Canon 7D Mark II Noise comparisons from TDP are now available

jrista said:
AlanF said:
Another test that is useful from TDP is the comparison of sharpness using the 200mm/2 L. The Mark II seems to my eyes to be an improvement over the 7D:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=673&FLIComp=0&APIComp=0

Although the 5D III still has a real edge, which is why the "extra reach" of the crop is not a factor of 1.6 because its image is more blurred.

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=792&FLIComp=0&APIComp=0


Hate to say it, but your assessment here is a little flawed. The 5D III does have the edge, however that is because it is a comparison of identical framing. Whenever framing is identical, more sensor area with similar pixel counts is always going to win. These tests are NOT tests of reach.


The 7D II appears softer (at f/2) than the 5D III only because the 5D III chart images were not taken at the same distance. If you DID change the framing with the 5D III, such that the chart was at the same exact distance from the sensor as it is with the 7D II...then the "softness" of the crop would at least be on par with the 5D III.


Furthermore, the softness is due to optical aberrations. For an adequate comparison of resolving power, you need to be more diffraction limited. If both cameras were tested at say f/4 at the same distance (which means different framing in the 5D III), the reach advantage of the 7D II should become much clearer.

In addition, Bryan Carnathan is fond of JPG straight off the camera in these comparisons, thus rendering them a less-than-perfect comparison. As we all know, JPG processing, picture styles, etc. vary from camera to camera. So I generally don't use his site's great comparison tool to rate camera bodies -- I use it for lenses, particularly sharpness vs. aperture.

- A
 
Upvote 0
ahsanford said:
jrista said:
AlanF said:
Another test that is useful from TDP is the comparison of sharpness using the 200mm/2 L. The Mark II seems to my eyes to be an improvement over the 7D:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=673&FLIComp=0&APIComp=0

Although the 5D III still has a real edge, which is why the "extra reach" of the crop is not a factor of 1.6 because its image is more blurred.

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=792&FLIComp=0&APIComp=0


Hate to say it, but your assessment here is a little flawed. The 5D III does have the edge, however that is because it is a comparison of identical framing. Whenever framing is identical, more sensor area with similar pixel counts is always going to win. These tests are NOT tests of reach.


The 7D II appears softer (at f/2) than the 5D III only because the 5D III chart images were not taken at the same distance. If you DID change the framing with the 5D III, such that the chart was at the same exact distance from the sensor as it is with the 7D II...then the "softness" of the crop would at least be on par with the 5D III.


Furthermore, the softness is due to optical aberrations. For an adequate comparison of resolving power, you need to be more diffraction limited. If both cameras were tested at say f/4 at the same distance (which means different framing in the 5D III), the reach advantage of the 7D II should become much clearer.

In addition, Bryan Carnathan is fond of JPG straight off the camera in these comparisons, thus rendering them a less-than-perfect comparison. As we all know, JPG processing, picture styles, etc. vary from camera to camera. So I generally don't use his site's great comparison tool to rate camera bodies -- I use it for lenses, particularly sharpness vs. aperture.

Sorry, but I'm not sure where you got that idea...

[quote author=Bryan Carnathan, TDP]
All test shots are taken in RAW format using Canon or Nikon's Neutral picture style (all parameters = 0). All aberration corrections are turned off both in-camera and during post processing. Using DPP (Canon's Digital Photo Pro), a sharpening setting of 1 is added to Canon shots.
[/quote]

Full details here: http://www.the-digital-picture.com/Help/ISO-12233.aspx
 
Upvote 0
ahsanford said:
neuroanatomist said:
Sorry, but I'm not sure where you got that idea...

He's largely JPG for his high ISO samples, so I presumed he did the same here. I stand corrected.

Well, I'm not sure where you got that idea, or to what high ISO samples you are referring. The ISO noise comparisons using the Kodak gray/color patch test chart, which are the subject of this thread, are also shot in RAW.

http://www.the-digital-picture.com/Help/ISO-Noise.aspx

Shooting out-of-camera JPGs for any sort of meaningful comparison (other than comparing JPG engines, for who-knows-what reason) would make for a pretty poor test, and that's now twice you've made that incorrect statement about TDP's testing.
 
