danmitchell wrote:
Your equivocal language does little to counter the claims about noise. "Little effects" are still effects and your "NOT entail significantly more noise" concept does acknowledge the possibility of increased noise - though the question of its significance could be open to debate.
Yes - my point is that its contribution is insignificant, e.g. far less than the enormous stride made each time a new generation of sensor is developed, so it can effectively be ignored. That's why I think people should stop moaning about high pixel densities.
The crop above shows a 10 Mpixel and 21 Mpixel file from the same generation of camera. You could argue that the 21 megapixel camera is cheating because it has 30% more area. But if you look closely, you see that in the shadows, where it counts, the 1D3 has about 30% lumpier noise. Canon has reached a stage with its sensors where, as long as your purpose is to make a print, rather than compare individual pixels at conflicting magnifications, there is practically no noise penalty for higher pixel densities at high iso.
> The "in the image" noise point would be valid if the noise produced by individual photosites was somehow average with the noise of the surrounding photosites...
But it is. There is the resizing process. You must at least allow the 21 Mpixel file to be downrezzed to your wonderful low noise figure of 10 Mpixels. For example, I could resize my 21 megapixel file down to ten megapixels in order to compare it to a ten Mpixel file. That would average a way a lot of noise (and detail too). So if you're arguing for fewer megapixels, hey presto! you can have them. Start with 21, downrez to ten in photoshop, and you're happy. No extra noise. So please everyone stop moaning about high pixel densities. It makes no significant difference to noise, unless you make the error of comparing 100% crops at different absolute image sizes.
brainiac - you've got me stumped! Based on noise alone, I'd say they're both good enough for me - and both.. not too different from where I'd expect my 1Ds II to be at 3200.. or the 5D... a a guess, 1D III on the left, 1Ds III on the right - but on noise alone, it's a tough one.
The high pixel density is a good thing for the 50D. Macro photographers will be all over this one as it's currently the most dense sensor on the market. Couple this with Canon's fine 180mm f/3.5 macro or their MP-E 65 (which isn't very sharp IMO but it gives amazing results) and you've got a winner combination.
EvilZardoz wrote:
brainiac - you've got me stumped! Based on noise alone, I'd say they're both good enough for me - and both.. not too different from where I'd expect my 1Ds II to be at 3200.. or the 5D... a a guess, 1D III on the left, 1Ds III on the right - but on noise alone, it's a tough one.
Yes, it is, but the overall noise pattern of the 1Ds3 seems smoother to me, and in the darkest shadows the 1D3 is getting quite lumpy, whereas the 1Ds3 manages to retain a smoother, more continuous tone.
I should add a thank you to Chris Miller for these files. Chris prefers his 1D3 at high isos in low light. I replaced my 1D3 with a 1Ds3, so I haven't had a chance to shoot the cameras side by side, but my experience has been commensurate with these crops. I think the 1Ds3 is better at isos 3200 - 12800 than the 1D3 and D3.
That's why I say that the only benefit of fewer pixels these days is that you save on disc space.
brainiac wrote:
Yes, it is, but the overall noise pattern of the 1Ds3 seems smoother to me, and in the darkest shadows the 1D3 is getting quite lumpy, whereas the 1Ds3 manages to retain a smoother, more continuous tone.
I should add a thank you to Chris Miller for these files. Chris prefers his 1D3 at high isos in low light. I replaced my 1D3 with a 1Ds3, so I haven't had a chance to shoot the cameras side by side, but my experience has been commensurate with these crops. I think the 1Ds3 is better at isos 3200 - 12800 than the 1D3 and D3.
That's why I say that the only benefit of fewer pixels these days is that you save on disc space....Show more →
You're welcome b... I do prefer my 1D3 for low light shooting, but not so much because of the noise (or lack thereof) at 3200. I will admit that I was pleasantly surprised by the comparison I did between the two cameras, thanks to you. And I did learn through the process that pixel pitch and density have less of a role in the generation of noise than I had expected.
