Optimally shot and then Enhanced to give an approx 240mp file...

[tashley]

Advertisement (gone after registration)

In another thread (RAW files access) I shot and showed a series of RAW files at each resolution and most major ISOs because some people had asked for that. 

I had purposefully pushed exposure far to the right of the histogram so as to give crazy 'blinkies' on the camera review screen, because I wanted to see if the lower resolution versions of the file protected the highlights better. You can judge that for yourself, if interested, by looking at the thread.

Anyhoo I ended up with a nice clean ISO 64 full res files shot in diffuse natural light with a good lens (50 APO) on a tripod with delayed release and it is 2 1/3rd stops over the exposure the camera wanted to give it but remarkably in LR there is little clipping and so I thought, let's give it the Enhanced treatment in LR.

Posted here: A thumbnail, a forum-sized crop of the up-resed file with my preferred sharpening etc in Lightroom and a link to the 60mp RAW file

I literally could not upload the Enhanced DNG to Zenfolio because the file size is too big at a tad over 900mb but if anyone is interested they can run the experiment themselves using the above linked RAW. I was literally gobsmacked. My hunch is that an 8 foot (yes, 8 foot) print would look none too shabby.... there’s a bit of artefacting going on from the upres but I bet most people wouldn’t notice it. I only see it clearly  in the shoulders of the cans in the crop, and nowhere else in the image is it obtrusive.

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

Edited January 23, 2022 by tashley

[adan]

2 hours ago, tashley said:

My hunch is that an 8 foot (yes, 8 foot) print would look none too shabby....

Well, one can enlarge almost any image to any size - provided one puts a limit on how close people can get to it.

Just ask outdoor-billboard advertisers.

If one gives human eyes free rein to get as close as possible and still focus their eyes (but no magnifying lenses allowed ) one needs to provide the eye with about 381 pixels per inch (15 ppmm) on the print surface.

The 9528 pixels of the M11 at full resolution will therefore allow a print of 25 inches wide (2 feet/63.5cm).

OTOH that is "theoretical ideal perfect eyes" - in reality one can often get away with 240 pixels per inch, or even as low as 180 pixels per inch (M8 10-Mpixel images enlarged to 21x14"/53x35cm - I sell a lot of those).

9528 pixels at 240 lines per inch = 27.2"/69cm = 2.3 feet long dimension

9528 pixels at 180 lines per inch = 53"/2067cm = 4.41 feet long dimension

However if you can keep viewers at least 5-6 feet away, yeah the M11 would produce a "sharp to the eye" 8-foot x 5-foot-4-inch print. With a little leeway for "leaners."

And if one assumes "normal viewing distance" that equals the diagonal of the print - 9.6 feet away for an 8-foot print, 14.5 feet away for a 12-foot print - yeah, it is probably almost impossible to NOT have a really crisp "looking" print from a well-handled M11 photograph. Just don't forget the railing.

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[adan]

2 hours ago, adan said:

9528 pixels at 240 lines per inch = 27.2"/69cm = 2.3 feet long dimension

Math correction: 39.7"/100cm = 3.3 feet

[tashley]

My math is as follows:

For the vast majority of non-peepers 180 dpi is fine. I would go lower only in exceptional circumstances. But as long as the capture is at low ISO and has no shake and was taken with a good lens and is properly proofed, 180 or more. 
 

An Adobe Enhanced file from the M11 Is 19056 pixels on the long side. And it looks like that was the native capture resolution in many cases and to most eyes. I’m happy saying that because of all the 150mp IQ4 files and files upreses from them, or stacked files etc. that I’ve peeped at on my monitor and then printed. 
 

19056 printed to 8” is therefore 198dpi and I’m pretty sure you could stand three feet away without it looking blurry. 
 

So I was being conservative! 🤣😂🤣

Edited January 24, 2022 by tashley

[adan]

Nothing wrong with your math - but I just question the assumption that "it looks like that was the native capture resolution in many cases and to most eyes."

