Go with the 5Ds or simply rely on Upscaling Software?

The new 5Ds large mega pixel sensor (50mb) seems to be excessive considering that its mainly used for large print purposes. As there are a few good upscaling software available, why would any professional choose a large sensor over - say 5D3 22mp sensor?

Is there other advantages apart from large print size? Afterall, the actual connected cost of needing bigger storage, faster computers etc. all build up...

I have not looked too deeply into the 5Ds. Does it have the anti-aliasing filter removed (usually reserved for MF, but with a pixel count that high, it's possible).

If there is no aa filter, then it will deliver intrinsically sharper images than other full frame cameras. (But at the cost of possible aliasing and moire.)

Take a look into that.

Rad

The more clean data you have to work with from the get go, the more you can do with it.

For example: stitching images together, while taking longer due to more processing time, can be more reliable if you are starting with higher resolution images. So even if you're down-sampling the final product to a standard 1080P resolution wallpaper, you may be able to get better results with less effort when starting at something with a far larger resolution.

Pixels are data points, NOT little coloured squares. If you have time I suggest digging up a copy of the paper "A pixel is not a little square" and giving that a read. Makes more sense if you have a bit of a math/data processing background, but covers enough that non-technical people should be able to get a bit of an idea of what's actually going on.

You can always discard data after the fact, but it is very hard to fill in gaps that you never collected in the first place. So you need to sort out what you actually want to do and figure out how much data is actually useful to collect. (This is why I prefer crop sensors if I'm doing telephoto work. If I know I'm going to be tossing out everything but the middle anyway, then I don't really see a point in recording it in the first place.)

that software can enhance detail that is already there, but it can't create detail that was missed when the shutter was released.

and it might gain you a couple of MP, but it isn't going to gain you 28 MP.

How about for billboards or prints so large that even images from the 5Ds will need to be upscaled?

I don't think storage cost is really even an issue anymore. At least not for those who would buy a 5ds. What about those professionals who make use of 50 and 60 mp medium format cameras?

If you don't print large, then you don't need the 5ds...if you do print large, then the extra pixels will be a big benefit.

I guess I can ask you why 24mpix is needed if we can shoot with 8mpix and just use software to uprez. The same logic applies to the 5ds versus 5d3.

Left Handed Brisket wrote in post #17521028
that software . . . . can't create detail that was missed when the shutter was released.

Bingo! That's the answer.

The upscaling software will not be able to replicate all of the fine detail that was in the actual scene that you photographed. This may not matter for many types of photography, but if you are trying to produce fine-art images of scenes that have an enormous amount of very fine detail, then the greater number of pixels will be a distinct advantage.

I think the 5Ds/5Dr vs. upscaling really comes down to how often you'll need the higher-resolution (or more tightly-cropped) images that would require the better detail from extra pixels on target. I wouldn't rush out to buy a 5Ds unless I expected to need the extra detail on a regular basis of weekly even more frequently (assuming the higher resolution couldn't be met by use of a backup 7D/7D2 or other "cropped" body).

^this

for me, everything I take is generally for web-sized use, up to about a 12"x18" print, which I actually have down scale to, even at 300dpi, so for me the 5DS/r would just be bigger files, slower workflow, more storage cost, with zero benefit. Bear in mind that even at the closest focusing distance of your eye you can't see more than 300dpi, and I doubt there are many printers with will pint it either (do not confuse ink-drops per inch on a consumer ink-jet with real pixels per inch!), so for me you have to be making large prints of a fairly regular basis to make it worthwhile and I just don't see myself doing that any time soon

I saw some stunning prints from a large format inkjet this weekend at a printing seminar from Stephen Johnson and it made me wonder why anyone would need more pixels than a 5D3 produces. I know I don't.

FarmerTed1971 wrote in post #17524822
I saw some stunning prints from a large format inkjet this weekend at a printing seminar from Stephen Johnson and it made me wonder why anyone would need more pixels than a 5D3 produces. I know I don't.

I was at the Tule Lake National Wildlife Refuge visitor's center last moth. There is a photo of Greater White Fronted Geese in flight - a whole flock of them - and it covers a 12' long by 8' high wall. The way the room is laid out, you really can't get a clear view of the photo until you are standing about 4 feet from it. The photo could be really cool looking, as it features a head-on, close-up view of about 20 or 30 geese flying right at the camera. But it looks like crap because there is no well-resolved feather detail in the geese ...... just soft areas of varying color where there should be intricate feather patterns on the breasts and distinct primaries and secondaries along the wings.

This type of shot could never be stitched, as the geese are in flight and it takes a shutter speed of about 1/1250th of a second to get the wings sharp - if you took a multi-frame burst there is simply no way that the wings would be in the same position for each of the frames. So the only way to capture this is with a single frame, which then must be blown up to 8 foot by 12 foot dimensions.

Uprezzing wouldn't be a good solution, because the software will fill in what detail it thinks should be there (via interpolation), rather than what really was there. If I am looking at a photo of geese, I want to see the feathers and feather segments that were really there on the goose's plumage - I do not want to see feather detail that was created by a software program that was trying to guess what the feathers looked like.

I don't know about you, Ted, but if I were shooting that photo for that application I would surely want more pixels than the 5D3 has.

Good example... but, was it meant to be viewed that close? I guess there was no choice.

here is a post of mine quoted by Tom R. from a thread about the 7DII's higher resolution.

https://photography-on-the.net …showthread.php?​p=17512994

while not directly about upscaling software i think it give some good insight into why a lot of people need higher resolution straight from the camera. And fwiw, i've used some of that up scaling software in the past and wasn't impressed.

Some people also just really enjoy very large prints which are loaded with lots of detail.

One of the few gallery views I've seen that truly impressed me where black and white landscape prints, that were enlargements from custom film in a home built camera. The photo started life on film that was already something like 4 feet by 8 feet, and they were then enlarged 'just a little' to take up entire walls. You could actually sit there with a loupe and find more details and things that you wouldn't have noticed in a normal photo, such as being able to barely make out a small song bird on a branch far in the background. (A single pixel on my 7D would have covered the whole bird.)


So the short version on resolution: Know what your output goals are, and record the data needed to produce your target work.

In most users' contexts, for cropping and reasonably detailed editing, around 20 MP is more than good enough. Print size in relation to sensible viewing distances tends to undermine arguments for huge megapixel images.

There's no real basis for insisting 30+ MP are better than 16 for billboards and vehicle advertising, etc.

Here's an interesting article on how many MP we may actually need:

https://photographylif​e.com …xels-when-i-could-have-50