Mini-Review: Initial 7D vs 7D2 Raw Comparison from Imaging Resource

50D banding with no work at all to pull it out. But yes all horizontal, not vertical. The other models did this too, when you pushed the exposure far enough.

legos?

TeamSpeed wrote in post #17197539
That would have indeed been interesting to have both the 70D and 7D2 released at the same time. It would have given people a choice..

The 70D has only been out a little over a year, amiright? I actually passed on it, awaiting the 7DII. I was hoping for 24MP, but I am a happy customer, at least for now.

TeamSpeed wrote in post #17197539
50D banding with no work at all to pull it out. But yes all horizontal, not vertical. The other models did this too, when you pushed the exposure far enough.

Or when you used high enough ISOs. Same thing, really. All of the major manufacturers had that issue up until that point, as I recall.

Even the D600 exhibits a small amount of that at extreme ISOs.

That was a very different problem than the 7D issue. I was very surprised to see the vertical banding issue in the 7D, because I'd never seen anything like it before.

That would have indeed been interesting to have both the 70D and 7D2 released at the same time. It would have given people a choice. However this way, Canon gets to double-dip. How many 70D owners are now going to upgrade? More now than if they had released them together, and people had to make a choice between 1 of the 2.

Well, I was actually thinking they might have released the 7D2 first, and then waited a bit to release the 70D. Kinda like they released the 7D (with the new 18 megapixel sensor) before the 60D...

It's possible that Canon learned from that experience that they should have released in the opposite order.


However, there's another way to put this: are you more likely to upgrade from a 70D to a 7D2, or from a 60D to a 7D2? Someone with a 60D now has an upgrade path to a 70D, but I expect it's unlikely they would turn around and upgrade to a 7D2 after upgrading to a 70D, because as much of an improvement the 7D2 looks to be over the 70D, it's not massive like the upgrade from the 50D to the 7D was, or like the upgrade from a 60D to a 7D2 would be. If, on the other hand, Canon had released the 7D2 first, people with a 60D would have been much more inclined to upgrade to that, precisely because it's such a massive upgrade, similar to the 50D to 7D transition. Keep in mind, in that scenario, the 7D2 would be the first on the scene with the dual-pixel sensor. Then, after the upgrade frenzy has subsided, Canon could introduce the 70D, which would appeal to Rebel shooters.

Tough to say which path would result in more sales, but I wouldn't be surprised if Canon erred with the release sequence they chose.

First of all thanks to TeamSpeed for all his work in looking at the performance of the 7DII.

This is a very exciting camera with lots of promise in terms of features and AF.

I too have been looking at the imaging resource CR2 files but am quite puzzled as to why everyone is saying the 7DII (and 70D) is 2/3 stop better for noise than the 7DI, I just don't see this in the data.

Quite the reverse, I measure a luminance mid tone difference in RMS noise of 0.58 stops against the 7DI.

This delta seems to be in agreement with the MkI eyeball too.

Can someone explain this discrepancy?

The headline plot is embedded below and the worksheet and methodology linked. Anyone with the latest DPP3 and Photoshop can reproduce this to test my results.

I will give some feedback, but it's pretty non-technical.

The noise on the 7D is "dirtier" meaning that if you push it, you end up with strange bright spots, or dark spots. It also has color blobbing on either the blue channel, or the red channel, depending on what copy you have. The images are a bit soft due to the AA filter, I assume.

The 7D2 seems to not have these same noise characteristics, and the detail is a bit sharper. That means that equal processing for each image would result in a rough equivalence of 2/3 stop.

Again, that is just my non-technical view of the different files, FWIW. I hope more technical folks will step in and help with the heavy pushing.

TeamSpeed wrote in post #17206745
I will give some feedback, but it's pretty non-technical.

The noise on the 7D is "dirtier" meaning that if you push it, you end up with strange bright spots, or dark spots. It also has color blobbing on either the blue channel, or the red channel, depending on what copy you have. The images are a bit soft due to the AA filter, I assume.