Looking at examples and commentaries of Adobe Enhanced files, it can handle sharper edges - but the textural details don't seem to really improve all that much.

Maybe 35-45% instead of the 100% the "doubling" would claim. https://photographylife.com/reviews/adobe-super-resolution

No algorithm can create and improve fine hairs or grass blades or sand grains or pores that were never there in the first place.

Especially if "double-dipping" - faking up the resolution (admittedly with better algorithms than we've seen before) - AND printing at 180 ppi instead of 240-plus ppi.

However the old saying probably still applies - "You can fool some of the people all of the time, and all of the people some of the time....."

And it might fool me at 6 feet.

[sebben]

I'm surprised to see so much chromatic aberration in the 50mm APO!

[Al Brown]

Advertisement (gone after registration)

We have had this "how much can I enlarge and at what dpi" discussion in the forum somewhere as I recall.

Edited January 24, 2022 by Al Brown

[tashley]

Me too! I am looking forward to Sean Reid taking a look at some of these lenses to see if new issues are revealed on the M11. Resolution, however, is great...

[tashley]

4 hours ago, adan said:

Nothing wrong with your math - but I just question the assumption that "it looks like that was the native capture resolution in many cases and to most eyes."

Looking at examples and commentaries of Adobe Enhanced files, it can handle sharper edges - but the textural details don't seem to really improve all that much.

Maybe 35-45% instead of the 100% the "doubling" would claim. https://photographylife.com/reviews/adobe-super-resolution

No algorithm can create and improve fine hairs or grass blades or sand grains or pores that were never there in the first place.

Especially if "double-dipping" - faking up the resolution (admittedly with better algorithms than we've seen before) - AND printing at 180 ppi instead of 240-plus ppi.

However the old saying probably still applies - "You can fool some of the people all of the time, and all of the people some of the time....."

And it might fool me at 6 feet.

I've generally had the same impression - that Enhance isn't much use for irregular, organic, high-frequency material and can in fact make it look downright odd! but it can be very good indeed for some subjects.

As to what level of detail is enough for whom, after years of peering into screens and at prints at close range - and after a couple of years out of photography spending time painting instead, I've come to a different view, which is that the vast majority of non-photographers (non serious at least!) just don't notice as long as it's reasonable. No excuse to be sloppy and better is always... better but for most people good is good enough. Sure, Ming can sell ultraprints at astonishingly high DPI and I've had a go at making them myself and they're great - but I think it's largely an enthusiast thing. The Leica store in Mayfair has prints on the wall which by the standards of anyone on this forum should never see a print beyond 10x8" and yet they blow them up large and because they're great shots, they look great.

I have (luckily) a HUGE Burtynsky on the wall at home. It's 80 inches wide. And I know from when I bought it that it can't have been shot at more than 80mp and because its shot from a plane it sure as hell isn't stacked or stitched. It's been much admired by lots of visitors and I am the only one I have ever seen standing with his nose to it, tutting quietly...

So here's bit of fun. Please, no one take it too seriously because I don't want to get into a great debate about methodology. It's rough and ready and yes I know you can't compare apples to oranges but sometimes there's only room in your lunchbox for one piece of fruit!

One of these is M11 with 50APO at f5.6. The other is IQ4 with Rodenstock 40HR at f8 on an Alpa field cam. Both on the same tripod (Gitzo 3 series CF with Cube) and with entirely natural diffuse light. Both bracketed at IO 64 and the chosen frame was the one that was hardest up against the right of the histogram but with only tiny highlight clipping. 

The M11 file was Enhanced and then down-resed to the pixel width of the Phase file. The Phase file was exported from C1 as a 16 bit tiff with no sharpening or NR and then cropped to the height of the M11 file and then given the exact same sharpening and NR as the M11 file. Both then exported as 92% quality JPEG in ProPhoto.