Upvote 0
neuroanatomist said:
Shooting out-of-camera JPGs for any sort of meaningful comparison (other than comparing JPG engines, for who-knows-what reason) would make for a pretty poor test, and that's now twice you've made that incorrect statement about TDP's testing.

The problem with having a beard is how the egg on my face gets stuck in there. :-[ Again, I stand corrected.

Me and my speed reading. I saw 'picture style' and thought he was referring to onboard camera JPG picture style, when he was in fact talking about picture style in DPP.

The lesson here? Never doubt Carnathan.

- A
 
Upvote 0
GraFax said:
Ya'll are a bit too technical for a simple guy such as myself. I am usually limited by reach shooting wildlife with my 400 f5.6 on a 5D3. I know there are better lenses out there but that lens fits my needs pretty well other than being a bit short. Performance with a tele can be a challenge as there is no IS, f8 focus speed etc. If the 7D2 yields images that out-resolve images from my 5D3 cropped (8mp?) or the 5D3 with a 1.4x tele (with all of the issues that brings) I assume I come out ahead. So far that seems to be the case. The fact that the 7D2 is in many ways a more robust camera for outdoor work is just icing on the cake. Am I missing something? I get that full frame has advantages if/when you fill the frame but that just isn't always possible when photographing wild animals.


This is exactly right. There is a reach benefit to a camera like the 7D II. A lot of vocal members here strongly push the notion that because of the bigger pixels, FF cameras like the 5D III or 1D X can be enlarged and have better quality than the 7D II. When it comes to noise, it was a fine line in the case of the 7D (and other 18mp APS-C parts), however there has clearly been an improvement on the 7D II high ISO noise. Maybe not by stops, but a third to two thirds of a stop.


More important than now much noise you have, though, is the simple fact that smaller pixels resolve more detail. More reach is all about detail. Grabbing an 8mp crop out of the 5D III is never going to result in the kind of sharpness or detail you can get out of a 7D II.


There are other benefits, though, to using a sports crop instead of a full frame when you are reach limited: Maximum lens aperture. You nailed it on the head here.


With the 5D III, you can always slap on a TC, but when you do, you lose a stop to two stops of maximum aperture. That in turn affects the number of AF points you can use, how sensitive they are, whether they are cross type or not, and how fast AF occurs. A 5D III with say a 400 f/4 +1.4x TC is going to lose the ability to use f/4 capabilities of many AF points, and the AF speed will drop. A 7D II with the bare 400 is going to have effectively the same reach, but it will be able to use all of it's 65 AF points, in full cross type mode, with more light than the 5D III.


So you get the AF improvement on top of the increased spatial resolution. Since you are reach limited, there is no full-frame benefit to cropping the FF, you aren't gathering more light in total, just more light per pixel. You can always downsample the 7D II image to the same dimensions as the cropped 5D III image, and your noise will drop while concurrently the image will get sharper. This fact should be more recognizable with the 7D II, with it's higher resolution sensor and slightly newer sensor technology, than it was with the 7D (which, to this point, is generally what these debates have revolved around...whether the 7D had a "real" reach benefit or not.)
 
Upvote 0
jrista said:
More important than now much noise you have, though, is the simple fact that smaller pixels resolve more detail. More reach is all about detail.

Excellent analysis, and you didn't even mention dr once :-p... esp. the drop in af system capabilities of the 5d3 with slower lenses is often forgotten, it's not like Canon is very verbal about announcing this fact.

Alas, the crop 18mp 7d1 af vs. ff "wisdom" has been passed down the generations for years, and now that at last a Canon crop camera with a top-notch af system and ~1/2+ better iso is here it'll take some time for people to adjust.

Just one note: The resolution of the "small pixels" demands a very good lens. I remember you stating that there is no "outresolving", but still, I feel with a 20mp ff you can slap on about every ef lens around as long as you don't magnify the corners, while with crop you have to be more picky. Probably the reason why the 100-400L2 appears now.
 
Upvote 0
During the age of CRT television sets, I developed a very strong appreciation for the incredible tvs that Sony used to make.

