5DS vs 5D Mark III - Low Light Performance?

Hi,

Thank you to anyone who takes the time to read and reply to this post, it is much appreciated. I have been doing my research before asking the forum for advice and feedback.

I am getting ready to buy my first full frame camera body, and I am debating between the 5DS and the 5D Mark III (5D3). My main concern holding me back from the 5DS is the difference in low light performance between these two, and I was hoping that people with experience with either of these bodies or both could shed some light on this matter.

The 5D3 has a vastly higher ISO range, going up to 102,400 ISO while the 5DS only goes up to 12,800 ISO. By the numbers alone, this would suggest that the 5D3 is superior in low light situations. I have read some reviews that say this very thing, but it appears they may just be going by the numbers and not real world performance. Some people who have tested the bodies write that they perform equally well in low light, and that the higher ISO settings on the 5D3 should basically not be used as they just make poor quality images.

For instance, Ken Rockwell writes in his review, that

"you can set the 5D Mk III above 12,800, at which point it looks worse and worse and worse. What I've shown is that the 5DS has the same low-light ability as other Canon full-frame cameras — and that you don't need to buy another camera for use at high ISO. You can push the 5DS to above ISO 12,800 using Photoshop. I haven't tried it; ISO 6,400 is good enough for me to shoot hand-held in moonlight." (link to review)

This seems to suggest that the 5DS performs perfectly fine in real world low light situations. That 12,800 ISO is more than enough to shoot in any situation that you would generally come across. Another review I read, when comparing the 5D3 to the 6D back in 2013, remarked that the 5D3

"struggles at ISO 25600, and definitely at ISO 512600. The most noticeable difference is when you view the ISO 102,400 images, where it's clear the 5D Mark III is struggling to control the levels of noise." (link to review)

This also suggests that the higher ISO settings of the 5D3 are not very usable, and won't be missed on the 5DS, making the 5DS just as good in low light situations.

Does anyone have experience using the 5DS in low light? At indoor events in with poor lighting? Outdoors at night?

Does anyone who owns the 5D3 actually use the ISO settings above 12,800? If so, how is the image quality? Are they keepers or are they actually quite poor and not worth even trying?

While 50 MP is certainly more than I will usually need, what it provides is flexibility in post. The ability to crop a significant portion of the image and still end up with a 22 MP result is just one example. And there are of course times I hope to get that great shot, utilizing the full frame and enabling very large prints with unsurpassed sharpness and image quality.

Most reviews state that the 5DS is a great camera for landscape or studio work, where you are going to mount the camera on a tripod and take advantage of the extra resolution. While the 5D3 is a better camera when it comes to being versatile, especially if you are going to shoot in low light. Landscapes are one of my favorite subjects, but I also shoot the photos for our family gatherings, do travel photography, and I would like to shoot more events and start doing some real estate photography as well.

Some things that are really attracting me to the 5DS are of course the resolution of the sensor, the ability to shoot-through flickering light (though this will be used not very frequently), the built-in bulb timer and intervalometer as I really enjoy night shooting and astrophotography. You don't have to worry about two other sets of batteries for the wireless shutter remote with programming. I also really like the new autofocus with the second sensor which it seems is improving the facial recognition of the 5DS over the 5D3.

One big advantage of the 5D3 over the 5DS to me is related to AEB. It seems the 5DS is limited to shooting just 3 frames, while the 5D3 can go up to 7 frames. Even more, the 5D3 has HDR blending built-in, while the 5DS seems to lack that (can somebody confirm this?). The 7 frames versus 3 frames is easy enough to work around for very still subjects. I can just shoot my 3 bracketed frames, adjust the exposure time and shoot another 3 frames. This extra delay in time, however, can lead to more of the subject moving, and to an accidental shift in the positioning of the camera. This issue is not a deal breaker for me, but if the low light shooting is a big difference then it adds to reasons to select the 5D3.

This is of course a big decision. I tend to buy a device and work with it for as long as I can. I haven't bought a new camera since 2008 when I got the Canon 40D. Unless you count the built-in cameras on my smartphones.

I am going to be building an entire new system around this body. The camera body, a few lenses, an EIZO monitor and a calibration device for the monitor. I already have a fast computer and plenty of available storage space.