It is clear after brief inspection which is which. I have made my own judgements but I'm interest in those of others!

https://tashley1.zenfolio.com/img/s/v-10/p834122565.jpg

https://tashley1.zenfolio.com/img/s/v-10/p576049924.jpg

 

Edited January 24, 2022 by tashley

[Jon Warwick]

....thanks for the new comparison, [AND THANKS FOR THE UPDATED LINKS! ]......my feel is both outputs look very similar indeed.  The jpeg ending 565 seems to have slightly fewer digital artefacts, and might look at bit more "relaxed" in its output ...so I'm guessing that is the IQ4?

That aside, I have never tried Enhance in PS, so just had a practice on one of my own images. My initial observation is it looks like a lot of extra sharpening is possibly applied as part of the process, so it might not be quite to my taste, but it's early days of me trying it.

I have to say, for the completely static "studio" image that you have taken above, this is perhaps where multi-shot mode ("pixel shift") would be absolutely ideal ....so something like the SL2 / S1R / GFX100S would probably all work really well here. I think of multi-shot mode as correcting the digital problems that can crop up with Bayer filters (like false color, moire) and this gives the appearance of higher resolution and also less noise.  As such, I regard multi-shot modes as giving very similar benefits to what we enjoy with Monochrom sensors including more fine detail despite the underlying sensor (in terms of megapixels) naturally having identical resolution in both "normal" and "pixel shift" mode.

 

 

Edited January 24, 2022 by Jon Warwick
Updated for having looked at links in #9!

[tashley]

2 hours ago, Jon Warwick said:

....thanks for the new comparison, [AND THANKS FOR THE UPDATED LINKS! ]......my feel is both outputs look very similar indeed.  The jpeg ending 565 seems to have slightly fewer digital artefacts, and might look at bit more "relaxed" in its output ...so I'm guessing that is the IQ4?

That aside, I have never tried Enhance in PS, so just had a practice on one of my own images. My initial observation is it looks like a lot of extra sharpening is possibly applied as part of the process, so it might not be quite to my taste, but it's early days of me trying it.

I have to say, for the completely static "studio" image that you have taken above, this is perhaps where multi-shot mode ("pixel shift") would be absolutely ideal ....so something like the SL2 / S1R / GFX100S would probably all work really well here. I think of multi-shot mode as correcting the digital problems that can crop up with Bayer filters (like false color, moire) and this gives the appearance of higher resolution and also less noise.  As such, I regard multi-shot modes as giving very similar benefits to what we enjoy with Monochrom sensors including more fine detail despite the underlying sensor (in terms of megapixels) naturally having identical resolution in both "normal" and "pixel shift" mode.

 

 

Interesting, and thanks for the observations Jon. I largely agree though there's an awful lot of 'spot the difference' one could play here. Both cameras were focussed in LV at shooting aperture but nonetheless, though the '0' is in perfect focus on both shots, one carries the its DOF more forward than the other. Also both lenses have slightly different field curvature characteristics and this is evidenced by the fact that the Rodenstock has more or less depth of field depending on where you look in the image and that can make it look less effective overall.

I agree with you that the Enhanced one looks a bit over sharpened - it only bothers me around the edge of the sage leaf where it is OOF so it might be worth giving the file a bit less sharpening after enhancing but one thing to note is that the Enhanced file, BEFORE it was downsized, displays that tendency somewhat less so it might have been better to leave sharpening entirely off until after both the enhance and the downsize routines.

Colours seem remarkably similar to me. What do you think?

Edited January 24, 2022 by tashley

[Jon Warwick]

Yes I think the colours between the two look highly similar indeed.  Separately I also think the colours from the DNGs in the Raw thread that you kindly posted are magnificent, it’s one thing (beyond the less sharpened naturalness of the output and not looking overtly “digital”) that is standing out to to me about the M11, especially the greens that look so natural and with so many tonal gradients within a given colour that sculptures a depth to the image; the greens are also without that excessive yellow that seems to be endemic in other digital files I’ve used over the years. 

