People get carried away too much by this image-level noise ethos.
It's like - as long as a camera has lots and lots of pixels, noise is not a problem anymore.
You just have to downsize and all the noise is magically gone and the 7D, for example, is as good as the 1D3 (if not better).
If that's what people want to believe, that's fine with me. I fully respect all beliefs.
I wonder when the hype started. The technical analysis never claimed more than the 7D was narrowing the gap between the 40D and the 1D3. For image-level noise.
I agree with the image-level noise idea to some extent but think firstly that it often overestimates the amount of noise removed purely by downsizing an image and secondly that it ignores other factors such as colour. There is also an assumtion that the photosensor 'real-estate' is the same for a given size of sensor but in practice (at least for any given sensor generation) there is likely to be more area given over to non-light gathering circuitry when sensels are more densely packed.
There are advantages to high MP count but also some advantages to lower density, larger pixels. It's very difficult to compare cameras across denerations and sensor formats but I believe there is bound to be a tradeoff, why else would Nikon have stuck with a (newly designed) 12MP sensor in the D3s?
I agree with the image-level noise idea to some extent but think firstly that it often overestimates the amount of noise removed purely by downsizing an image and secondly that it ignores other factors such as colour.
With noise reduction through pixel averaging, noise decreases with the SQRT of the downsizing.
For example, to decrease noise 2x, a 40mp image needs to be downsized 4x to 10mp.
A 7D -> 40D resolution downsizing will result in ~sqrt(2) noise decrease, or ~1.4x.
Considering that that the 7D pixel-level noise is worse than on the 40D, a downsized 7D image to 40D dimensions will have practically the same noise as a full-size 40D image (if not worse?).
Therefore, there's no way that a downsized 7D image could even be close to 1D3 levels - unless the 1D3 has the same noise as the 40D
There is also an assumtion that the photosensor 'real-estate' is the same for a given size of sensor but in practice (at least for any given sensor generation) there is likely to be more area given over to non-light gathering circuitry when sensels are more densely packed.
Exactly !!!
This is something that the image-noise guys are never considering.
As pixels shrink, the on-sensor circuitry stays the same, so the photo-sensitive area shrinks more than the pixel decrease would imply.
There are advantages to high MP count but also some advantages to lower density, larger pixels. It's very difficult to compare cameras across denerations and sensor formats but I believe there is bound to be a tradeoff, why else would Nikon have stuck with a (newly designed) 12MP sensor in the D3s?
jorkata wrote:
A 7D -> 40D resolution downsizing will result in ~sqrt(2) noise decrease, or ~1.4x.
Considering that that the 7D pixel-level noise is worse than on the 40D, a downsized 7D image to 40D dimensions will have practically the same noise as a full-size 40D image.
So you are saying 0.9 x 1.4 equals 1.0?
This is something that the image-noise guys are never considering.
As pixels shrink, the on-sensor circuitry stays the same, so the photo-sensitive area shrinks more than the pixel decrease would imply.
No.
The 7D is worse than the 40D at the pixel level but I have not measured exactly how much worse.
If the 7D has 1.4x more noise at the pixel level, there's zero noise advantage from downsizing.
That's why it seems reasonable to assume that 7D downsized is no better than the 40D (and even if it's better, the difference would be negligible).
Ever heard of 100% coverage microlenses?
Yes. What about them?
The 50D has them too - and the the pixel level, where microlenses are applied (you see, you don't apply a single microlens to a 50D image), the 50D is worse than the 40D.
The lighting is obviously different in these two test shots (look at the color of the wall in the background) which simply invalidates any conclusion one would have otherwise been able to draw from the shots.
This kind of overhype of new cameras happens on all new releases
Remember the "50D is 1 stop better than 40D" farce?
Or the "5D2 is 2 stops better than the 5D" farce?
Nothing's really changed...it's just that some tend to be sucked in by the hype with new cameras.
Take the 1D3 for example. Many objective tests have shown this camera have be cleaner than the 1Ds3, 40D etc etc. Suddenly the 5D2 comes out and it is "2 stops better" even though the sensor is similar to the 1Ds3. Now we have a 7D that has come out which is suddenly cleaner than the 1D3 as well.