The 7D2 seems to not have these same noise characteristics, and the detail is a bit sharper. That means that equal processing for each image would result in a rough equivalence of 2/3 stop.

Again, that is just my non-technical view of the different files, FWIW. I hope more technical folks will step in and help with the heavy pushing.

Thanks for coming back.

I was wondering about these noise quality aspects.

I did seem to find just as much white speckling upon using my standard masked USM sharpening with the 7DII and the 7D, this was just using the old mk 1 eyeball. The rms measurements still showed the 7DI leading with this sharpening.

For the general noise qualities I could look a spectral analysis and see if I can see a difference. (I get uncomfortable if I can't see a measured result).

I would love to think the 7DII is the same or better in performance, but so far I can get near convincing myself.

Can you kindly point me to some samples that illustrate the quality difference, I can look to see if can understand what is going on.

My 50D banded pretty badly at 6400. That was why I purchased NIK NR. It was good at taking it out.

TeamSpeed wrote in post #17197246
That is much, much fainter than anything in the past though, even if it does happen. The banding was VERY apparent in older bodies, FF included. There are no good raw files yet from the 7D2 that I have found which would be good candidates for such a test. The one I used above didn't seem to show this, though, or at least I didn't notice it. If Canon has made any improvements over that last 5 years or so, it has been this banding issue. It is greatly reduced over the days of the 5D, 5D2, 20D through 50D etc.

It looks like you now have to be 5 stops underexposed to show the banding issues, instead of 2-3.

One thing I have been exploring recently is Roger Clark's article on how to recover dynamic range in the Canon Raw files: Here

By eliminating the built-in offset in the Raw data using Photoshop he is able to pull a lot more detail from the shadows and remove the color noise problems that are exaggerated by the added digital "amplification". I've tried it with varying success on both 5D3 and 7D2 files.

I have to say that without doing any rigorous analysis on the 7D2 files, standard Lightroom processing seems to get me within about 1 stop of the 5D3 files - the noise is very easy to clean up and often if I'm working at ISO1600 or below I don't even bother with NR. Even when I do, NR levels are about half of what I used on the 7D mk I and I have never seen any banding. ETTR does an excellent job on this sensor; I accidentally overexposed a bunch of ISO3200 frames last night (far more than I should have) and when I pulled them back down they were quite good. I don't want to do this on a regular basis because I was starting to lose highlight data on some of them, but it was surprising even to me how much better the shots looked than they should have.

umphotography wrote in post #17186684
That doesnt look very promising for noise. 3200 is barely usable to my eye. MKIV was better than that

Defintely agree with you.

DPR added new comparison tool for exposure latitude. They added 70d, d5500, e-m5 II. Although they think 70d is still behind
EM5. I think 70d did pretty well compared to even d5500 until +3ev push. All off them are started loosing lot of details after that.

http://www.dpreview.co​m …3955&y=0.975098​0507081942

I've been testing my 7D mk II at ISO 1600 and 3200 with pretty good results. Definitely an improvement on the original 7D, and if you are careful to ETTR and think about post-processing when shooting, 3200 is pretty darn good. I have used it for a couple of sports events and even without any NR the images are good at ISO3200 for web resolution. The mk II noise cleans up very well in Lightroom without any artifacts or excessive loss of detail if you're going for prints at 8x10 or slightly larger. The specs and sample images don't fully express how much better this camera is than its predecessor - the AF in particular clearly behaves better than my 5D3 in actual use, and that's with the case settings as nearly identical as they can be.

Bottom line, the mk II is going to be used for most of my sports shoots and astrophotography, and the 5D3 will be used where I need the slightly better noise performance or DoF control. It's great to be able to leave the 5D3 in the bag more often; I was putting a lot of clicks on it in situations where the 7D2 can now be used.