Edited January 24, 2022 by Jon Warwick

[tashley]

6 hours ago, Jon Warwick said:

Yes I think the colours between the two look highly similar indeed.  Separately I also think the colours from the DNGs in the Raw thread that you kindly posted are magnificent, it’s one thing (beyond the less sharpened naturalness of the output and not looking overtly “digital”) that is standing out to to me about the M11, especially the greens that look so natural and with so many tonal gradients within a given colour that sculptures a depth to the image; the greens are also without that excessive yellow that seems to be endemic in other digital files I’ve used over the years. 

I totally agree; there’s nothing worse than dragging down on saturation with an eyedropper over something that should be green and seeing the yellow slider move left…

[tashley]

22 hours ago, adan said:

Well, one can enlarge almost any image to any size - provided one puts a limit on how close people can get to it.

Just ask outdoor-billboard advertisers.

If one gives human eyes free rein to get as close as possible and still focus their eyes (but no magnifying lenses allowed ) one needs to provide the eye with about 381 pixels per inch (15 ppmm) on the print surface.

The 9528 pixels of the M11 at full resolution will therefore allow a print of 25 inches wide (2 feet/63.5cm).

OTOH that is "theoretical ideal perfect eyes" - in reality one can often get away with 240 pixels per inch, or even as low as 180 pixels per inch (M8 10-Mpixel images enlarged to 21x14"/53x35cm - I sell a lot of those).

9528 pixels at 240 lines per inch = 27.2"/69cm = 2.3 feet long dimension

9528 pixels at 180 lines per inch = 53"/2067cm = 4.41 feet long dimension

However if you can keep viewers at least 5-6 feet away, yeah the M11 would produce a "sharp to the eye" 8-foot x 5-foot-4-inch print. With a little leeway for "leaners."

And if one assumes "normal viewing distance" that equals the diagonal of the print - 9.6 feet away for an 8-foot print, 14.5 feet away for a 12-foot print - yeah, it is probably almost impossible to NOT have a really crisp "looking" print from a well-handled M11 photograph. Just don't forget the railing.

PS I have no idea where you get the 381 number from but take a look at the  resolutions of current state of the art screens. Most of them are in the region of 220-240 PPI.

It might well be that you have found some source that says that at a certain fairly close distance 318 is the most that the human eye can resolve. Not most eyes, I would say. And pixels are not a measure of printed material output….

https://blog.eyewire.org/what-is-the-highest-resolution-humans-can-distinguish/

Other cherry picked opinions are available 😂

Edited January 25, 2022 by tashley

[adan]

4 hours ago, tashley said:

PS I have no idea where you get the 381 number from but take a look at the  resolutions of current state of the art screens. Most of them are in the region of 220-240 PPI.

It might well be that you have found some source that says that at a certain fairly close distance 318 is the most that the human eye can resolve. Not most eyes, I would say. And pixels are not a measure of printed material output….

https://blog.eyewire.org/what-is-the-highest-resolution-humans-can-distinguish/

I assume by "art screens" you mean the stand-alone LCDs that can be used simulate framed prints? I don't know that a consumer product is a "standard" for anything.

As to "381," that comes from one of Erwin Puts' essays in his Leica Lens Compendium. p. 87 - although it really should be "406" (my memory error):

"It has been established that any object smaller than 1/16th mm will be seen by the eye as a single point, [and] two small objects separated by at least 1/16th mm will be identified as two separate objects."

16 times 25.4 (millimeters per inch) comes out....... 406 lines/dots/pixels per inch

Which does correspond to my own practical tests of glossy silver-gelatin paper resolution - about 400 lines/dots per inch. Slightly better than the usually-recommended digital output of 300 lines per inch.

And to the data in the web site you yourself just directed me to (red quotes):

Quote

Q: What is the highest resolution humans can distinguish?

