35mm Scans VS FF digital

[138S]

Ektar beats the 850 in all but shadow detail.

Hmmm... a side by side would be interesting. I would say that Ektar does not reach the D850 resolving power by some margin.

Parkin measured that with Portra

Mamiya 7 Portra (microscope) 115Mp

If we apply the proportionality, and with the suposition that color film is the limiting factor, then a 36x24mm shot would yield around 23.7MPix effective with Portra. Ektar won't be much far.

And this is in the microscope, once you count the losses in the scanning then you may easily loss a 50% of those 23.7MPix even with a 8000dpi drum scan https://www.onlandscape.co.uk/2011/12/big-camera-comparison/ to 12 or less perhaps

See this:

From the 157MPix a 6x7cm frame has under microscope, the 8000dpi scan extracted 80MPix, and the 4000dpi drum scan 51MPix effective, this would yield 8 to (say) 16 MPix effective in the 35mm frame, depending on if the drum scan is 4000 or 8000dpi

The D850 yields 35 to 40 MPix effective with a good lens...

 

[George Mann]

Hmmm... a side by side would be interesting. I would say that Ektar does not reach the D850 resolving power by some margin.

Parkin measured that with Portra

Mamiya 7 Portra (microscope) 115Mp

If we apply the proportionality, and with the suposition that color film is the limiting factor, then a 36x24mm shot would yield around 23.7MPix effective with Portra. Ektar won't be much far.

I have conducted these tests myself. Portra struggles to deliver more than half of the resolution of Ektar.

 

[138S]

I have conducted these tests myself. Portra struggles to deliver more than half of the resolution of Ektar.

This is not my experience... also Parkin rates (under microscope) 157Mp for TMax and 115Mp for Portra, Ektar is under TMax so it cannot double Portra, Ektar may be slightly over Portra but still quite far from TMX.

But anyway Ektar resolves well less than TMX, and TMX delivers 51MPix effective (Parkin) after a 6x7cm drum scan at 4000dpi, this would be 12MPix for a TMX 35mm frame, or 19MPix if it is a 8000dpi drum scan. Ektar will be well under those TMX numbers, making good that 8 to 16 MPix interval for Ektar.

 

[George Mann]

My tests show that Ektar outresolves Tmax 100 in 35mm, just as Ektachrome outresolves a D850!

Where besides flawed scans is your data on Ektar coming from?

 

[138S]

My tests show that Ektar outresolves Tmax 100 in 35mm, just as Ektachrome outresolves a D850!

Where besides flawed scans is your data on Ektar coming from?

There was an ancient Ektar ISO 25 that was sharper than TMX, but Ektar 100 is less sharp than TMX

There are many tests around:

http://www.cacreeks.com/films.htm,
http://www.tmax100.com/photo/pdf/film.pdf
https://www.onlandscape.co.uk/2014/12/36-megapixels-vs-6x7-velvia/
http://jpbuffington.com/?p=167
http://www.boeringa.demon.nl/menu_technic_ektar100_resolution.htm

Parkin rates Velvia at 24MPix microscope, which would deliver 16 after a drum scan. I doubt Ektarchrome is sharper than Velvia, so IMO not outresolving D850. Parkin also finds 35MPix for Adox CMS 20 after scanning, probably yielding 50MPix (or limited by the taking lens) in the microscope, but this is wild monodisperse microfilm...

TMX vs Ektar 100

 

[George Mann]

There was an ancient Ektar ISO 25 that was sharper than TMX, but Ektar 100 is less sharp than TMX.

Sharpness is irrelevant!

Parkin

Forget this guy's results. Try Henning Serger.

 

[George Mann]

This link you posted is a good one however. Pay very close attention to the variability Henning spoke about.

http://jpbuffington.com/?p=167

My tests were of real world scenes of very fine detail under optimal sunlight conditions.

It does surprise me that TMX achieved 150 lp/mm, but I expected it to outperform Ektar in laboratory tests.

 

[George Mann]

Let me explain further by stating that TMX tends to get a little course under highly dynamic conditions, which tends to obscure the finest details that it can capture.

Delta 100 does better here.

 

[George Mann]

I also need to remind you all that digital cameras utilizing color sensor arrays have an actual resolution of 1/3 of their claims at best.

The D850 has a native resolution of around 16mp!

 

[138S]

Forget this guy's results. Try Henning Serger.

Before trying to discredit a source like Parkin, read well the test... (https://www.onlandscape.co.uk/2011/12/big-camera-comparison/)

Sharpness is irrelevant!

Sorry if those words were inaccurate, please subtitute by "resolving power". If sharpness is irrelevant (which can be debated), why do we discuss about all that?