I don't have the results cause haven't updated my DPP yet, but just looking at the high ISO NR off Jpgs, when I tested the 7D against my 40D, the 7D had stronger shadow chroma and luminance noise than even the 40D, let alone the 1D3
alundeb wrote:
Ever heard of 100% coverage microlenses?
-- jorkata wrote:
Yes. What about them?
The 50D has them too - and the the pixel level, where microlenses are applied (you see, you don't apply a single microlens to a 50D image), the 50D is worse than the 40D.
lol I remember when Canon was making a bit announcement about them - turned out to be all hot air and no substance where the 50D was concerned imo
ChrisDM wrote:
The lighting is obviously different in these two test shots (look at the color of the wall in the background) which simply invalidates any conclusion one would have otherwise been able to draw from the shots.
Chris Miller
www.imagineimagery.com
I'm not sure Chris... I think you may want to look at the lighting across the frame (and the noise in different parts of the image) before you make this call. Again, these samples are posted at Imaging-Resource.com:
Unlike some of the other contributors to this thread (apparently), I don't have a horse in this race; I feel the 7D and the 1D3 are both great cameras. However, the samples I have downloaded (full images) don't show that the 1D3 is significantly cleaner than the 7D when equalized for output, at least based on in-camera JPEGs shot at ISO 1600.
I'm happy to be shown evidence to the contrary though; I think it would move this discussion forward in a constructive way.
You haven't measured, but you speak in absolute terms as if you had. I understand why people hide you.
Yes. What about them?
The 50D has them too - and the the pixel level, where microlenses are applied (you see, you don't apply a single microlens to a 50D image), the 50D is worse than the 40D.
When you confuse microlenses and pixel level / image level that way, I can't take you comment serious.
Do you know, the photon collection efficiency per unit area is actuelly a bit worse for the 50D than for the 40D. The use of gapless microlenses compensate for that, and make the SNR equal at image level again.
The photon collection effieincey per unit area is preliminarily measured to be higher for the 7D than for the 40D.
We will see about the SNR when DxO gets this one out.
Unlike some of the other contributors to this thread (apparently), I don't have a horse in this race; I feel the 7D and the 1D3 are both great cameras. However, the samples I have downloaded (full images) don't show that the 1D3 is significantly cleaner than the 7D when equalized for output, at least based on in-camera JPEGs shot at ISO 1600.
Thats just it. You dont equalize the image. The 7D makes an 18mp image...NOT a 10MP image. And at that size, it doesnt even come close to the 1D3. Still waiting on some examples.
If you were to make an 8x10 from both cameras, you wouldnt downzide the 7D for the 8x10 in total pixels, you just chop off a small portion of the image....leaving still roughly 15mp's. No one completely downsizes images for print...at least they shouldnt. Just clip the ends that need to go for the particular aspect ratio you are using.
To evaluate in print, it would be easier to just use the whole file. Use a print that doesnt need cropping like 8x12 or 12x18. Then use what each file produces, not downsampled images.
Yeah its done....cause once again, no one has posted these fantastic 7D images that are supposed to be just as good as the 1D3.
I put some 1D3 images up...ISO 1600 and 3200, but where are these 7D images. We wont expect anything unreasonable, surely the 7D can produce a good ISO 3200 shot, I mean, the 3yr old 1D3 did it and the 7D is supposed to be soooomuch better, so where....
Thats because there isnt any that can match the 1D3. And the ones that claim they can, never post the image and even further arent willing to caugh up a RAW file.
As the OP's question... I have neither but I am curious how a nearly 3yo camera compares to the latest and greatest APS-C crop from Canon. We've seen the 7d vs 5d Mark I comparison with surprising results. So how does the 1D3 do?
So far, I'd say from my examples(real world examples) its not even close. Why? CAUSE CANON DOESNT KNOW WHEN TO STOP WITH THE MP RACE. Its negating what improvements can be made with the new sensor. Its not like Nikon is pushing them in MP race. They keep everything they have at 12mp and lower, except for one big boy. Why? Cause packing more in negates the real sensor improvements they make. So, we have an 18mp APS-C sensor, thats noisy, and its seen by its detail being smeared by NR techiques. Hey, I dont mind a little crow sometimes, I'll eat some if someone can prove me wrong. Just be ready to caugh up that RAW file.