A: “The visual resolution of the human eye is about 1 arc minute.

At a viewing distance of 20″, that translates to about 170 dpi (or pixels-per-inch / PPI), which equals a dot pitch of around 0.14 mm.  LCD monitors today have a dot pitch of .18mm to .24mm.

A 30″ monitor with a 16:9 aspect ratio would be sized around 26″ x 15″. To achieve 170 dpi, it would need a resolution of 4400 x 2600 pixels [a MacBook Retina display is 2880 x 1880].

If you want to lean in closer, say to 10″, and still not see the pixels, then you would need to double that resolution.”

170 dpi/ppi doubled is.....340 dpi/ppi.

300 or 340 or 381 or 406 - take your pick. That range is the target resolution for photographic works to be viewed at the closest unaided viewing distance - and reproduce every detail as sharp points ("infinite" detail, as far as the eye is concerned).

..............

Pixels certainly "are a measure of printed material output."

As someone who was in the publication industry (newspapers and magazines) during the transition to using digital files in pre-press work (1985-2008), I saw it all happen. Scitex to Leaf to Coolscan to Pagemaker to Quark Xpress to InDesign to an Agfa/Hyphen imagesetter. Photoshop 1/2/2.5/3/4/5. I worked with all those pre-press systems in mind.

The question came up as to just how high a pixel-resolution was needed for high-end glossy magazine reproduction (e.g. National Geographic) using a 150-line half-tone screen. And the industry decision was: 2 image pixels for each half-tone line, or 150 x 2 = 300 pixels per inch.

Nat. Geo's first* published picture from a "general purpose" digital camera was made with a Nikon D100. 6 Mpixels (3000 x 2000). 214 ppi when printed across a two-page spread (10" x 14"). Decent - but clearly a bit soft (darn Nikon AA filtering ) compared to their usual laser-scanned Velvia/Kodachromes.

(*that doesn't count the various remote off-planet digital images captured and transmitted from Viking/Mars, etc. Those were published almost as soon as they arrived - 1976).

From the other side, chemical prints from digital using laser LightJets, Lamdas and similar came to the same number: 300 ppi to get close to the resolution of optical prints on silver gelatin. Not quite as good as silver paper with an excellent enlarging lens (406 lines) - but pretty close.

Now - we do agree that there are times that that can be fudged quite a lot, depending on the subject, situation, or the intended viewing distance - if one can corral the audience.

Edited January 25, 2022 by adan

[jonoslack]

It's wonderful to have you back Tim! (I'm sure Andy thinks so as well!)

I thought the jpg finishing with 65 was quite noticeably better when zoomed in . . . but the colours are remarkably similar.

All the best

Jono

[tashley]

5 hours ago, jonoslack said:

It's wonderful to have you back Tim! (I'm sure Andy thinks so as well!)

I thought the jpg finishing with 65 was quite noticeably better when zoomed in . . . but the colours are remarkably similar.

All the best

Jono

And it's wonderful to BE back! Thank you.

You're right about the files but out of curiosity, I think there are some downsides and upsides and that though '65' is the winner, it's not a comprehensive win. Out of interest, what are the areas you think it wins in in particular? And, from your experience, would you anticipate that an approx 180DPI print from the file ending in '24' would look good?

Edited January 25, 2022 by tashley

[tashley]

10 hours ago, adan said:

I assume by "art screens" you mean the stand-alone LCDs that can be used simulate framed prints? I don't know that a consumer product is a "standard" for anything.

 

Nope, in fact I wrote 'state of the art' screens - referring to displays like the Apple Pro Display XDR and the various current generation MacBook and iPad and iMac Pros, which all shoot in the area between 218 and 250 PPI. The 'frame' style TV screens, which I did not refer to at all, have much lower resolution but look OK from normal TV viewing distances.

10 hours ago, adan said:

170 dpi/ppi doubled is.....340 dpi/ppi.

