Concerns over the Nikon D5 DR Strategy

[BernardLanguillier]

With all due respect, there are quite a few pointless gear discussions at LL, and I have wasted my share of time in many of those, but this one is shaping out to be record breaking!

The reality, plain and simple, is that no single camera is best ar everything but that many cameras are excellent at a lot of things.

Every one of us just needs to know our exact needs and be objective about the actual performance of the cameras we are considering relative to these needs. I do landscape and action and have decided to own both a D810 and a D5. I'd rather have used my cash on something else and have zero pride in camera ownership, my focus was only to get the best tool for the job. And having used those cameras for their intended purposes, I believe I took the right decision.

Every one of us has different needs, financial situations, skills,... so there is zero value in trying to convince others that we know better than them what is best for them.

Cheers,
Bernard

[AlterEgo]

If I wanted an ideal, 'do everything/go everywhere' camera, I'd take the guts of the D5, replace the sensor with the one from the A7r2 (or the upcoming 80MP Sony one), put it in a compact body with optional grip and alter the firmware to include both 40MP mRAW and 20MP pixel binning.

if you do binning post sensor you can't get 14fps though (otherwise you can do 80mp @ 14fps and you are not)...

[shadowblade]

if you do binning post sensor you can't get 14fps though (otherwise you can do 80mp @ 14fps and you are not)...

Why not? The limitation is almost certainly going to be in the write-to-card speed - that's the slowest thing in the pipeline by far.

[shadowblade]

With all due respect, there are quite a few pointless gear discussions at LL, and I have wasted my share of time in many of those, but this one is shaping out to be record breaking!

The reality, plain and simple, is that no single camera is best ar everything but that many cameras are excellent at a lot of things.

Every one of us just needs to know our exact needs and be objective about the actual performance of the cameras we are considering relative to these needs. I do landscape and action and have decided to own both a D810 and a D5. I'd rather have used my cash on something else and have zero pride in camera ownership, my focus was only to get the best tool for the job. And having used those cameras for their intended purposes, I believe I took the right decision.

Every one of us has different needs, financial situations, skills,... so there is zero value in trying to convince others that we know better than them what is best for them.

Cheers,
Bernard

That's why many - maybe even most - people shoot at least two different bodies, depending on what they're shooting and where they're going. It's why the D4s/D810 combo is so popular, as is the 1Dx/A7r2 (the 5Ds having arrived three years too late to gain much traction) and as was the 1D3/5D2.

[shadowblade]

Hi,

Here are some data from Bill Claff: http://www.photonstophotos.net/Charts/PDR.htm#Nikon%20D5,Sony%20ILCE-7RII

I expect Canon's 1Dx2 to be a better all-round camera than the D5, if it uses the same on-sensor ADC tech as the 80D.

AF should be comparable between the two bodies, and will be top-of-the-line anyway. The 1Dx2 also shoots 2fps faster.

The Canon sensor should have better DR at low ISO and comparable high-ISO performance (not sure about ridiculous-ISO peformance, such as ISO 102400 and up, which no-one actually uses).

Canon's generally also been stronger in video, and I suspect the 1Dx2 will also incorporate most of the functions of the 1Dc, making it much more useful for journalism and journalistic-style travel photography, which increasingly incorporates video (I don't expect there to be a 1Dc2).

Finally, Canon currently has a stronger lineup of supertelephotos. The 200-400L is incredible - the inbuilt 1.4x TC adds a huge amount of utility, particularly when shooting action where you can't afford to take 15 seconds out to change teleconverters, or in environments where you really don't want to be taking the lens off (e.g. on the deck of a ship or Zodiac with seaspray everywhere). It would be nice to see future supertelephotos all incorporate a 1x/1.4x/2x TC switcher.

[hjulenissen]

Why not? The limitation is almost certainly going to be in the write-to-card speed - that's the slowest thing in the pipeline by far.

I would guess that the limitation is usually one of:
1. The sensor itself overheating
2. The attainable sampling rates of A/D converters (on sensor or off)
3. The bandwidth of poorly conditioned, relatively long cable runs from sensor to supporting electronics

I am not an expert on (analog) circuit design, but I guess that performance usually is a compromise between cost, size, power drain, image quality, pixel-rates, emi, etc. If you want pixel-rates that are 2x or 10x the current industry norm (=comfort levels), you probably will have to pay by a shift in said compromise.

