Why 80MP for Landscapes

[mtomalty]

This is problem with the internet, I write as a matter a fact and it does not come off very well on the internet.  I'm not trying to offend anyone but everyone should see all the data before making opinions without seeing some well executed work.

Fair enough Tim and, yes, it doesn't take much for ones comment to be misinterpreted.

To your second point, I guess part of the problem is that very few of us will ever have
the opportunity to see ALL the data in order to inform our opinions and, thus, rely on
interpreting second hand information from those more experienced such as yourself to
gain more insight.
When that information is shared with aggression, whether intended or not, it sort of makes
it difficult to treat such information with much more credibility than that spouted by some of the pretenders who seem to relish in reading their own words.

Moving on, I look forward to hearing more about your new printing technology

Mark

[ErikKaffehr]

Yes, Sir!

One hundred percent with you!

Best regards
Erik

Why 80 instead of 60, or even 50 megapixels may be a more relevant question considering the available options. If you like high resolving power 80 megapixels is the current best (except for stitching) so just go for it. However if you have economy considerations, or lens compatibility considerations for tech cams a 60MP back is probably wiser option.

I'd like to have say 400-600 megapixels or so to really outresolve the optic system, make sure all digital artifacts are nullified by diffraction. That's not available today though, and today I don't think it's too big a difference between 50 and 80 megapixels.

To me making a print look nice big is not necessarily that the eye is out-resolved on nosing distance (although that is cool), but that the print lacks ugly artifacts on nosing distance, and what's ugly is a matter of taste. For example I prefer the structure of film grain much more than digital jaggies.

I prefer an oversized digital print look fuzzy at nosing distance rather than having false detail. I don't like the look of fractal upsizing either, I rather have that standard bicubic blur. Previously I shot at f/11 most of the time, but nowadays I shoot at f/16 with my 6um pixels (50-60 MP backs). Why? To reduce aliasing artifacts and give oversized prints a smoother look up close. May sound backwards to reduce resolution by diffraction to print larger, but that's exactly what I'm doing. I've done my own tests and come to the conclusion that I prefer that nosing look, and welcome the increased depth of field. Of course on normal viewing distance it doesn't matter, resolution is fine. Digital sharpening techniques makes it much easier to control the negative effects of diffraction than film too.

It also matters which type of landscapes you shoot. 90% of my landscape photos are close distance shots often 4:5 portrait format rather than wide panorama shots with lots of tiny details at infinity. Actually the background is often slightly out of focus in my shots due to limits of depth of field (and my desire to get a subtle layering effect in the scene). The grand panorama type of shot gains more from having high resolving power as it's more likely to be printed big and invites the viewer to step close and look at the details.

I've attached one shot to show an example where very high resolving power doesn't really provide much value. It's made with a 33 MP back and you don't really gain more meaningful detail with more resolving power. I actually didn't have any recent shot which show the opposite. But the best example where you really want lots of resolution is huge environmental group portraits in panorama format where you have relatively small faces in the frame, the viewer is then very much invited to step close.

Personally I don't stitch any longer as I don't think it's a satisfying way to do photography. For the same reason I don't focus stack either. One scene in one shot, ideally without need to crop in post -- that's how I like to make photographs. It's all in my head though, and if you don't have any issues regarding photographic enjoyment when it comes to stitching it's a great way to gain that extra resolution when you need it, and I'd say in most cases when that extra resolving power actually means anything are in such situations where stitching and stacking works well. Of course stitching with 80MP back gives you more pixels faster than a 50MP back, but I think the importance of high MP count is in any case further reduced in this scenario.

[Joe Towner]

Why not?  80MP is a good starting point for a lot of images, but it's all relative to the output and subject matter.  Stitching doesn't work well when things keep moving on you.

[Phil Indeblanc]

Thanks Mike for the kind feedback. This image is a single shot from my shoot last fall with my friend Tim Wolcott. I have stitched and have some files at 1.7gb as the unrezzed up file.