The other advantages of the 5D3 seems to be video related, in that they have a headphone jack for audio monitoring and uncompressed HDMI out. I don't anticipate ever using the headphone jack, and only very rarely might I ever use the HDMI out. I'd probably want to preview and edit the video on my computer before showing it, so I won't be showing it from the camera; and if I do show it from the camera I'm not worried about the compression. So video is not a real factor in my decision.

One last, but very real, concern is the camera's forgiveness of bad technique. I am not an expert yet. I have much to learn still, and a lot of room for improvement in my shooting technique. The increased resolution is going to pick up blur when I move the camera a little. And I think that shooting in a crop mode will not alleviate this issue, as it is just cropping out the edges and not averaging neighboring pixels for example. The pixel density remains the same. Have people been having good experiences with the 5DS while handholding it for stuff like landscapes, and people at events?

I have been reading review after review after review of these cameras and I have been flipping back and forth between them. It seems like the 5D3 was the camera to get until the 5DS came out, but the 5DS is not objectively better in every respect like the 5D3 was to the 5D2. Thus the decision is not straightforward.

Again, thank you for reading through this post, and thanks in advance to anyone who may reply with their advice and thoughts. I appreciate any and all feedback. And please ask me any questions that you think might help advise me.

Cheers,
heldGaze

What are you shooting? And quoting ken Rockwell is never a good thing. The two variants of the 5Ds are really meant for studio and landscape use.

gonzogolf wrote in post #17837125
What are you shooting? And quoting ken Rockwell is never a good thing. The two variants of the 5Ds are really meant for studio and landscape use.

Landscape, astrophotography and night shooting are some of my favorite subjects. But I do want to be able to shoot for family gatherings, events like birthday parties and weddings, real estate and our travels.

My main concern is the low light performance of the 5DS, it really is the main thing that is holding me back and seems to be biggest draw to go with the 5D3 if you don't care about video. So I am hoping to hear about the real world experience people here have on these two bodies in low light.

Do people regularly utilize the higher ISO settings on the 5D3? Have people noticed the 5DS struggle in low light? Does the 5DS work well with your travel photography and shooting family events?

Out of curiosity, what's the issue with Ken Rockwell? I'm not entirely familiar with him, but his wasn't the only review I've read stating the higher ISO of the 5D3 perform poorly with lots of noise and to avoid shooting with those settings. It was just the most direct pull quote I had when writing this post. It's the conflicting information across many reviews on the low light performance of these bodies that has me seeking the real world experience of those in this forum.

Again, thank you for your reply.

Heya,

heldGaze wrote in post #17837118
For instance, Ken Rockwell writes in his review, that

Oh boy... here we go.

***************

More to your thread though...

So the 5Ds actually has good ISO performance in general. It's not worse, really, than the 5D3. Some tests have shown it to be slightly better than 5D3 & even 6D. I'm not sold on that though. So far, I don't see any real improvement. But I didn't expect improvement. I expected ISO to suffer on this release, because of the massive increase in resolution (shrinking photo bucket surface are per pixel). They did a great job of retaining good noise handling and ISO performance, despite that. So really, that's a good thing. But I wouldn't call this camera a low light titan. Sony has that covered handedly. And the 6D is a more practical for price low light camera for general photography anyways.

I wouldn't split hairs over ISO on the 5D3, 5Ds, 6D, etc. They're all usable at ISO 6400 and 12,800 with good processing. They're all able to have horrible noise if you under-expose at high ISO too. So I will stress the processing knowledge of high ISO, and the skill/technique to expose with high ISO in mind (a little ETTR goes a long way for high ISO shooting is key to getting good clean files to work with, with any of these cameras. And one is not totally significantly better than another at high ISO. The 6D is a hair better than the 5D3 at high ISO. But it's not enough that I would worry about it. The 5Ds is like that, it's not significantly better nor worse than the 5D3 at anything, so don't split hairs on this.