Less likely candidates:
4. Bandwidth/processing speed of cpu (storage of raw files more likely to be bandwidth limited than computation limited)
5. Speed of memory buffer (DDR memory tends to be OOM slower than cpu speeds)
6. Memory card bandwidth (not really an issue for "burst")

[shadowblade]

I would guess that the limitation is usually one of:
1. The sensor itself overheating

Not likely. Current, even very dense sensors can handle live view and video (which still reads every pixel on the sensor) and can be used for hours-long exposures without overheating.

2. The attainable sampling rates of A/D converters (on sensor or off)

I doubt it, since current 42MP and 50MP sensors can sample video at 25 or 50fps, utilising every pixel on the sensor (although downsampling to 1080p or 4k for storage). The sampling rate of that is much faster than even reading 80MP at 14fps.

3. The bandwidth of poorly conditioned, relatively long cable runs from sensor to supporting electronics

This is still much, much faster than the 350MBps write speed of XQD cards, or even the write speed of next-generation 1GBps XQD cards. I don't know if this would present any bottleneck to getting data from the sensor to the processor at 1960MBps (80MP at 14 bits and 14fps), but the write speed is a much bigger bottleneck. Hence the need to downsample to 40MP or 20MP for the faster burst speeds.

Less likely candidates:
4. Bandwidth/processing speed of cpu (storage of raw files more likely to be bandwidth limited than computation limited)

Agree

5. Speed of memory buffer (DDR memory tends to be OOM slower than cpu speeds)

Memory used in computer graphics cards handle ridiculous amounts of data at very high speeds. Also less necessary if you use multiple banks running in parallel, or if you process and downsample in real time and write it straight to card for unlimited burst durations.

6. Memory card bandwidth (not really an issue for "burst")

This is probably the biggest issue, actually. Burst rates at full resolution would be very slow, unless you used a large, fast buffer. But it can be alleviated by downsampling to mRAW or sRAW for the faster speeds.

Don't forget, high-speed video cameras, shooting hundreds of frames per second, handle more data than even an 80MP sensor shooting at 14fps. And they're not always attached to computers to do this.

[Rob C]

With all due respect, there are quite a few pointless gear discussions at LL, and I have wasted my share of time in many of those, but this one is shaping out to be record breaking!

The reality, plain and simple, is that no single camera is best ar everything but that many cameras are excellent at a lot of things.

Every one of us just needs to know our exact needs and be objective about the actual performance of the cameras we are considering relative to these needs. I do landscape and action and have decided to own both a D810 and a D5. I'd rather have used my cash on something else and have zero pride in camera ownership, my focus was only to get the best tool for the job. And having used those cameras for their intended purposes, I believe I took the right decision.

Every one of us has different needs, financial situations, skills,... so there is zero value in trying to convince others that we know better than them what is best for them.

Cheers,
Bernard

Goodness me, Bernard, it's taken you a long time to realise that!

I have a very good reason for following this thread, even though zero interest in upgrading cameras: the little bar that usually serves me lunch has one major flaw: the guy who serves soup is very unstable, and will invariably spill some during the delivery journey from his serving pot to my plate, resulting in wet eating implements at the very least, if not drops on my own person. In an effort - so far successful - to prevent this happening to me, I have taken to switching on the cellphone as I wait for him to come fill my soup plate, placing the cellphone on the place mat before me, and shifting the actual soup plate well to the side, away from me, and preventing him, through this shocking breach of etiquette, from giving me any extra helping of said soup other than where it belongs.

The longer this thread becomes, the safer I feel, and the less energy I need expend in an attempt to keep reading without switching around. As my actual, really concerned attention is rivetted on the soup server and not the thread, this is a completely satisfactory arrangement for me.

Anyway, looking at your shots with the 2.8/400 helps me better guess which lens Hans Feurer might be using most of the time - I think it's perhaps a tad longer than yours, but just as far out of my financial acceptability paradigm. All of this personal and deeply photographic research fills mealtimes with sweet oblivion and necessary distraction from the task at hand!

;-)

Rob

[Josh-H]

"Goodness me, Bernard, it's taken you a long time to realise that!"

AMEN!