Anyway testing film is quite complex because contrast-exposure matters a lot, as sensitive particles are of different size, testing what a DSLR does is straighter.

__________

IMO, last model DSLRs do surpass well the practical resolution of common pictorial 35mm film, if you don't go to things like CMS 20, if you consider the loss in the scanning (even a drum) then there is no doubt.

But this is irrelevant for a different reason: we don't shot film because of pixel peeping. What the medium resolves ultimately is not that useful in the presentation because at these levels image quality is degradated, and the ultimate rating may not be well related to the practical result.

From film we have an "organic" feel, a dedicated spectral response that can be quite specialized for the job, and results we love. Regarding resolving power, we have what we want, I don't shot 8x10" because of resolving power... that plain overkill comes as a bonus... when my digital colleagues grasp their heads I tell them that hose 400 or 600MPix effective are there but I don't know why

__________

35mm film is amazing, a 35mm SLR provides amazing agility. But... Want ultra Pro results ? Go MF !!! like many Pros did since ever.

MF has no compromise in the IQ, you won't get much more in practice with LF (beyond the look). Also MF ha a distinctive look (focus roll-off) that makes a difference compared to the small 35mm and FF. with MF you have less shots in the roll, but you waste quite less frames. In LF you may target not wasting a single shot.

35mm is powerful, what brings the film medium to total excellence is MF, as eye capability is mostly surpased when covering the print with our field of view.

 

[Alan Edward Klein]

For what it';s worth, when I bracketed Portra, bracketed +1, 0, -1, the Epson V600 scanned flat results had differences in color rendition. While each could probably stand on its own, there were obvious hue differences when comparing all three.

 

[George Mann]

Before trying to discredit a source like Parkin, read well the test.

I did. Parkins findings are well below mine and Hennings.

If sharpness is irrelevant (which can be debated), why do we discuss about all that?

Because people overstate its importance.

Anyway testing film is quite complex because contrast-exposure matters a lot

Which is why Ektar was able to edge out TMX, which I explained in a previous response.

IMO, last model DSLRs do surpass well the practical resolution of common pictorial 35mm film

My tests show a noticeably greater amount of fine detail from the film I had tested.

35mm film is amazing, a 35mm SLR provides amazing agility. But... Want ultra Pro results ? Go MF !!! like many Pros did since ever.

I get sufficient results shooting 35mm slide film. In fact, it is much more realistic than the results from a digital slr.

 

[138S]

Which is why Ektar was able to edge out TMX, which I explained in a previous response.

George, it would be interesting if you can post your results... What exposure for the white bars ? What one for the black ones ?

Marco Boeringa (http://www.boeringa.demon.nl/menu_technic_ektar100_resolution.htm) showed that Ektar and Portra (VC version) had a very similar yield, samples are scanned 8000 dpi in an Imacon (flextight), also showing a clearly higher MTF in the TMX case.

Also the ancient (2008) Apha 900 (5.91 µm pixel pitch) was showing something (the doc explains why the bars look smaller, film is relatively oversampled).

 

[George Mann]

Scans can't show you what film is capable of!

 

[138S]

I agree, of course, post 101 shows the difference between a microscope evaluation and high end scans, we may loss a 10% to 30% linear resolution, in MPix it would be the square. A microscope shows exactly what film recorded...

Still those 8000 dpi flextight scans are top notch, probably they are 6000 dpi effective or beyond. If ektar and portra look similar there then we may concede that in practice we get the same, ultimate numeric resolving power (not that useful pictorial concerns) should be seen in the microscope.

Anyway scans show what it is to be printed from CN film. Today darkroom RA-4 optic printing is testimonial, nearly extinguished... the vast majority is displayed in monitors or digitally printed, so the scanned yield is a principal metrics.

Personally, I've inspected with a good microscope side by side negatives made with ektar, portra, cinestill 50-800 (vision 3) to see the color clouds, scenes but not targets, identifying the same feature in the different situations, what I perceived is that most of the times the shot does not exhaust the film capability, also with digital it happens the same... well, all what is shot handheld without flash... all what's not totally flat... all what's not shot at the optimal aperture... Real photography is not shooting flat targets on a technically vibration isolated tripod.

For sure this debate is interesting to learn the medium capability, but this is more academic that practical.

 

[Alan Johnson]

I don't understand why the results you are quoting with scanned film are so poor.
Even with my inexpensive Plustek 8100 I got results comparable to figures often quoted for 36 mp digital sensors, IIRC about 80lppm
My result was similar to that obtained by the professional scanner dealers for true, not nominal resolution.
https://www.photrio.com/forum/threads/resolution-of-the-plustek-8100.156680/

 

[removed account4]

I shoot FF digital, and I'm learning about printing digital negatives etc. A question that I have though is, what are the differences between a scanned 35mm negative, and a full frame digital file? Why would someone prefer to start with a scanned digital file, instead of a straight digital file?