Steven

http://www.friedmanphoto.com

Maybe not the best thread to comment, specially since I see others great work as well, but I think the point to the OP has already been made.....
Your images of dense trees are just wonderful. Color and my imagination of detail when enlarged...but most importantly the feeling you must have in the location.  Are any of those images artfully created by multiplying trees and backgrounds, or are there locations with such density of trees? (Autumn Glow, Sunshine in Aspen, etc).  Great work! If you have created the images with creative editing, it is great work, and totally expected in scenic work that is supposed to be aesthetically perfect.

[stevenf]

Phil

Thank you for the kind feedback. All the images on my website are real. Nothing has been removed or added to the images. Recently,  I just started stitching panoramic images, all but a few of the panoramic images are from a panoramic film camera. The images were scanned and colour matched to the slides. I have spent over 12 years for almost a month a year photographing the north american aspen forests. Sometimes it has taken hours or days in the same location waiting for the right conditions. Fyi, we have printed these images at lengths up to ten feet.

Steven

http://www.friedmanphoto.com

[BobShaw]

Is anyone using a 80mp to capture nature? Why? Is there a practical reason for shooting with 80mp? How much is enough?
Mike

If you are asking the question then you don't need more.
I know a very famous landscape photographer with an 80MP Phase One that he was given because he is a very famous landscape photographer. However he still shoots with the same film camera that made him a very famous landscape photographer. Whatever pixels you need to give the required PPI you require is enough.

[torger]

Here in Europe film has a very strong position in the art world. It's common that photographers which do both commercial and art do their commercial work digitally and their art using film. It seems to me that in the US there's a higher acceptance for digital.

I think part of the reason artists often use film is that the layman audience (and gallerists) thinks it's more "real photography", so it's not really about quality. That the film more often than not is scanned and post-processed as any digital picture they don't really know or care about, but there's still a few that use an analog process end to end.

[KevinA]

Tim,
I'm just curious, I don't do any printing myself or very often prints, but I would like to know if you can divulge what your new printing technique or theory is please. I quite understand if you want/need to keep it tp yourself.

[Wayne Fox]

Is anyone using a 80mp to capture nature? Why? Is there a practical reason for shooting with 80mp? How much is enough?
Mike

I shoot with an 80mp back because I can’t buy a back with 120mp yet.

[Ken R]

I shoot with an 80mp back because I can’t buy a back with 120mp yet.

Yep. I think with a bayer sensor the more pixels the better if the quality per pixels can be retained or even improved. Lets say we upscaled the Sony A7RII sensor up to the size of the one in the IQ180. That would make it what? about 135MP? That should make a lot of us more than happy. With full color pixels 80mp would be plenty.

[Don Libby]

I shoot with an 80mp back because I can’t buy a back with 120mp yet.

Best comment yet!  I shoot with an IQ180 because I can. Plus I just produced an 80x40 panorama of the Grand Canyon even the smaller 60x30 and 40x30 shows incredible details and people love them.

Like Wayne I'll keep shoot the 80 until a 120 or larger is released....

[ErikKaffehr]

Hi,

More is better for sure. But there is also something called good enough and something called diminishing returns. Technically, I would guess that we may run into diminishing returns in many cases, due to workflow limitations and diffraction. But, I think good enough sets in before that.

Again, definition of what is good enough may depend on both subject and observer. Just a small example:

Once upon the time I was shooting 12 MP APS-C. With that format I could do decent level A2 size prints. Than 24MP full frame came around and I obviously made some comparison. Surprise, the 24 MP full frame where much better, the prints not so much. In at least one case I could not make any difference in A2-size prints.

So, I went on to 39 MP (on a P45+). But I still make a few prints from my 12 MP APS-C era, as it was my most productive period. An A2-size print I made for my latest exhibition was made from APS-C at 12 MP and even hand held. But, another image had a lot of dust specs. I realised that those dust specs were birds, but they were not resolved. Surface texture, like sunlit mountains, were just fine.

I am pretty sure that good lenses of today exceed the resolution of sensor over a large part of the image area. No, you don't go into Otus or Rodenstock territory for that, just have a decent workflow. But, it may just happen that there will be little benefits much of time. But, when the subject is demanding, more pixels will always shine.

