allnak wrote:
"As for the caps in the magazine...Japanese generally don't pixel peep - they seem more impressed with the subject, and therefore quite a lot of shots in the media are little missfocused ever now and again. Wouldn't necessarily mean that AF is out of order, would it?"
Scuse me for the frankness, but - you're talking out of your @ss. Not only this quote - everything you've said so far. But, hey, you're doing a good job of it
Sorry mate, I don't feed the trolls... Happy shooting..
garyvot wrote:
Could you perhaps have fallen into the trap of comparing images at 100% from cameras with very different resolutions? If the 7D at 18Mp looks even close to an 8Mp camera at 100% (prior to NR), then that's a big win.
Another way to think about this: shoot the 7D in MRAW mode, then compare the files with the 30D. My guess is that 7D MRAW will show a clear noise advantage over the 30D, perhaps as much as a full stop, yet still offer higher resolution.
I always compare files resized in a similar resolution, that's the only way to do it reasonably. Anyway, the 7D is going to have clear advantage at lower ISO's, the extra megapixels help there. The M-RAW is an aspect that I am definitely interested in and will test it for sure (I am going to get a 7D even if the image quality isn't significantly better than previous bodies - the other aspects of the camera are good enough).
n0b0 wrote:
I just remember, the 30D RAW is only 12bit yes? That means there's a lot less information compared to 7D 14bit RAW.
Packing pixels more densely (to my understanding) reduces dynamic range at least at higher ISO's (increased noise vs signal). But the 14bits is a good thing and might still help give more exposure latitude.
arn_old wrote:
Packing pixels more densely (to my understanding) reduces dynamic range at least at higher ISO's (increased noise vs signal). But the 14bits is a good thing and might still help give more exposure latitude.
If that was an absolute truth, the 7D would've had worse high ISO performance than my 450D but that's not the case.
I'm not an expert on bit depth but I think it's safe to say that in this case, more is better.
I can't give you any specific source for that information... It just something that makes sense to me and I've seen reasoning along those lines many times over the years. If you want a link to some source, you can check for example this: http://www.sphoto.com/techinfo/dslrsensors/dslrsensors.htm
I'm sure you can google for more. Like I said "to my understandstanding", but I still could be wrong.
n0b0 wrote:
I'm not an expert on bit depth but I think it's safe to say that in this case, more is better.
Certainly more bit depth *is* better
But having all those extra bits availabe for data storage doesn't neccessarily mean that it can be all used effectively. There's much more to dynamic range than just the bit depth.
I thought that no one could actually prove that 14bit did anything with SLR camera type files? Was that old information or have I missed something? I seem to recall there not being any more information in those bits. I've never shot with a 14 bit camera though, just some vaguely remembered threads from a while back. Any updates very welcome.
Beni wrote:
I thought that no one could actually prove that 14bit did anything with SLR camera type files? Was that old information or have I missed something? I seem to recall there not being any more information in those bits. I've never shot with a 14 bit camera though, just some vaguely remembered threads from a while back. Any updates very welcome.
Well yes, the extra bits don't mean that there neccessarily is any extra information in them. Some one else can google a quote.
The noise component that arises from less bit depth, quantization noise, ideally has a standard deviation of 0.5 ADU / sqrt(2)
Say the black frame noise for the 7D at ISO 100 has a Std.Dev of 5.0 ADU.
The quantization noise is 0.35 ADU, and subtracted as RMS, we get noise level from other sources than quantization to be 4.99 ADU
If we replaced the A/D converter in the 7D with a 12-bit one, the quantization noise would be 1.4 (in 14-bit scale)
Adding that quantization noise to 4.99 as RMS, we get 5.19.
The difference between 5.0 and 5.19 is 0,05 stops, meaning that if the 7D had a (perfect) 12-bit converter, the DR would only be 0,05 stops worse.
A/D converters working at high speed often introduce more noise than the quantization itself, so the real-world difference is bigger.
If the A/D noise is say 1,5 ADU (a realistic value at these conversion speeds), a 12-bit converter would give a total noise level of 7.8, and we lost about 2/3 stops DR.
As long as we don't know exactly how good the converters are (we can't separate conversion noise from amp noise), we don't know the exact benefit of using 14 bits. What we do know, is that the (12-bit) 450D has about 0,5 stops less DR than the (14 bit) 40D.
brainiac has posted a link in this thread. You might want to check out that link.
arn_old wrote:
I can't give you any specific source for that information... It just something that makes sense to me and I've seen reasoning along those lines many times over the years. If you want a link to some source, you can check for example this: http://www.sphoto.com/techinfo/dslrsensors/dslrsensors.htm
I'm sure you can google for more. Like I said "to my understandstanding", but I still could be wrong.
alundeb wrote:
The noise component that arises from less bit depth, quantization noise, ideally has a standard deviation of 0.5 ADU / sqrt(2)
Say the black frame noise for the 7D at ISO 100 has a Std.Dev of 5.0 ADU.