What good would it do to post completely different images than your 1d3? If the images appear "better", you'll just cry "not fair" because the light/subject/whatever isn't the same.
You've been fighting this campaign for weeks now, and you still don't get it. If you don't resize either the 7D or 1d3 image, you are not comparing apples to apples. Since you seem to be having SOOO much trouble understanding this, let me put it in another context. Let's do this thought experiment:
1. I will take a shot of anything with my 18mp 7d, and print it at 12x18.
2. Next, I will crop the center 50% of the shot, and print that at 12x18.
3. Now the question: will the 1st shot be better, worse, or the same as the second when viewed at the same distance? (obviously the shots will not be identical, but thiunk about the NOISE/overall IQ)
So far you have conceded that the shots you are comparing between the 1d3 and 7d have to be the same angle of view - I would agree with that. However, when you print both unresized images THE SAME SIZE, the DPI of the 7d shot will be higher and it will therefore have the POTENTIAL to have more detail. Obviously, if the 7D image is soft the extra rez won't matter. SO, the LENS absolutely matters as much as everything else. Even using the same lens, if you use it at different FL's you can potentially hit the weak spot in one shot and sweet spot for another. Naturally, DOF also comes in to play.
What good would it do to post completely different images than your 1d3? If the images appear "better", you'll just cry "not fair" because the light/subject/whatever isn't the same.
You've been fighting this campaign for weeks now, and you still don't get it. If you don't resize either the 7D or 1d3 image, you are not comparing apples to apples. Since you seem to be having SOOO much trouble understanding this, let me put it in another context. Let's do this thought experiment:
1. I will take a shot of anything with my 18mp 7d, and print it at 12x18.
2. Next, I will crop the center 50% of the shot, and print that at 12x18.
3. Now the question: will the 1st shot be better, worse, or the same as the second when viewed at the same distance? (obviously the shots will not be identical, but thiunk about the NOISE/overall IQ)
So far you have conceded that the shots you are comparing between the 1d3 and 7d have to be the same angle of view - I would agree with that. However, when you print both unresized images THE SAME SIZE, the DPI of the 7d shot will be higher and it will therefore have the POTENTIAL to have more detail. Obviously, if the 7D image is soft the extra rez won't matter. SO, the LENS absolutely matters as much as everything else. Even using the same lens, if you use it at different FL's you can potentially hit the weak spot in one shot and sweet spot for another. Naturally, DOF also comes in to play....Show more →
Man, obviously we cant take the same shots. I'm just saying, its not hard to judge image quality even if they are of different subject matter. All I asked was, post some. Lets see the images. There not being posted. You can look at my 1D3 images and see the sharpness and noise free images for what they are.
The look at some 7D files, bet no one will be more impressed with the 7D files.
I'm just using a common 70-200 f2.8L IS and 24-70 f2.8L in the shots I posted. These are very common lenses. And weak spot in FL, come on, now you are splitting hairs.
Look, I've posted some ISO1600-3200 shots from 2.8 to f5, lets see some 7D shots. Anyone with any knowledge can look at the files and see if they compare.
Thats the thing, the 7D files wont be as sharp and they will have more noise. Prove me wrong, thats all.
If ,on the whole, the 7D is found to be "better" why would anyone buy a 1d? For me its a bit like buying a 3 series BMW or a 7 series. They both get you there but in varying degrees of comfort & with different gadgetry. Its up to personal choice how much you are willing to spend. I find that I am more refreshed at the end of a long journey in a more expensive tool built for the purpose than a cheaper one. However bang for buck opens a whole new ball game-that's where the debate starts.
Tony B wrote:
For me its a bit like buying a 3 series BMW or a 7 series. They both get you there but in varying degrees of comfort & with different gadgetry.
It goes the other way, too: you may find it hard to park the 7 series Smaller, lighter bodies have a place, even if they are "inferior" in so many other ways. That's why I have the 5D2 next to the 1Ds3. Money has nothing to do with it.
Smaller, lighter bodies have a place, even if they are "inferior" in so many other ways. That's why I have the 5D2 next to the 1Ds3. Money has nothing to do with it.