Hey CC
getting back to the original question you asked. ....

it depends on the size of the digital negative you need to make. You need to remember there are people who claim you can't enlarge a 35mm negative past 5x7 and 8x10 well its pushing the limits of the film, and then there are other people who gleefully enlarge a 35mm negative to 30x40", so asking which is. better is kind of personal taste. its not hard to get a 8x10 print from a digital phone file ( yes I have done that ), and the difference between starting from a scanned file instead of straight digital is like asking if you want whipped cream and nuts on the ice cream sundae. some people like one or the other, other people like none or both...
the way I treat these things is that a digital file from phone or camera, or a scanned piece of film or scanned paper negative are all capable of making great digital negatives for anything you want to print.

In the end none of it really matters, or it matters so much that it is an obstacle.

Have fun !
John

 

[Lachlan Young]

I don't understand why the results you are quoting with scanned film are so poor.
Even with my inexpensive Plustek 8100 I got results comparable to figures often quoted for 36 mp digital sensors, IIRC about 80lppm
My result was similar to that obtained by the professional scanner dealers for true, not nominal resolution.
https://www.photrio.com/forum/threads/resolution-of-the-plustek-8100.156680/

Probably because they're quoting other people's results (through filters of their own agendas) at each other, rather than on the basis of real practical experience. Much to my amusement, you can see one of them attempting to cover his ass after spending years badmouthing Hasselblad/ Imacon scanners.

 

[Down Under]

it depends on the size of the digital negative you need to make. You need to remember there are people who claim you can't enlarge a 35mm negative past 5x7 and 8x10 well its pushing the limits of the film, and then there are other people who gleefully enlarge a 35mm negative to 30x40", so asking which is. better is kind of personal taste. its not hard to get a 8x10 print from a digital phone file ( yes I have done that ), and the difference between starting from a scanned file instead of straight digital is like asking if you want whipped cream and nuts on the ice cream sundae. some people like one or the other, other people like none or both...
the way I treat these things is that a digital file from phone or camera, or a scanned piece of film or scanned paper negative are all capable of making great digital negatives for anything you want to print. ... In the end none of it really matters, or it matters so much that it is an obstacle.

Sensible advice from a poster who likes to tell it as it is, if at times blunt. But not what some posters like to be told. Personal agendas, yes. Opinions quoted as facts. Rolleiflex and Hasselblad users claim they can print murals from their Tri-X negatives, 35mm shooters are obsessed with perfect 16x20s. Why? Nobody seems to consider this one word.

I can't remember the last time I made a print larger than 8x10. Some time in 2012, I think. My darkroom work prints for both formats are 4x5s (one 8x10 sheet cut into four pieces) as work prints and 5x7s as gift prints. Ditto scanning and inkjet printing.

The only good way to test a scanner, any model, is to pick a negative you want to scan and make prints from, and do tests. Scan it every which way and carefully study your results. After doing this with several negatives and slides on my Plustek 7600i (a true antique!) I decided i get my best results at 2400 dpi with everything (especially sharpening) turned off and all tones set at what I consider to be the closest to the old Gamma index of .6. Even with my equally ancient Silverfast 6.6 I get acceptable results to 8x10 and as I've never really seen the need to print any bigger, this satisfies me.

I now shoot mostly digital but I still have some 30,000 not yet scanned negatives and slides of architecture subjects. Occasionally I sell color digital images to publishers of architecture books. All want large jpegs without sharpening or any color fiddling - their art directors do all that. I have sold photographs of buildings that were so dark and off-color as to be unusable, or so I thought. Published, they were more than adequate. Lesson for us all here.

Perfection in scanning and printing has its price, and some will probably scoff that a Plustek 7600i with Silverfast 6.6 is not that price. For me it suits. What do I care what the obsessives think?

Good one, John.

 

[138S]

I don't understand why the results you are quoting with scanned film are so poor.
Even with my inexpensive Plustek 8100 I got results comparable to figures often quoted for 36 mp digital sensors, IIRC about 80lppm
My result was similar to that obtained by the professional scanner dealers for true, not nominal resolution.
https://www.photrio.com/forum/threads/resolution-of-the-plustek-8100.156680/

Yes, your 80 lp/mm rating in the test would yield 24MPix if the center performace was the same in all the field, while the averaged result in all the field (accounting mid-corner) would be lower it shows the D100 potential to record 24MPix.