The 5Ds is a high megapixel camera with a great AF engine (1D series) with some other new(ish) features and abilities. But it's meat & potatoes is the megapixel count. You either need resolution, or you don't. Resolution is useful for heavy crops. High resolution helps with pixel density advantage (which is what APS-C is normally all about), so if you're always looking for more reach, this is useful. But you really need to "need" that reach. High resolution also matters for big prints. You can already easily print 16x20 with today's less than 20MP resolution files. How big do you need it? 20 x 30? Do you honestly print this big? Ever? Some do, mind. Maybe you do. But if you don't even begin to print near that size, I don't see why you'd want a 50MP camera at this point in time. So I'm just playing devil's advocate about that resolution because while it is key to the camera, if it's not something you will actually benefit truly from, you're just talking yourself into a numbers game, instead of looking at the reality of what you're doing and the bottom line. If you do print a big file, ask yourself, was it possible to just stitch a few images into a pano and make a large resolution image anyways? If you do this once a year, type thing, it would be easier to just work around not having 50MP directly. And stitching 4~6 images from a ~18~20MP sensor is going to be fine too. Just something to consider.

I actually sort of chuckle when I see things like "this is for landscape photographers" and what not. Sure, that might be true. But it really needs to be in context. It's important for someone who's printing really large prints. Not little or medium size prints. Large ones. Large prints also cost big money, and are pointless if they're not done by a high quality printer on high quality mediums, unless you are ok with viewing it from several yards away or going "billboard" on it. If you've looked at really huge prints, up close, you've probably noticed...they're not meant to be seen at the same distance as you looking at a monitor. Large prints, even high quality ones, really are meant to be looked at from farther away to look correct. If you "pixel peep" a large print, you'll never be happy. Pixel peeping in general will always result in dissatisfaction anyways, but I digress. This camera is for ANYTHING you can use it for. Plain and simple. It's resolution is important for large prints. Again, that's the real meat & potatoes of this camera. Sure, you can argue that it has more data to work on in post, better cropping (pixel density), etc. But that's not important unless that's something important to your work. If you don't already do a ton of post work, or already struggle with reach, then these things are not as critical.

I don't know a lot of event/low light folk who are using the 5Ds. Most of them either use 5D3, 6D or have moved to Sony & Nikon. Maybe it's just local meta. It's not like there's a true census out there watching this that we have access to.

If you're looking at these cameras, you most certainly do not want to use an in-camera HDR engine. So if this is a concern, I'll just break this bubble now. Save your money. By the way, the 5D3 can be configured to do 2, 3, 5, and 7 in AEB via custom functions.

I would not build a system around a body. The body will not hold up, the way lenses will. Build a set of good lenses and they will be great the next 20 years. Build with a body, and sure you get a few years, but it will not compare to how well lenses age. Get the body you need. Don't focus on the body and then build around it. Build everything together as a system, but just know, lenses are the better place to focus your money along with a good support system. Also, going high resolution also means you have to significantly invest and maintain a system that can deal with that file size whilst editing, and for storage. It will go fast, depending on how much you shoot and what you retain. This is an added cost over time. Higher resolution is a blessing, but also a curse, depending on how you look at it.

No SLR is forgiving. High resolution does have something in the form of micro-shake as a potential issue if shooting with a low shutter speed hand held or on an unstable mount. But if you're on a tripod, or using an appropriate shutter speed, this will not be something to hinder you. But the problems you will encounter early on getting into photography will apply to virtually all bodies and not just this one. I wouldn't stress it. But again no SLR is really forgiving, so keep away from greenbox and focus on learning the craft, that will help you in the long run.

I would sit back and take a hard look at the reality of what you actually do with the images you take. If you're honest with yourself, you'll know if you need a 5Ds or not.

On the other hand, if you can wait a bit more, next releases are coming which will drive down current prices on these cameras, maybe opening some new doors.

Very best,

Direct quote from: http://www.kenrockwell​.com/tech/notcamera.ht​m

Your equipment DOES NOT affect the quality of your image.


ie, use 5D3 or 5DS or 10D. Results will be of the same quality.

He is right, to an extent. Do you really think the average person cares about noise and sharpness, especially given today's mobile phone cameras, and social media sites downsizing images with less than stellar final result? Content matters more to the person on the street, not resolution, sharpness, and noise. Those are things reserved for photography forums.

I shoot astro with the 5D3 at 8,000 ISO. It's fine, there. If you are printing, 12,800 is fine, too. If you pixel peep, 12,800 under decent lighting will be fine, with a good RAW noise reduction technique.