I ended up buying into a whole Sony system. Surround sound. VCR (don't laugh). DVD player. I was very pleased.

Then with the advent of flat panel screens, Sony was no longer the player they used to be.

So there I stood one day, next to a salesman, weighing up just which LCD screen I wanted to spend my money on. I had this desire to buy Sony, as it would seamlessly fit into my little system. But my eye, my eye told me that the Samsung models were way better.

So I'm chatting to the salesperson and voicing some opinions when he nudges me and says; "Hey. You do know that you're only going to hang one screen in your room, right?"

The simple reality is that every single camera out there has it's shortcomings but when you take a good pic, what else matters?
 
Upvote 0
Marsu42 said:
jrista said:
More important than now much noise you have, though, is the simple fact that smaller pixels resolve more detail. More reach is all about detail.

Excellent analysis, and you didn't even mention dr once :-p... esp. the drop in af system capabilities of the 5d3 with slower lenses is often forgotten, it's not like Canon is very verbal about announcing this fact.

Alas, the crop 18mp 7d1 af vs. ff "wisdom" has been passed down the generations for years, and now that at last a Canon crop camera with a top-notch af system and ~1/2+ better iso is here it'll take some time for people to adjust.

Just one note: The resolution of the "small pixels" demands a very good lens. I remember you stating that there is no "outresolving", but still, I feel with a 20mp ff you can slap on about every ef lens around as long as you don't magnify the corners, while with crop you have to be more picky. Probably the reason why the 100-400L2 appears now.


I do believe that good glass matters, however I don't really think it matters more on crop than FF. I mean, people buy $10k lenses to use on FF cameras for the same reason...to get the most out of the format. You can try to use a 400mm lens on a 5D III for birds, but at that focal length, you have to get very close to get the most out of the system. On the flip side, a 400mm lens is an effective 640mm FoV on a 7D II. Just from a "filling the frame" standpoint, that makes it a lot easier to do just that, fill the frame, and get as many pixels as possible on target (which is ultimately where both lower noise and more detail is going to come from.)


Then you still have that added benefit of the potential for a faster aperture, which means potentially better AF, and therefor sharper results (I do notice that AF slows down on my 5D III with the 1.4x TC...it doesn't seem like it slows down a ton, but just enough to be a little frustrating).


I think the whole "you need a better lens" thing is itself mostly a myth. For one, particularly with Canon, the old 100-400mm is definitely not up to snuff by todays standards, or even yesterdays standards. It's a 1990's lens design, which was still back in the film era. I still have that lens (really need to sell it), and it was sent into Canon once for tuning, and even after that...at f/5.6 that lens is very soft, and it doesn't really sharpen up until at a minimum f/7.1-f/8. I still don't think it's critically sharp at f/8, and beyond that your into diffraction limited territory, so you just can't ever get the most out of such a lens. I think the very, very common pairing of the 100-400 with the 7D has in large part helped to create a deeper myth about how poorly the 7D (which then translated to all crop cameras) could resolve detail.


Once I started pairing other lenses with the 7D, including the 100mm macro, the 50mm f/1.4, and a few other lenses I've rented over the years, I realized that you can get critically sharp results with the 7D pretty easily. Often even hand-held. Yet every time I put my 100-400mm lens on it, I just never felt as though I was getting the kind of sharpness I knew the camera could deliver. (I also feel similar about the 16-35mm f/2.8 L II...I just never feel like I am getting enough out of that lens to even get the best sharpness on the 5D III, let alone the 7D.) It wasn't most lenses that had problems...it was really just that a small specific few lens that had problems.


With 4-stop IS, fast primes and even fast zooms (24-70 f/2.8, 70-200 f/2.8, etc.), getting critically sharp results in a tripod is a no-brainer, and getting critically sharp results hand-held does not really require special equipment. It requires a little bit of technique...however I believe, especially if you want to use all the pixels that FF cameras offer, that you need that same technique when hand-holding with FF as well.
 
Upvote 0
JMZawodny said:
Reach is one reason I got the 7D2. I was just out shooting a little while ago with the 300mm f/2.8L IS II on the 7D2. I'm very happy with that combo. Show me another way to get an effective 480mm F/2.8 lens.

