This image quality thing... (10 years in camera development)

I recently hired a Canon 5Ds for an assignment that needed a lot of resolution.

Since I'm mostly an architecture photographer I thought it would be cool to test it in the field that is my expertise and where this camera should excel.

To make it a fair (but not scientific) comparison I returned to a spot I've been a lot and tried to recreate one of my favorites.

A bit of technical background, I do most of my prints at 90cm by 60 cm, so that is about my reference and where I thought the extra resolution would come in handy.

It's not that I'm disappointed, but I'd think the difference would be way more apparent, both files are about 8200 pixels wide and both files are in my opinion edited and ready to be printed.

Could you please tell me what you see? Was I expecting too much from the 5Ds, is my technique lacking or was my editing not on par?

Shot 1 is taken with the rented 5Ds and a 17mm TS-E

Shot 2 is taken with my old and trusty 5D and a Sigma 35mm 1.4

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

Some 100% crops

HOSTED PHOTO
please log in to view hosted photos in full size.

HOSTED PHOTO
please log in to view hosted photos in full size.

Well it's kinda hard to comment, as you have not, for some reason, included the EXIF/exposure information on your pictures.

That being said, what comes to mind is this:
-The 5D shot is better exposed imo (possibly as a result of the different weather?) and less busy/better framed imo (the other shot is a bit busy in the corners, not in a nice way)
-In the 5D shot, as well, you're getting more out of the lamps on the right-hand side of the frame, the pattern and the star shapes are nice.

However, I think that this is the extent of how the 5D could be said to be better.

In the 5Ds shot, you can almost read the labels on the metro station (isn't it "Rotterdam Central"?). The posters in the more central part of the photo are super clear.

In any case, since the framing of the shots are different and you're using different lenses, all this could be affecting your results.

I'm not sure how valid this comparison is given that the lighting is different and you were using two different lenses. Also, were both shots processed in the same way?

But setting all that aside, I find the 5D shot much more pleasing, especially the beautiful starbursts around the lights.

Maybe it's just me, but I feel like the focus is off on the first 5DS shot. Almost as if there was a bit of camera movement or something. I'm going to chalk a lot of this up to the different lenses and possibly the settings. 5DS shot is overexposed IMO.

Thanks to all of you for your replies, I've manually included the EXIF information now.

As said before, I knew this wasn't going to be a scientific comparison and since there's 3 years between the shots I can't rule out any difference in editing.

Maybe, as suggested, it comes down to exposure and composition and the difference is for the most important part on the user side.

Besides, if there's indeed camera movement in de 5DS shot, I should practice a bit more before upgrading anyway

The differences are most likely related to lenses used. The reason ones goes to the 5DS is to get the resolution out of it. All that you managed to do is reduce the effectivity of all those extra pixels by reducing your optical zoom and expecting the digital zoom to take over. Not only that, you are then setting the expectations that both lenses resolve identically down to the pixel level, and I am quite sure that isn't true with these two.

Optical zoom is always preferable first and foremost to digital zoom, which is what you do when you do the 100% view.

Run both cameras side by side with the same lens. Either increase the size of the 5D image up to the 5DS or resize the 5DS down to the 5D to do your comparisons. I can guarantee you that the 5D will never match the 5DS when you crop and resize up to the same pixel dimensions, it will likely look as if you put a low quality 2x TC on the 5D.

Aside... I have read many times how the high resolution in the 5Ds will expose ANY camera/hand shake movement.

Just curious, why such a low ISO and small aperture? ISO 50 is a calculated ISO and not native to the camera. You could have bumped it to 100 at a minimum or even 400 without loss of IQ. Also, opening the aperture to f8 would have gotten you another stop. that would have reduced you 30 second exposure to 4 seconds.

Lastly, as all your verticals look pretty good, I assumed that you dialed in some tilt. How did you focus, through the viewfinder or live view? I have found that live view works much better as you can zoom in 10x and look across the frame.

chuckmiller wrote in post #18906269
Aside... I have read many times how the high resolution in the 5Ds will expose ANY camera/hand shake movement.

Only if you pixel peep. There are more pixels to record movement, so when you zoom into the pixel level, what should be a crisp line or dot may actually now be spread across several more than a larger pixel, lower resolution sensor.