At the same *print size*, the 5Ds and sr will look better. They have newer electronics, and more pixels will look cleaner.

Regarding blur, it's 'increased' only because of the increased magnification when pixel-peeping at 100% view. The 5Ds has better-controlled mirror.

If you want to do multi-image HDR, don't do it in-camera. You can get better results with software.

One point I want to address is the idea that more resolution will make motion blur worse. I hear this kind of thing a lot. Like "I have so many pixels with the XYZ camera, I have to double my shutter speeds to avoid blur." or "Canon needs to release all new lenses to work with camera XYZ because it outresolves everything they have now."

None of this makes any sense. More resolution can never make an image worse from the standpoint of lens resolution or motion blur. All that more resolution will do is resolve more detail....provided there is more detail possible in the image in the first place.

So if you have a super soft, low resolution lens and mount it on a high resolution and low resolution camera, the absolute worst possible case for the high resolution camera is that it will resolve exactly the same image as the low resolution camera. And if the lens is better than you thought, the high resolution camera may well extract more detail from it.

That's it.

There are some negatives associated with higher resolution, but blur or lens detail are certainly not among them.

blur is certainly masked more with a low megapixel camera, however, it's clear that man L lenses and likely all primes, maybe even most, can outresolve the typical 24mp canon sensor. They wont outresolve the entire frame, but the center certainly. Once you view at the pixel level, there is more blur, however, doing an apples to apples comparison at the same magnification, it will be no worse than a low MP body. Of course you buy a high MP body to use the pixel advantage, hence higher handheld shutter speeds and better glass is highly relevant.

Charlie wrote in post #17837486
blur is certainly masked more with a low megapixel camera, however, it's clear that man L lenses and likely all primes, maybe even most, can outresolve the typical 24mp canon sensor. They wont outresolve the entire frame, but the center certainly. Once you view at the pixel level, there is more blur, however, doing an apples to apples comparison at the same magnification, it will be no worse than a low MP body. Of course you buy a high MP body to use the pixel advantage, hence higher handheld shutter speeds and better glass is highly relevant.

Right. As in - if you buy a much higher resolution camera you may not get 100% of the increased resolution possible if A) your lenses are not up to the higher resolution and B) your technique does not provide for an image with low enough blur and/or perfect enough focus / DOF to actually resolve the detail possible via the new sensor.

But people so often phrase this in such a way to imply that higher resolution will be worse. Like in this thread, the OP is worried that a 5DS might be worse for motion blur than a 5D Mark III because it has higher resolution. That worry, to me, reflects a misunderstanding of what is going on with sensor resolution.

There is no possible situation where the 5DS can resolve less detail than a 5D Mark III. So it is illogical to cite a concern of motion blur as a reason to avoid higher resolution.

The discussion takes me back 6-7 years ago when people talked about diffraction and pixel densities. Back then the confusion was about the diffraction limited aperture, which was something like f/11 for a 30D and f/5.6 for a 7D. And people thought this meant the 30D was better, because it wasn't diffraction limited until f/11 while the 7D was running into diffraction limitations on resolution by f/5.6. Never mind that this meant that the 7D at worst was merely equal to the 30D in resolution beyond f/5.6 (itself a dubious conclusion related to resolution and diffraction).

Jeffrey is absolutely right. This phenomenon has happened several times in the last 5-10 years; motion blur, diffraction, resolving capabilities; there have been so much misinformation circulating that the unsuspecting user will understandably start to believe that we're taking 1 step forward, 2 steps back with today's higher res cameras. And that is mostly because we tend to forget about "common output size", viewing different megapixel images at one common size (same size print, monitor, etc.). Like Jeffery stated, an image will not get worse, less detail, more motion blur just because we're using a higher resolution camera as long as we're comparing the same output size. Of course if we pixel peep a 50 MP image at 100% compared to a 12 MP 5D at 100%, we may see a bunch of differences regarding perceived sharpness ("oh, my old 5D looks crisper at 100% than my 5Ds both at 100% view"), motion blur ("hey, why do I have some motion blur at 100% on my 5Ds images which I don't see on my old 5D??"), and the same may be true to image noise compared at 100%. We must remember that we're viewing 2 images at drastically different magnifications. Print or display them both at the same output size and all these "problems" disappear ...and now the 5Ds will look better in many cases.