But the reason I much prefer the 1D3 over the 1Ds3 for low light work is twofold. First, the ability to shoot at 6400. This is huge. Candlelight? No problem... And even more relevant are the file sizes. If I'm shooting a low light event, it is a wedding, party,etc... And I may shoot 1000 frames and none of them will get printed larger than 13x19, while 99% of them are going to end up as 4x6 prints. There's simply no need for a 21mp file, and it is inefficient in terms of processing and delivery. 10mp is the perfect, most efficient resolution for wedding/event work. I also do a lot of commercial and landscape work, where I'm making gallery-sized prints, 2 page magazine spreads, etc, and this is where the 1Ds3 earns it's keep. Also, these types of images are shot at low ISOs, so the 3200 ability of the 1Ds3 is a side bonus at best. Once again, I'm not going to be making gallery-sized prints at 3200 anyways, so a 100% crop of a 1Ds3 file at 3200 simply isn't relevant to my work.
As I was typing this, I just remembered last weekend. My 1D3 was in for repair, so I took my 1Ds3 to shoot a MMA Cagefight. I was glad to know I did have the ability to shoot at high ISOs, as the lighting was very difficult. So thank you brainiac!
Ya know Chris, you are the classic example of those who are completely justified in buying the cutting edge since you need every last milligram of performance. Dolts like me are just fine with 2 generations ago
Of course for the trust fund babies who cares....they can get whatever they want.
Have you tried shooting sRaw with the 1Ds3 set to iso 3200 and a stop of under-exposure? It works quite well for me. At very high ISO's I seldom need detail beyond 5 megapixel.
What all this means is that the 50D is going to satisfy the claims being made about it, and those in the market for 15 megapixels and 12800 iso shouldn't hesitate on the grounds that high pixel density will increase noise significantly. mRaw sounds cool too.
brainiac - and the disk space argument is completely smashed by the TWO LEVELS of sRAW..... Two years ago I joked around at how sRAW would be a really really kickass feature but there would be no way to convince Canon to implement it. There are many times when I don't need the resolution but I still need to shoot in RAW.
Re: the noise, it's really personal preference. Initially I preferred the image to the left - but the 1Ds III seems a bit smoother. I'm really keen to see where the 50D's sensor tech goes in a 1D or 5D replacement
Have you tried shooting sRaw with the 1Ds3 set to iso 3200 and a stop of under-exposure? It works quite well for me. At very high ISO's I seldom need detail beyond 5 megapixel.
What all this means is that the 50D is going to satisfy the claims being made about it, and those in the market for 15 megapixels and 12800 iso shouldn't hesitate on the grounds that high pixel density will increase noise significantly. mRaw sounds cool too.
Thanks Richard. No, I haven't tried that, that's what I've got the 1D3 for... However if the 1Ds3 had an "mRAW" of about 10mp, and went to 6400, I might not have any use for the 1D3.
I too own both Mk IIIs and can testify that the noise is practically identical (but with the "way over 100%" edge going to the 1Ds). So brainiac is right even if you can't understand his reasoning or math because, after many 100's of MK III shots (and plenty of enlargement), I'm telling you it just looks that way.
Now Canon has introduced (characteristically in their prosumer line first) an improved gapless pixel design. That means that other direct comparisons are invalid since this is their only newly designed DSLR sensor (but only for long!).
I am hoping that the 50D is not a cynical advertising ploy and that, at least, some Canon lenses can "realize" better images at this unprecedented pixel density. I can see the circumstances and opportunity (slow moving or inanimate scene) to switch to a (relatively inexpensive) 50D instead of a Mk III to 'extend' the range of the user's longest super tele (if only because the Great Whites have some of the best resolution).
Exciting possibilities are there for new Mk IV's (and also copious depreciation of Mk IIIs??). CAUTION!! Comparative speculation follows: My 12 MP FF Nikon D3 has "gap" sensor technology and captures wonderful files at ISO 6400. So I would expect that a new Canon FF sensored body would have the potential of 15 MP and up to 12800 ISO, for similar noise performance. This may be the long awaited 5D Mk II?