[BernardLanguillier]

Anyway, looking at your shots with the 2.8/400 helps me better guess which lens Hans Feurer might be using most of the time - I think it's perhaps a tad longer than yours, but just as far out of my financial acceptability paradigm.

It really depends on how much you spend on soup and cars.

Cheers,
Bernard

[BernardLanguillier]

AMEN!

It's been a veeeeeery long time I haven't been involved in any discussion of the sort, believe me.

I do admit that I wasted lots of time genuinely trying to convince the world about the value of high DR bodies, only to realize that this was perceived by many as a brand discussion which I never cared about.

Cheers,
Bernard

[Rob C]

It really depends on how much you spend on soup and cars.

Cheers,
Bernard

In my dotage, as little as possible on either! The best soup is the stuff (infinitely variable) that I make for myself. But the consumption of that's reserved for weekends when I try to avoid pubic places as much as I can. I doubt the public either appreciates the gesture or is even aware it's being offered.

;-)

Rob

[dwswager]

Just to add a little perspective to this discussion:

1. If you can't make a quality large print from either the D5 or the D810 you are doing it WRONG!

2. Considering that a D810 image is 7360 pixels on the long edge and the D5 image is 5568 pixels on the long edge, if you make a 60" print from the D810, then a D5 image would yield a 45.39 inch print at the same resolution.  Considering that probably less than 0.00001% of all images get printed larger than 45" inches you can draw your own conclusion as to how beneficial the additional pixels are in the grand scheme of things.  Full Disclosure, I own and love the D810.

[Josh-H]

2. Considering that a D810 image is 7360 pixels on the long edge and the D5 image is 5568 pixels on the long edge, if you make a 60" print from the D810, then a D5 image would yield a 45.39 inch print at the same resolution.  Considering that probably less than 0.00001% of all images get printed larger than 45" inches you can draw your own conclusion as to how beneficial the additional pixels are in the grand scheme of things.  Full Disclosure, I own and love the D810.

I wish this wisdom was more commonly held. Well said.

[shadowblade]

2. Considering that a D810 image is 7360 pixels on the long edge and the D5 image is 5568 pixels on the long edge, if you make a 60" print from the D810, then a D5 image would yield a 45.39 inch print at the same resolution.  Considering that probably less than 0.00001% of all images get printed larger than 45" inches you can draw your own conclusion as to how beneficial the additional pixels are in the grand scheme of things.  Full Disclosure, I own and love the D810.

I'd consider 20-24MP good for around 20x30" or 24x36", and marginal for larger prints. I'd consider the D810/A7r as being excellent at 24x36", good at 32x48" and marginal for 40x60", although the larger format is more forgiving (e.g. where I might want 200ppi for a smaller print, a larger one may only require 150ppi, or even less at the very largest sizes). It's not a strictly linear relationship - as the print size increases, you tend to need more overall pixels, but fewer PPI. Ideally, I like to produce 40x60" prints using stitched images, even with the A7r2 (although a 70-80MP sensor would likely change this), but would still be happy to produce one from a single frame if that's all I had. I couldn't say the same about blowing up most 21MP 5D2 files to that size, particularly low-key images or ones with lots of shadow details.

You don't often need to print that big, but, with the right image, it can be the difference between, 'Here's $5k, I want that thing blown up huge to hang in my lounge' and 'Sorry, I can't print it that big and still have it look good'

Bottom line is, for 99% of applications/shots, any of the D810, D4s or D750 will be more than adequate and will give you indistinguishable final results. But, when you're comparing top-tier gear, it's all about getting that last 1%, where the different cameras will give you noticeably different final results.

A top athlete would do a lot to shave 0.1s over 100m.

[Josh-H]

'Here's $5k, I want that thing blown up huge to hang in my lounge' and 'Sorry, I can't print it that big and still have it look good'

You know.. over the last 7+ years I have sold prints up to 60 x 90 inches from 1DX 18 MPX files for well into the thousands of dollars. NOT ONCE have I had a client say 'its not sharp enough', or 'it doesn't have enough resolution". Not once.

[shadowblade]

You know.. over the last 7+ years I have sold prints up to 60 x 90 inches from 1DX 18 MPX files for well into the thousands of dollars. NOT ONCE have I had a client say 'its not sharp enough', or 'it doesn't have enough resolution". Not once.