For the OP, a good tool to compare image noise at various ISOs is DPreview's studio comparison tool, where you can compare different cameras at a common output size, where the image noise will show results different from most other test sites that may only test noise at 100% view (at the pixel level)

Here it is, just change the second camera to the 5D3 (or 6D, or whatever). Select "Comp" or "Print" in the upper right corner of the studio scene to display common output size. There you go, the true noise comparison is displayed. NOTE: this is for the 5DsR, so you may want to change that first cam to 5Ds, although it won't make much difference.

http://www.dpreview.co​m/reviews/canon-eos-5ds-sr/8

right, tdp reports that the 5Ds is better in low light than the 5D3 when doing an apples to apples comparison. So if TS wants the camera best for landscapes AND low light events, that would be the 5Ds and not the 5D3 IMO.

Yes, ...and the 5Ds is much more forgiving when pushing shadows. The 5D3 is really not the best at that, ...to put it mildly. That's why I shoot my real estate shots with the 6D I have (much better at that).

JeffreyG wrote in post #17837491
Right. As in - if you buy a much higher resolution camera you may not get 100% of the increased resolution possible if A) your lenses are not up to the higher resolution and B) your technique does not provide for an image with low enough blur and/or perfect enough focus / DOF to actually resolve the detail possible via the new sensor.

But people so often phrase this in such a way to imply that higher resolution will be worse. Like in this thread, the OP is worried that a 5DS might be worse for motion blur than a 5D Mark III because it has higher resolution. That worry, to me, reflects a misunderstanding of what is going on with sensor resolution.

There is no possible situation where the 5DS can resolve less detail than a 5D Mark III. So it is illogical to cite a concern of motion blur as a reason to avoid higher resolution.

The discussion takes me back 6-7 years ago when people talked about diffraction and pixel densities. Back then the confusion was about the diffraction limited aperture, which was something like f/11 for a 30D and f/5.6 for a 7D. And people thought this meant the 30D was better, because it wasn't diffraction limited until f/11 while the 7D was running into diffraction limitations on resolution by f/5.6. Never mind that this meant that the 7D at worst was merely equal to the 30D in resolution beyond f/5.6 (itself a dubious conclusion related to resolution and diffraction).

I recently visited my local camera shop, WEX in Norwich, and they had a Canon rep working in store that day. I had virtually this conversation with him, and even he did not get it! He was giving the you need higher shutter speeds/better resolution lenses stuff that you see in many of the reviews that just look at images at 100%. When I suggested that I would like to use a 5DS with a 600 f/4 for my aviation work, he really tried to suggest that I would get better results from the 7DII when shooting a 1/160 with a 600mm lens. Even though in many instances I would be using the same approx 15×10mm crop from both sensors, with virtually identical numbers of pixels in each case. This is actually a situation where the new 5DS mirror mechanism would give you an advantage though. I don't really need more than about 5 fps, the buffer depth is fine too, and the AF is basically the same, so no issues there. To me the advantage of the 5DS with the 600mm is that in the situations where I am not cropping down to use less than approx 22.5×15mm of sensor area you get with an APS-C sensor I can fill the larger sensor area with image and have the 600mm prime act as if it were a 375-600mm zoom on a 7DII, that is the zoom range that I use for about 75% of my shots. As I am zooming out by using a larger area of sensor, the overall noise at least should actually get better as I "zoom" out, since overall image noise is dependent on total sensor area used, not the pixel size. I really had a job to convince the guy (actually I'm not sure I did convince him) that if I use the same area of sensor, with the same resolution I would get the same results. I know that I didn't win the same size output argument with him, I guess Canon have him to well trained in the party line.

Alan

If you have higher resolution (smaller pixels) you do have the ability to capture smaller things as "blurs" that would not be captured as "blurs" on the sensor with bigger pixels. The whole confusion is whether you will see them or not. The size of the blur will determine how much the image needs to be elarged for viewing before it is perceived as a blur. This enlargement affects all kinds of blur, whether motion blur, diffraction or DoF. SOme people who discuss this think "I always print the same size" so more pixels doesn't affect them, while others think "I like to print to the max at e.g., 240 ppi so with more pixels I can print larger" - they are the ones who will need to think about blur differently