Another way to see it is that more pixels render the image more accurately, thus having more leeway for post-processing.

The simple truth is that more is more in science and technology, and the art of photography is based on technology.

Best regards
Erik

Best comment yet!  I shoot with an IQ180 because I can. Plus I just produced an 80x40 panorama of the Grand Canyon even the smaller 60x30 and 40x30 shows incredible details and people love them.

Like Wayne I'll keep shoot the 80 until a 120 or larger is released....

[tjv]

I always said that when 60mpx backs with a decent interface (Credo / IQ etc) were "cheap" enough, I'd jump into the digital fold. Scanning 4x5" and 6x7cm film with a Flextight 949, I got/get a file that is approximately 24" on the short edge at 300dpi. With my Credo 60 I get a file that's approximately 22" on the short edge at 300dpi but with an astonishing increase in captured detail due to a lack of film grain. What I've found however is that with my limited practice in upresing, film scans tend to hold up better–give a more pleasing look due to random grain structure and lack of jaggies–when pushed to monster sizes. I'm sure the 80mpx backs are awesome and better in this regard (marginally, perhaps) but I've never used one.

I think a pertinent question to this thread is what techniques do people use to upres their digital files to the monster sizes they print at?

[jng]

I think a pertinent question to this thread is what techniques do people use to upres their digital files to the monster sizes they print at?

Since one of the advantages of using high resolution MFD sensors is the ability to print large, I agree that this is a pertinent question. So far I've been upressing my 60 Mp (8984 x 6732) files in Photoshop CC - this seems to work fine on modest "enlargements" (up to 45" on the long dimension @ 300 ppi, or about 1.5x) but I'm guessing there are more sophisticated ways to do this, which may make a difference for even larger prints.

-John

[Wayne Fox]

More is better for sure. But there is also something called good enough

While I agree, the problem I have is when I take a shot I don't know how much will be good enough ...

Ctein had an interesting article  about where things might be going over on the online photographer.  Always an interesting read.  I responded to a featured comment which sort of talked about the good enough idea, which is also featured below the article. Rather than rewriting my entire concept here, those interested can take a look.

[tjv]

Since one of the advantages of using high resolution MFD sensors is the ability to print large, I agree that this is a pertinent question. So far I've been upressing my 60 Mp (8984 x 6732) files in Photoshop CC - this seems to work fine on modest "enlargements" (up to 45" on the long dimension @ 300 ppi, or about 1.5x) but I'm guessing there are more sophisticated ways to do this, which may make a difference for even larger prints.

-John

When you use CC to upres, do you use the new resampling algorithms? Do you do another round of sharpening afterwards?

[ErikKaffehr]

Hi,

Ctein's article is interesting, but I also feel he is a bit on the optimistic side. He is obviously right about bayer sensor wasting a significant part of the light, but we have not yet seen a better solution. Also, I would also think that pixels can be shrunk quite a lot, but shrinking pixels reduces DR. Low DR (at the pixel level) and need of high bit readout is depending on adequate pixel size. It seems that the  Nikon D810 is close to needing > 14 bit of data.

The way I see it, sensor makes improve their designs and technologies in small steps. Thinner wiring shifts optimal pixel sizes downwards. Sony's latest sensors put the photodiode in front of the wiring, just as an example. I am pretty sure the current MP counts are chosen to match current technology.

I would also think that lenses keep up with high resolution sensors. As long as aliasing artefacts like jaggies, false detail staircase effect are visible the imaging system has potential for higher resolution sensors.

But, diffraction sets a limit on the optical system at least when we need to stop down. A way to get around that is increasing the size of the sensor and using techniques to extend stop at non diffracted apertures, like using "Scheimpflug" and stacking. Technical cameras with tilts and proper workflow come to mind…

Another thing is, in film times we were shooting Tri-X, but for best image quality we used Panatomic-X. For larger formats we used 120 film (Pentax 67 in my case), than we had Technical Pan. High resolution sensors are a bit like high resolution film, although they often have good high ISO performance due to their noise free readout.

