5D4 v. 5D3; High ISO noise?

[Scott Stephen]

Hi,

I am thinking of taking a plunge and getting a 5D4. I currently have 6D, and I would peobably be keeping it as a 2nd body. Obviously the autofocus would be just incredibly better. On the other hand, (presumably) AF will be at least considerably improved on the 6D2 (in a year). The one thing in particular that I care about, however, is high ISO image quality (ISO 3200 - 6400 really).

I have looked at the review on this site, which is my most trusted site for reviews. I have also looked at DXOMARK, which is rather puzzling in many ways. I have also seen a few other reviews.

I am not clear on whether the new 5D4 is better, the same or even worse at high ISO image quality. I may very well be missing it but I am not sure I understand if the review on TDP makes a clear statement on this. Also, on my monitor, the noise comparison tool shows a larger image on 5D4 than the test chart image for 5D3, which confuses me. DXO MARK's "sports" rating may (or may not) suggest that the camera is better at high ISO's, though I do not understand their methodology on that even after reading their explanation. Another site I saw seemed to say the 5D4 does less well at high ISO image quality. Perhaps part of the confusion is the combination of resolution loss + noise gain at high ISO. Maybe some sites drill down on the resolution improving while others just look at noise. I was wondering which is best overall IQ at high ISO, once you have applied some noise correction.

Is there a verdict on this one?

Thanks.

[Busted Knuckles]

My look at the 5D4 is really close - perhaps a tiny bit better from the 5dIII.

The differences are subjective and depending on the image, so not much of a change from III to IV. What are the other reasons why you would change as the noise is seemingly indifferent.

Make the decision on other issues - even if means - no camera.

[Minerve101]

I haven't used a 5D IV, so can't answer your main question directly. However regarding the different size of the display of the color test chart, there is an answer that Bryan has built in. One of the options for the 5D IV is "Standard Down-Sized to 5D III." That is the last choice in the pull-down box immediately under the camera name.

With this option the display of the test chart is the same size. The down-sampling of the 5D IV to produce this mildly improves the noise. Most post-processing software easily lets you down-sample in the same way, so this is a realistic comparison of something you could do with yourself with 5D IV images.

As I see it the 5D IV at 6400 is better with this comparison, but not a full stop. That is to say, the 5D IV Standard down-sized at ISO 12,800 is noisier than the 5D III Standard at 6400.

I agree with Busted Knuckles that sensor noise is not a very important distinguishing feature between the 5D III and 5D IV. There are lots of other features to consider, though.

[Kayaker72]

The 5DIV and 5DIII are very similar at "high ISO." The real difference in sensor performance is low ISO (100-800). Granted, I've heard some reports that the noise pattern is better even at high ISO on the 5DIV, which you can see a bit in Bryan's test (some of which is likely due to down sampling). DXOMark has the 5DIV as ~1/3-0.5 stops better than the 5DIII at ISO 3200-6400. This is likely only noticeable in very specific and likely controlled conditions. Up to you if benefits of down sampling and ~0.5 stop improvement in that range is worth the upgrade price. Personally, as a 5DIII owner, what appeals to me mostly are the reports of faster/better AF, improvements at low ISO (almost 2 stops), some reports of general IQ improvements, and to a lessor extent the 30 MP, 1 fps increase, improved LCD, weight, and a few other things. The ~0.5 stop improvement at high ISO is on my list, but fairly far down. While I do expect to eventually upgrade as I believe the 5DIV is all around generally an improvement over the 5DIII, I am not in a rush to do so as the 5DIII is very capable.

[Scott Stephen]

As long as the 5d4 high ISO IQ is not actually worse than 5d3 due to the increased pixel density, as was suggested somewhere else I was reading. If I see a nice flash sale on cyber Monday I may be a taker. I'd also love a crystal ball to preview the 6d2 for comparison.

[Busted Knuckles]

I might rent a 5dIV and really test out the differences in image quality at low ISO. I am not a high ISO shooter, I am a low ISO shooter 95% of the time.

I would like to see what almost 2 stops of DR really does

[Kayaker72]

I would like to see what almost 2 stops of DR really does

From the comparison's I've seen posted, there is a significant difference in shadow detail (I know, not surprising). But I've seen some comparison shots of the 5DIII vs IV and most of the scene looked very similar if not identical (slight resolution difference, again, as expected). It wasn't until they looked at, and usually pulled shadows up, that you start to see a difference. But what I saw was impressive. Something that was exposed at near black could be pulled into a range that still looked possible and natural.

This could actually benefit what I do. I do not do much HDR and tend to work from a single shot. So with waterfalls, there are instances, not frequent, but not rare either, where the DR between the water and shadows (caves, crevices, etc) results in banding if I try to pull those shadows up on my 5DIII image. For now I just pull those up to the extent I like the image or keep them darker.

[DavidEccleston]

From what I've seen, it means you can get more detail back from the shadows. Which should help any time you have bright and dark in the same scene, like a typical bride/groom shot for example, or landscapes with a bright sky.

[Busted Knuckles]

I did a quick test of running up the shutter speeds and ISO six stops worth and looking at the images.

Out of the camera, I was surprised out good the 6400 looked compared to 100 - I just hadn't ever done it before.

Per Kayaker's suggestion, though, I am going to do a little bit more rigorous shadow detail test. ( I deleted the images before trying to pull the shadows.)