Also 21Mp are too "needed" (useful) if you want to print very large (24x36" +) or crop substantially and I do-frequently.
EvilZardoz wrote:
Re: the noise, it's really personal preference. Initially I preferred the image to the left - but the 1Ds III seems a bit smoother.
I should point out that Chris had to use different focal lengths to get the same angle of view, so the 1Ds3 file has slightly less d.o.f..
My experience is that grain like the left grain will be more visible in print, whereas the grain on the right is slightly more likely to disappear at any given print size. It's just finer.
What interests me though, is that in the deep shadow the 1D3 file seems lumpier. That's where noise tends to be a problem, and that's why I prefer my 1Ds3 for shooting ISO 12800. I honestly believe it's due to the 1Ds3's larger sensor, which, despite the slightly lesser dynamic range, seems to handle very low light better.
The efficiency of the D3/D700 sensor is only as good as the one on the 1Ds3. No doubts about that.
However, if Canon's claims are indeed true, we'll see at least 1 stop of advantage over the competition. I'm eagerly awaiting the arrival of the 50D... and of course, the 5D replacement (whatever it's gonna be called).
thw2 wrote:
The efficiency of the D3/D700 sensor is only as good as the one on the 1Ds3. No doubts about that.
I have doubts about that. Below is a 1Ds3 crop and a D3 crop AT THE SAME MAGNIFICATION, both at iso 3200. The Nikon file seems to have had some kind of chroma blurring NR, which is what the camera seems to do, but the 1Ds3 performs better than the D3 even though it has not had any NR. That suggests to me that Canon's 21 megapixel sensor is less noisy than Nikon's 12 megapixel sensor: http://cyberphotographer.com/1ds3/1Ds3v1D3.jpg
Nikon is of course 'in the lead' as everyone keeps saying, but Canon seems to make better sensors.
Before anyone questions how my opinion can be so at odds with all of the wonderful skilful online reviewers who spread so much wisdom, the answer is simple: find me a review where the reviewer compares crops at the same magnification. Practically all of them are making a basic, and very misleading error. Now ask how many normal people are carefully comparing the performance of the two cameras without having bought one or the other. Practically none. That is how internet mythology develops.
My experience is that the Nikon D3 looks great on people (where you only want limited detail for pores and pimples). Nikon isn't afraid to use aggressive NR so I routinely shoot my dancers at ISO 6400 and use little or no NR on them (but often a lot on OOF BGs )
Even Nikon loving Ken Rockwell (gasp! don't flame me now) is up front in saying, in his tests, the (obsolete) 5D has better sharpness/detail than the D3. So it makes real good sense that the 21Mp 1Ds should do an even better job on the veins in tree leaves.
SoundHound wrote:
My experience is that the Nikon D3 looks great on people (where you only want limited detail for pores and pimples). Nikon isn't afraid to use aggressive NR so I routinely shoot my dancers at ISO 6400 and use little or no NR on them (but often a lot on OOF BGs )
Even Nikon loving Ken Rockwell (gasp! don't flame me now) is up front in saying, in his tests, the (obsolete) 5D has better sharpness/detail than the D3. So it makes real good sense that the 21Mp 1Ds should do an even better job on the veins in tree leaves. ...Show more →
Soundhound,
Have you ever compared the 1D3s with the D3 with NR off on both? I am just curious to know about the raw shadow noise.
brainiac wrote:
That's why I say that the only benefit of fewer pixels these days is that you save on disc space.
I am fully prepared to except this. Just for my curiosity why do point-n-shoot cams look so bad? It is poor sensor tech? Everyone online was stating it was all the megapixels.
pcunite wrote:
I am fully prepared to except this. Just for my curiosity why do point-n-shoot cams look so bad? It is poor sensor tech? Everyone online was stating it was all the megapixels.
Sensor size.
However if the 1Ds3 had an "mRAW" of about 10mp, and went to 6400, I might not have any use for the 1D3.