That's if you want to look from a distance of 10" or closer. My point is that pretty no one healthy does. The reason the Mac Pro Display XDR has a lower resolution than the MacBooks and iPads is that people sit further back from a monitor than they do from an iPad or laptop screen. In fact another 'state of the art' manufacturer of high-end proofing monitors, Eizo, has mostly 109 PPI monitors.

10 hours ago, adan said:

Pixels certainly "are a measure of printed material output."

They very very definitely are not. I can state that confidently and as a fact. Pixels are on screens, dots are on giclee prints. PPI is often used loosely but in fact it refers ONLY to the number of pixels per inch of a screen, though people can quite correctly state that it is their opinion that if a print viewed at 100% on screen at a resolution of 250 PPI looks great, then it will look great when printed if you print it to [number of pixels wide the file is / 250] physical inches wide. So for example if I look at a 6,000 pixel wide print on a 200PPI monitor and it looks great then I know I can print it to 30" wide.

Print resolution, for printing processes that use dots or drops, are expressed in DPI (dots per inch) and a good quality large format giclee printer will produce something in the region of 2,400 DPI so theoretically you could make a 3.97" wide print that would have more detail under a loupe than, say, a 5" print. SO where do you stop?

Michael Reichmann (RIP) at Luminous landscape did a fantastic long series of workflow videos (subscription only) and he covered all of this in as much depth as you'll find anywhere. As a side comment after much experimentation and discussion with people who make pro quality prints all day long for a living, he concluded that if you had a really good quality file that you wanted to print lager than 180DPI, in other words to say 160 or 140 DPI, you are nearly always better off letting the printer and its drivers do the interpolation for you. However that was 'back in the day' and the up-resing software options have improved greatly since then.

In short, there's nothing 'fudged' about prints made well from good source files are resolutions as low as 180 DPI. Nor do viewer have to be 'corralled' into viewing them. They are absolutely standard fare amongst some of the very, very best art photographers currently working today - such as Burtnksy and Gursky - who often sell very large prints indeed for extremely large amounts of money. They aren't counting Erwin's angels on pinheads, they're counting the $$$ 😂😂😂

So for the Burtynsky print I referred to above: it is from his Dryland Farming series, released in 2012 so with the best will in the world it cannot have been shot with more than 80mp which with a Phase One back would be around 130 DPI... and he is at the very top of the tree.

In the end, we all have different rendering intents. If one wants to produce the tightest looking print quality imaginable so that people can get to less than 10" away and marvel at the detail, then that's fine. I merely contend that from my experience, the only people who ever do that are some photographers. The vast vast majority of people who buy photographic images to hang on their walls will never get anywhere near that critical. For them, often, they want something bigger and they are trading something they do care about (size) for something they don't even know they might think about (resolution). So if 130DPI is good enough for Burtynsky, it's good enough for me!

 

 

 

 

Edited January 25, 2022 by tashley

[tashley]

18 minutes ago, tashley said:

 

 

Edited January 25, 2022 by tashley

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[jonoslack]

1 hour ago, tashley said:

 

In the end, we all have different rendering intents. If one wants to produce the tightest looking print quality imaginable so that people can get to less than 10" away and marvel at the detail, then that's fine. I merely contend that from my experience, the only people who ever do that are some photographers. The vast vast majority of people who buy photographic images to hang on their walls will never get anywhere near that critical. For them, often, they want something bigger and they are trading something they do care about (size) for something they don't even know they might think about (resolution). So if 130DPI is good enough for Burtynsky, it's good enough for me!

 

 

 

 

Funny you should say that - I was talking to a photographer who had sold an extremely large amount of money's worth of large 'fine art' prints over the last few years. He said that the  things you absolutely knew were that if someone went up close to a print and looked carefully :

1. They were a photographer

2. They weren't going to buy anything!

All the best

Edited January 25, 2022 by jonoslack