 format correct?
I suspect the reason the 350D and 400D produce a CR2 file in it which has a embedded JPEG that is equal to the size of the JPEG that gets saved separately is due to processor speed or some hardware issue. I know Canon can and does intentionally cripple these sometimes, but I wonder if maybe it is easier and/or quicker for them to create a single size JPEG and then embed that size in the RAW, rather then trying to create two different images... This is only a guess... but then that still seems rather silly, since if you do shoot RAW+LJPEG, the camera has to write more data then it would if it just saved a small JPEG in the RAW file.
I don't know why Canon does not just provide a way to extract the JPEG from the CR2 file, rather then saving the Large JPEG twice... (on the 400D, this is 4MB of data for each of these JPEGs, the RAW data alone is only 8MB. They are saving 33% more data then they need to. More and more I'm glad I returned the 400D.
astrolucida wrote:
...
Interestingly enough, the exiftool seems to be able to extract the small JPEG directly from the RAW. This would help my workflow a lot, so I need to study how to use it in a mass operation.
BreezeBrowser Pro will extract the embedded jpg's directly.
For what it's worth, I can't see them putting the thumbwheel that far across: it just wouldn't work.
I would love these images to be right (that's a nice big screen). But they're not. The wheel are positioned different on the 2nd and the 4th picture.
Also on picture 1 it clearly says 40D but on picture 3 it says (not as clearly though) 20D.
Finally I don't think they are going to remove the print button.
DynoMoHum wrote:
Actually I'd like to see Canon take a stand and say that 10MP is really all that is necessary... and that going beyond that just for the sake of keeping up with the Nikons, etc... is not their game. But then maybe someone can make a better argument that 12MP is actually better... so far I have not seen any compelling reason to believe we need more then 10MP on a APS-C sensor...
Oh, and I do like mfurman's idea that the 40D should have at least one high precision AF point that works at high precision with a F4.0 lens...
...Show more →
I believe that you are right. With same technology we might very well be much 'happier' with a 10 MP sensor than a 12 MP one.
BUT does it matter really.
Canon need to sell cameraes and that would be their main motivation.
Obviously higher quality will be good. Short term (because some geeks will find out and tell about on the net) as well as long term (more satisfied customers - like: well, I don't know. But I can just see that this camera makes better pictures than that). On the other hand more megapixels will be selling more cameraes and probably the difference in IQ will be quite small actually (just like the benefit of 12 MP over 10 MP). More cameraes sold mean more (happy) customers who will tell about their positive experience.
So, - at the end of the day. Should Canon make a better camera or one that sells more
Sinope wrote:
Those images look real to me. If they are photoshopped it is top quality work.
Yep.
Compared with the mind-bogglingly mediocre Photoshop fakes published by the akihabaranews.com and fofor.se sites, the pics published by disitu look so real, that I'd be totally flabbergasted if they were not of the real EOS 40D.
Here are some slightly larger versions of these ultra-realistic pics I'm talking about:
The images shown, with the huge screen, were posted a while ago by someone clearly stating that he was just playing around with Photoshop, he even left the 20D name unaltered. The only 40D shots available are the ones posted recently in sites like Northlight Images.
Yes... I agree that Canon will likely do what is best to sell cameras... and I also sadly agree that many camera buyers don't or would not understand that 12MP is not necesasarlly any better then 10MP. I do however have some small hope that Canon actually would rather make a camera that has the best IQ possible, and if that is done with 10MP as opposed to 12MP, then I wish they would do the right thing and educate people about things, rather then just market the extra 2MP that didn't really do anything for IQ.
DaDane wrote:
I believe that you are right. With same technology we might very well be much 'happier' with a 10 MP sensor than a 12 MP one.
BUT does it matter really.
Canon need to sell cameraes and that would be their main motivation.
Obviously higher quality will be good. Short term (because some geeks will find out and tell about on the net) as well as long term (more satisfied customers - like: well, I don't know. But I can just see that this camera makes better pictures than that). On the other hand more megapixels will be selling more cameraes and probably the difference in IQ will be quite small actually (just like the benefit of 12 MP over 10 MP). More cameraes sold mean more (happy) customers who will tell about their positive experience.
So, - at the end of the day. Should Canon make a better camera or one that sells more...Show more →
Oh, by the way... I think maybe I figured out why the 1DM3 uses close to 13MB per RAW image doing it's 14bit color... Here's my theory...
Hexadecimal numbers consist of four bits each. 12 bits consist of 3 hexadecimal numbers... to add two more bits, you really need to go to 4 hexadecimal digits. So, the file size probably increases by 33%, not 17% as previously thought by myself.
Xti's RAW size is 9.8MB, multiply that by 1.33, you get 13MB, which is exactly the size Canon says the 1DM3 RAW files are...
So... if the 40D is 10MP and 14bit color data... it would likely end up with RAW files that are close to 13MB.
romesdsanthem wrote:
I have just read from the upcoming product list for a major electronics company will be starting to carry the 40D on sept 2nd.
Thanks for the info. but please cite sources and company so we can read it too!
If Canon makes the announcement on or about August 20th... I could easily believe September 2nd, in stores in the USA. They put the Xti out in about that much time from the announcement.
350D does this but 30D does not. I would assume the 1DM3 would not be crippled like the 350D.