However when looking at an image at a normal viewing size, you will see no additional shake over a lesser resolution body if all factors are the same.

So this only comes into play if you crop substantially.

TeamSpeed wrote in post #18906387
So this only comes into play if you crop substantially.

That.

chuckmiller wrote in post #18906269
Aside... I have read many times how the high resolution in the 5Ds will expose ANY camera/hand shake movement.

Many APS-C cameras have higher pixel density, including some 1.5x crop cameras that have no AA filter at all (weaker than the 5Dsr, which has a mild, partially-self-cancelling filter) and m43 cameras that would be 80MP as FF, used hand-held with telephoto lenses. I sometimes handhold my 400DO II IS with 2.8x worth of TCs on my 7D2 (same pixel density as the 5Dsr), equivalent to 400mm on a 400MP FF camera, and at least 60% of images show no camera motion blur at 100% pixel view.

Either we are getting all these negative anecdotes because the 5Ds actually has some unique slap problems totally independent of pixel density, or many people who use FF cameras have not developed stability skills to the degree that they think they have.

Forget about automatic tripod stability. A tripod only returns the orientation of a camera to the same mean point of orientation, and it oscillates about that point readily, unless you dampen such motion or eliminate causes of oscillation. Just mounting on a tripod is not a guarantee of better-than-handheld stability for medium-length tripod exposures!

gjl711 wrote in post #18906385
Just curious, why such a low ISO and small aperture? ISO 50 is a calculated ISO and not native to the camera.

That doesn't matter. The fact is, unless you blow out highlights you needed, an "extended" ISO 50 is fine, because it really, really is ISO 50, even better, noise-wise, than if the camera actually had half the analog gain of ISO 100.

John Sheehy wrote in post #18906520
Many APS-C cameras have higher pixel density, including some 1.5x crop cameras that have no AA filter at all (weaker than the 5Dsr, which has a mild, partially-self-cancelling filter) and m43 cameras that would be 80MP as FF, used hand-held with telephoto lenses. I sometimes handhold my 400DO II IS with 2.8x worth of TCs on my 7D2 (same pixel density as the 5Dsr), equivalent to 400mm on a 400MP FF camera, and at least 60% of images show no camera motion blur at 100% pixel view.

Either we are getting all these negative anecdotes because the 5Ds actually has some unique slap problems totally independent of pixel density, or many people who use FF cameras have not developed stability skills to the degree that they think they have.

Forget about automatic tripod stability. A tripod only returns the orientation of a camera to the same mean point of orientation, and it oscillates about that point readily, unless you dampen such motion or eliminate causes of oscillation. Just mounting on a tripod is not a guarantee of better-than-handheld stability for medium-length tripod exposures!

It is quite simple. The issue has been brought about because people crop heavily and/or zoom into nearly 100% view of their files.

If conditions are exactly identical, and you shoot two different pixel density sensors with the same exposure settings on the same subject, either cropping heavily and/or zooming down to the pixel level would indeed give the indication of more blur. Now many of us know that is simply because we are "magnifying" the image with more resolution, so that blur is visually exaggerated, but that is where these discussions have gone over the years.

This diagram shows what I am trying to depict. Let's say that red line is a hair that is blowing in the wind. One will show "more blur" when zoomed in or cropped heavily than the other, because there were smaller, more pixels to record that hair and its movement, creating the look that there is more blur. But display both images so that they have the same basic enlargement factor, they will show no such thing.

Back when the 5Ds first came out I had a longish conversation with a Canon rep at my local camera store, WEX, regarding use, and shutter speeds. I would still love a 5Ds for use at airshows with my 150-600. The thing is that I shoot mostly propeller driven aircraft, and to get good prop blur I keep my shutter speeds at or below 1/160s. The rep was horrified at that, claiming that I would have nothing but blurred images. He then suggested that I should really look at a 7DII for this sort of photography. It took a great deal of explaining that since the linear resolution of the two sensors was the same, any blur on the sensor would be identical, and that the 5Ds would give me 2.56× more pixels in a larger window to work from when I was focal length limited. The larger sensor also helps once the subject is close enough that I would no longer be FL limited, so it would be a double win.

I think I finally got it through to him, but I also got the feeling that the watch out for blur, use a higher shutter speed bit was company policy, just to avoid the people who can only compare image quality at 100% view.

Alan