Your 7D MkII makes your 300 f2.8 an effective 480mm f4.5 from a fov, dof and iso point of view, when compared to a 135 format camera.

Most people who have actually tested same generation cropped ff against crop cameras for the same fov almost universally agree the difference in resolution is very small.

Having said that, the price of a 10fps ff camera makes the 7D MkII an amazing buy, but for different reasons than the largely imagined crop reach advantage.
 
Upvote 0
zlatko said:
Actually, the 7D2 can do even better! The Digital Picture has found that with a very fast 1066x UDMA 7 CF card the 7D2 can exceed its 31 frame RAW buffer by about 50%. It can do 47 to 49 RAW frames:
http://www.the-digital-picture.com/News/News-Post.aspx?News=13858
And after the buffer is *full*, it can still shoot at 5.7 fps, faster than some cameras shoot with an empty buffer. :D

Darn it Canon, by exceeding the rated buffer, you're doing a bad job of the "crippling" that people are always talking about. ;)

Music to my ears. I want to always shoot RAW, and 3s worth of burst at max fps is just short of what I would like. I can hit the buffer limit on the 7D in RAW.

Nearly 5s should cover what I do perfectly.

And 5.7 fps when the buffer is full? Nice work Canon.
 
Upvote 0
MichaelHodges said:
Your logic could be used to excuse away any differences between any cameras.

Obviously not differences that are outside of the capability of post processing to erase.

If you want to compare sensor performance then RAWs converted with neutral settings (as neutral as you can make them; don't think it's ever perfectly neutral) is best. But I'm most interested in practical results when the entire imaging chain is in play. In this case there's very little to pick between formats at low ISO.
 
Upvote 0
GraFax said:
Super! That's about how I had it figured in theory but I was having second thoughts. Finally stopped raining here in Pennsylvania so I'm going to go and see if I can find some critters and put the theory to work. Still have about ten days left in the return window for the 7D2 just in case it doesn't pan out. I guess I could get a longer lens but I'd prefer not to deal with the extra weight.

Don't be disappointed when you do actual comparisons and realise there is actually very little difference.

For all the "simple fact that smaller pixels resolve more detail" proponents I have never, and I have asked lots of times, ever seen anybody actually illustrate that point to any worthwhile degree. There are lots of comparisons out there, I have posted my own many times, of same generation cameras but when you process each to their optimum it becomes very hard to distinguish the two even at 100% magnification.

Now a 1.4 TC is a hell of a lot cheaper than a 7D MkII, but that doesn't mean there is no point to getting the 7D MkII, and I have never said that. What I have said is if you already own a ff camera the difference in resolution in your actual images doesn't come close to the expected figures so make the purchase choice based on more realistic differences, like frame rate, AF etc etc.
 
Upvote 0
privatebydesign said:
JMZawodny said:
Reach is one reason I got the 7D2. I was just out shooting a little while ago with the 300mm f/2.8L IS II on the 7D2. I'm very happy with that combo. Show me another way to get an effective 480mm F/2.8 lens.

Your 7D MkII makes your 300 f2.8 an effective 480mm f4.5 from a fov, dof and iso point of view, when compared to a 135 format camera.

The light cone produced by the lens is determined by the lens alone and is f/2.8 no matter what camera it is attached to. Similarly, the difference in pixel size and number of electrons a pixel can hold is compensated for by the gain of the electronics so that ISO means the same thing from one camera to the next. And that is independent of lens. FOV of view does scale with crop factor. For a given solid angle, the APS-C puts more pixels in there. Depth of field is slightly reduced with the smaller pixels on the APS-C sensor. The problem I see here with folks throwing numbers around is that they are assuming very different things or performing some assumed scaling in order to match some arbitrary aspect between two different cameras. That was also the case with my original post. From an FOV perspective it appears to be a 480mm lens. DOF is reduced by ~1.6. Any discussion of ISO is irrelevant, but noise will most certainly increase as compared to a 480mm lens on a FF body.
 