Fair enough if there's nothing to compare it to. The impact of the size and scale of such large prints can overwhelm any effect of the loss of detail.

Put it side-by-side against a high-resolution version and, suddenly, it becomes, 'Why's this one blurry?'. It's like putting a 480p TV next to a 1080p one, or a 1080p one next to a 4k one. The lower-resolution one looks just fine on its own, until you put it next to a better one.

[Manoli]

You know [...] I have sold prints up to 60 x 90 inches from 1DX 18 MPX files for well into the thousands of dollars. NOT ONCE have I had a client say 'its not sharp enough', or 'it doesn't have enough resolution". Not once.

I'd consider 20-24MP good for around 20x30" or 24x36", and marginal for larger prints. [...]
A top athlete would do a lot to shave 0.1s over 100m.

IMO you're both right.

A good case can be made that it's quasi impossible to tell the difference between prints made at a base resolution of around 250ppi. The examples above rely on a base resolution of around 88 ppi.   I think that's too marginal to make a hard and fast rule. It can be done, has been done but will depend on subject matter and, at these limits, the quality of pp.

For a great majority, at reasonable viewing distances, resolution or lack thereof won't be the first thing they'll notice - an abruptness in (subtle) tonal transitions will be - unless they're viewing your 60" masterpiece through a loupe.

If you want to play safe, it's wiser to try and stay above 150-180 ppi unrezzed base resolution - but that doesn't mean the limits can't be pushed - they can, just not all the time.

[dwswager]

IMO you're both right.

A good case can be made that it's quasi impossible to tell the difference between prints made at a base resolution of around 250ppi. The examples above rely on a base resolution of around 88 ppi.   I think that's too marginal to make a hard and fast rule. It can be done, has been done but will depend on subject matter and, at these limits, the quality of pp.

For a great majority, at reasonable viewing distances, resolution or lack thereof won't be the first thing they'll notice - an abruptness in (subtle) tonal transitions will be - unless they're viewing your 60" masterpiece through a loupe.

If you want to play safe, it's wiser to try and stay above 150-180 ppi unrezzed base resolution - but that doesn't mean the limits can't be pushed - they can, just not all the time.

Quite right.  My 60" example came from a previous post and I used it to make my point.  I rarely print larger than 28" and when I do it is a multi-shot stitch.

But having worked with Epson on a military tech project, when they say their dither algorithm tops out by 360ppi and even that is rarely distinguishable from it's output when fed 240ppi, I believe them.  I used Genuine Fractals back in the day to work pretty decent magic with 2.1 and then 6MP images.  Guys feeding it 720ppi are the same guys whining they could see the dots back in the late 1990s.  Yeah, with a 4x Schneider loupe!

I think the sharpness is over done today the way over saturated was back in the Fuji Velvia days.  Velvia gave a wonderful look, but was just over done for everything.  Images sell for the emotional impact the evoke.  Some images need sharpness and high resolution and other just don't.  Obviously, it is better to start with high resolution and soften it when needed, but I just think we have gotten a little anal retentive over it compared to all the other attributes that make a great photograph great.

[shadowblade]

I think the sharpness is over done today the way over saturated was back in the Fuji Velvia days.  Velvia gave a wonderful look, but was just over done for everything.  Images sell for the emotional impact the evoke.  Some images need sharpness and high resolution and other just don't.  Obviously, it is better to start with high resolution and soften it when needed, but I just think we have gotten a little anal retentive over it compared to all the other attributes that make a great photograph great.

On the other hand, many images sell well precisely because they have so much visible detail, in addition to an interesting and aesthetically-appealing subject. I often hear comments about things looking so real and being so detailed that they almost don't look like photos, instead having a holographic, almost 3D effect, and of people saying that, each time they look at certain photos, they see something new - a detail or feature they hadn't seen before - thus giving the image great staying power rather than being interesting merely for a single viewing.

Sure, not every photo needs this, but better to have the capability for those photos that benefit from it than to lack it when you most want it.

Also much of the 'excess sharpness' you see in certain images is just that - excess sharpening - rather than excessive detail. Sometimes, you will see prints which appear far too crisp, but, on closer inspection, actually don't show all that much detail - just lots of sharpening.