The quantization noise is 0.35 ADU, and subtracted as RMS, we get noise level from other sources than quantization to be 4.99 ADU
If we replaced the A/D converter in the 7D with a 12-bit one, the quantization noise would be 1.4 (in 14-bit scale)
Adding that quantization noise to 4.99 as RMS, we get 5.19.
The difference between 5.0 and 5.19 is 0,05 stops, meaning that if the 7D had a (perfect) 12-bit converter, the DR would only be 0,05 stops worse.
A/D converters working at high speed often introduce more noise than the quantization itself, so the real-world difference is bigger.
If the A/D noise is say 1,5 ADU (a realistic value at these conversion speeds), a 12-bit converter would give a total noise level of 7.8, and we lost about 2/3 stops DR.
As long as we don't know exactly how good the converters are (we can't separate conversion noise from amp noise), we don't know the exact benefit of using 14 bits. What we do know, is that the (12-bit) 450D has about 0,5 stops less DR than the (14 bit) 40D.
from what I hear the 12bit ones especially (and even 14 bit ones somewhat) are not that great but some of the 16bit ones are very good. I'd almost bet we could make use of using a 16bit one in terms of getting the best ADC even if then it would be reasonable to pack it back down to 13 or 12bit afterwards.
Can the 7D wirelessly control 580EX Mk I? I know it can wirelessly control 580EX Mk II but I have not found any reference to the Mk I. Is it a master like all masters which can wirelessly control any flash with a slave capability?
arn_old wrote:
Certainly more bit depth *is* better
But having all those extra bits availabe for data storage doesn't neccessarily mean that it can be all used effectively. There's much more to dynamic range than just the bit depth.
There is no connection between bit depth and dynamic range whatsoever. At all.
The dynamic range tells you how high contrast a scene the camera can capture while retaining detail in the highlights and the shadows. Bit depth tells you how smoothly tonal data can be stored in the file. There is simply no connection whatsoever between bit depth and dynamic range.
Bit depth = smoothness of tone
Dynamic range = retention of highlight and shadow detail
Stop confusing them.
To put it more bluntly: it should be possible to build a 4 bit camera which has 50 stops of dynamic range. I don't know what you would use it for, but it shows that bit-depth and dynamic range are not related.
brainiac wrote:
There is no connection between bit depth and dynamic range whatsoever. At all.
The dynamic range tells you how high contrast a scene the camera can capture while retaining detail in the highlights and the shadows. Bit depth tells you how smoothly tonal data can be stored in the file. There is simply no connection whatsoever between bit depth and dynamic range.
Bit depth = smoothness of tone
Dynamic range = retention of highlight and shadow detail
Stop confusing them.
To put it more bluntly: it should be possible to build a 4 bit camera which has 50 stops of dynamic range. I don't know what you would use it for, but it shows that bit-depth and dynamic range are not related....Show more →
well that is kinda of silly you could just take a 5D2 and chop off the lower 10 bits and then use this scaling from black to saturation and have low, low read noise impact and say you now have tons of stops of DR?
I don't think it makes sense to say you have more stops of DR than you have bits.
brainiac wrote:
There is no connection between bit depth and dynamic range whatsoever. At all.
Until a camera designer chooses the encoding and transfer curve. Adding more bits is in practice always done without changing the transfer curve. So, in Canon DSLR's, bit depth affects the smallest distinguishable signal.
Yakim Peled wrote:
Can the 7D wirelessly control 580EX Mk I? I know it can wirelessly control 580EX Mk II but I have not found any reference to the Mk I. Is it a master like all masters which can wirelessly control any flash with a slave capability?
Happy shooting,
Yakim.
Yes, quite confidenrt it will. The Canon wireless E-TTL protocol predates these products and the there is not, to my knowledge, any change to the E-TTL system being introduced here. Should be no different than using an ST-E2 or an external speedlight in 'master mode'.
garyvot wrote:
Yes, quite confidenrt it will. The Canon wireless E-TTL protocol predates these products and the there is not, to my knowledge, any change to the E-TTL system being introduced here. Should be no different than using an ST-E2 or an external speedlight in 'master mode'.
With the 5DmkII, I could control some 580 mkI functions but not as many as I can with the mkII version. Honestly forgot what the differences were but I would expect the same with the 7D.
I'm a bit puzzled by the uprezzing and downrezzing of files when comparing the 5dII and the 7D.
My understanding, and correct me if I'm wrong, is that the 7D file is supposed to be uprezzed to 21mp in order to do a fair comparison with the 5DII. It seems to me that to compare the 5dII and the 7D we ought to be cropping the 5dII file so that it's the same phyisical size as the crop camera [i.e., by the appropriate 1.6 factor], and then uprezzing it to equal the density of the 7D.
Putting it differently, assume we have two cameras, one crop and one FF, that are made from the identical sensor material, such that the FF files, when cropped by the appropriate 1.6 factor, contained the same number of pixels as the crop sensor files. It seems that in this case there is no need to uprez or down-rez either one to examine them at the pixel level.
In the present case, since the 7D has much greater density than the 5DII it seems that the
5dI files are the ones to be adjusted rather than the 7D files.
Happy shooting..