Upvote 0
privatebydesign said:
Don't be disappointed when you do actual comparisons and realise there is actually very little difference.

For all the "simple fact that smaller pixels resolve more detail" proponents I have never, and I have asked lots of times, ever seen anybody actually illustrate that point to any worthwhile degree. There are lots of comparisons out there, I have posted my own many times, of same generation cameras but when you process each to their optimum it becomes very hard to distinguish the two even at 100% magnification.

^^ THIS.

Theory is wonderful. I like theory. But empirical testing is far more relevant to real-world applications. Ample empirical testing has shown that the 'reach advantage' of smaller pixels in smaller sensors is much less than theory would suggest. In actual practice, the 'reach advantage' ranges from small to nonexistent to a disadvantage, depending on lens used, lighting, and subject distance.

Interestingly, many of those empirical tests come from people who used only crop bodies and espoused the 'reach advantage'...then started using a FF body.
 
Upvote 0
Hmm, it's looking like it's time to reach for the
19594d1371339101-summer-solstice-karma-popcorn-smiley.png
 
Upvote 0
jrista said:
AlanF said:
Another test that is useful from TDP is the comparison of sharpness using the 200mm/2 L. The Mark II seems to my eyes to be an improvement over the 7D:

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=673&FLIComp=0&APIComp=0

Although the 5D III still has a real edge, which is why the "extra reach" of the crop is not a factor of 1.6 because its image is more blurred.

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=458&Camera=963&FLI=0&API=0&LensComp=458&Sample=0&CameraComp=792&FLIComp=0&APIComp=0


Hate to say it, but your assessment here is a little flawed. The 5D III does have the edge, however that is because it is a comparison of identical framing. Whenever framing is identical, more sensor area with similar pixel counts is always going to win. These tests are NOT tests of reach.


The 7D II appears softer (at f/2) than the 5D III only because the 5D III chart images were not taken at the same distance. If you DID change the framing with the 5D III, such that the chart was at the same exact distance from the sensor as it is with the 7D II...then the "softness" of the crop would at least be on par with the 5D III.


Furthermore, the softness is due to optical aberrations. For an adequate comparison of resolving power, you need to be more diffraction limited. If both cameras were tested at say f/4 at the same distance (which means different framing in the 5D III), the reach advantage of the 7D II should become much clearer.

It's not the distance that is important in determining sharpness in those TDP tests, it is the sharpness of the sensors. The camera and lens are at a distance so that the chart fills the sensor in all cases and the conditions are not such to cause light shimmering, haze etc that gets worse with distance. If the distance were the crucial factor, then a 600mm L lens would be much worse than a 100mm lens in his tests, which it isn't. And you can compare a 1dsiii with 300/2.8 + 1.4xTC with a 60d + 300/2.8. Both are about the same distance away from the chart but the 1dsiii is still sharper despite having a TC which degrades the image slightly.

http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=739&Camera=453&Sample=0&FLI=1&API=1&LensComp=739&CameraComp=736&SampleComp=0&FLIComp=0&APIComp=2
 
Upvote 0
GraFax said:
Whoops. Didn't mean to wade into the middle of the crop vs FF debate. What was I thinking? Seems like folks already have their clearly staked out positions on this one. Think I'll stay out of it.

Anyway, just ran off a couple hundred frames and I am really beginning to love the handling of this camera. The in-viewfinder level is awesome. Wish my 5D's had that. Its so easy to lose track of the horizon when you are photographing wildlife and then have to do a major crop to square things up. Given that it's already a "crop" sensor you don't want to needlessly be throwing away any precious pixels for that. 10 FPS is waaaaaay faster than 6. Hard to just get one frame off in high speed mode. It really wants to run!

In fact, you have a great handle on the crop vs. FF debate. It's really about cost, and about features for a given price point.
 
Upvote 0
AlanF said:
It's not the distance that is important in determining sharpness in those TDP tests, it is the sharpness of the sensors.

If so, then the lack of sharpness of older sensors in their test shots is misleading. With a 60d which appears blurry as hell on tdp, I can get shots I couldn't imagine to be sharper with a 100L lens. Or am I missing something here?
 
Upvote 0