Getting back to 80 MP digital, I would say that there may be a visual advantage of say using an 80 MP system over a 50 MP system in large prints (say from 30"x40" and up). Visible or not may depend on processing, viewing conditions and the viewer. A 120/160/240/360/520 MP system would have an advantage over that 80 MP system - how much advantage? It depends.

Just to say, it is easy to test, just make a stitched pano with your 80 MP back vertically mounted, that would essentially double the number of pixels to 160MP (if keeping field of view). Than just print it at you preferred size. I am pretty sure that you have done that.

There is a very good reason to use the best technology available and that is that shooting opportunities don't come back.

Best regards
Erik

While I agree, the problem I have is when I take a shot I don't know how much will be good enough ...

Ctein had an interesting article  about where things might be going over on the online photographer.  Always an interesting read.  I responded to a featured comment which sort of talked about the good enough idea, which is also featured below the article. Rather than rewriting my entire concept here, those interested can take a look.

[landscapephoto]

Ctein had an interesting article  about where things might be going over on the online photographer.  Always an interesting read.  I responded to a featured comment which sort of talked about the good enough idea, which is also featured below the article. Rather than rewriting my entire concept here, those interested can take a look.

The article is interesting, but the idea that technology still has plenty of room for improvement omits a hard reality: the quantum nature of light. With the present high iso fashion, we are pretty close to counting individual photons. For example, at ISO6400, typical values collected by your individual camera pixels will be between 1 and 1000 photons.

[Christoph B.]

I think the only reasons for a high MP count are the ability to have a lot of flexibility when it comes to cropping and the quality when it comes to big prints.
Sure, you can get a usable 100cm / 40inch print from say..a 20 or 30MP file - but 60 or 80MP offer much more detail and it appears sharper on close inspection (even though all are 'sharp').

Just think about buying/selling prints - you would have a hard time selling blurry, interpolated photos compared to detailed prints where you can see the leaves on a distant tree...

[Bart_van_der_Wolf]

Ctein's article is interesting, but I also feel he is a bit on the optimistic side. He is obviously right about bayer sensor wasting a significant part of the light, but we have not yet seen a better solution.

Hi Erik,

I'm not too sure about his comment: "Those Bayer arrays we use today are only about 25% efficient, at best. Those colored filters toss out out three-fourths of the light before it even gets to the sensor". It sounds like he is falling in the old trap that is often used by Foveon proponents.

It is true that to separate Chromaticity from Luminance, a filter is used, and filters absorb or reflect part of the spectum. However, in the Bayer CFA, the filtered out colors, say 2/3rd in a worst case of full spectrum light, are added back by interpolation from neighboring sensor elements. So no light is wasted, it is just partly sampled and partly interpolated, and the sum is larger than the filtered part alone.

In addition, some parts of the spectrum are filtered out, but maybe they do not exist in the subject to begin with, in its color we are capturing.

The only thing that can be said is that filters do lose some of the incoming energy (they are not spectrally pure, peak transmission is less than 100%) and convert it it to heat. I'm sure there will be methods available to separate color in a more efficient way, but the cost will be relatively high for the gain that can be had.

I would also think that lenses keep up with high resolution sensors. As long as aliasing artefacts like jaggies, false detail staircase effect are visible the imaging system has potential for higher resolution sensors.

But, diffraction sets a limit on the optical system at least when we need to stop down. A way to get around that is increasing the size of the sensor and using techniques to extend stop at non diffracted apertures, like using "Scheimpflug" and stacking. Technical cameras with tilts and proper workflow come to mind…

Lenses and diffraction will limit the ultimate resolution we can get, but denser sampling also increases the potential for Deconvolution Capture sharpening. The more accurate we can model and analyze the (spatially variant) image blur, the more of the original input signal/resolution we can restore.

There is a very good reason to use the best technology available and that is that shooting opportunities don't come back.

Indeed, and because better technology often also produces better image quality.

Cheers,
Bart