Your 80 lp/mm rating is not rare at all, tests like those made by C. Perez show 85 lp/mm peaks for exceptional LF lenses, but those tests overexpose the white lines by 2 or 3 stops (see the "Please Note:" notice in the Perez test).

You made a Flash exposure, this is good because you cancelled vibrations, still probably it can be an overexposed shot for the chart, nothing is reported, when it would be interesting to know the density of the white and black areas.


Film resolving power depends a lot on exposure and contrast, For example the TMX datasheet reports 63lp/mm at TOC 1.6:1 contrast and 200lp/mm at TOC 1000:1. At 1000:1 TMX resolves 10 times more megapixels that at 1.6:1 ( (200/63)2 )

To show the effect, if you could find a way to project on film 1000:1 microcontrast in all the scene that TMX 35mm frame would contain 200x2x24 x 200x2x36 = 140 MPix effective, using the Kodak's 200 lp/mm rating. For this reason a film performance rating should tell very well what the conditions are, if not we say little. Note that the 63lp/mm rating for low contrast would yield an optimistic 14 MPix instead, before taking lens and scaning degradations that it may leave the practical thing at 10 MPix effective. What contrast do we consider? 1.6:1? 30:1? 100:1? 1000:1?

If you made the shot with TTL flash metering and background was darker or farther than the chart then for sure the chart was overexposed, chart overexposure (if happened) could explain perfectly that D100 boost in the resolving power, My view is that the discrepances we find in film tests can be perfectly explined by exposure conditions, if your white lines were overexposed by two or three stops (like in the Perez test) then for sure you could get 24mpix. Film resolves well more at +3 than at -/+0.

What density has the white area of the chart? this is a key point...

Just pointing that a film test may state an impressive x10 MPix varibility (for the same fim!), depending on tested conditions.

We should understand the difference betwen shooting flat targets and real photography. IMO, in practice most of the times we take real advantage of up to around 40 lp/mm and not much beyond.

 

[138S]

I have sold photographs of buildings that were so dark and off-color as to be unusable, or so I thought. Published, they were more than adequate. Lesson for us all here.
Perfection in scanning and printing has its price, and some will probably scoff that a Plustek 7600i with Silverfast 6.6 is not that price. For me it suits. What do I care what the obsessives think?

Absolutely, the most important factor in scanning/edition is operator: sporting a good enough workflow, making an accurate optimization and having a sound aesthetic criterion.

High end scanner performance is useful only in some cases, in general pictorial film yields very diminishing results beyond (say) 40 lp/mm or 60lp/mm. While ultimate resolving power may go a lot beyond this has not much pictorial interest: as we take advantage of high resolving power we also show several kinds of noise an contrast degradation from film and from lens. Anyway, the higher the overexposure the higher the resolving power... because smaller crystals play a role...

There are some exceptions, like wild CMS 20, which is a great microfilm adaptation to pictorial usage, knowing it's limitations it is a great stuff.

All want large jpegs without sharpening or any color fiddling - their art directors do all that.

OK, a big jpg may be enough, but not always. A good Pro edition may require a 16 bits/channel source, if the scene is challenging, so tiff or the like would be better in some cases. Better using a TIFF if possible, for no compromise.

 

[chris77]

Very interesting thread, thank you!

 

[removed account4]

Good one, John.

pleasures mine ozmoose
yeah. ..
30K negatives left to scan . sounds like either a greek tragedy or lots of fun. !
"and behold ozmoose, who displayed hubris to the gods and was left to scan a never ending stack of negatives. some of the gods smiled at the display as the stack of film never receded, zeus was thoroughly impressed and gave him the gift of auto-levels (that worked)"

==
I know what you mean about small prints. classmates are consistently making beautiful mural sized images, like you, I'm happy with something small

John

 

[Alan Johnson]

.

We should understand the difference betwen shooting flat targets and real photography. IMO, in practice most of the times we take real advantage of up to around 40 lp/mm and not much beyond.

Amateur Photographer Test Report Book 11 p5
"To measure the resolution of a lens , Dr Bell [Their Lens Tester] uses it to shoot a picture of a specially designed chart [Parallel & perpendicular lines of increasing frequency] at 30% contrast and 70% contrast - probably the outer limits of of everyday shooting situations." -my square brackets.
For the Canon 135mm he obtained about 133 lppm at high contrast f5.6 and about 96 lppm low contrast.f8.
Amateur Photographer Test Report Book on the Pentax [ 50mm] p29
High Contrast , about 127 lppm f8, low contrast about 88 lppm f11
He says "The graphs show the printable resolution on T-Max 100developed in Acutol for 8.5 min at 20C"
IMO this illustrates that if a good lens is held steady [or image stabilization is used] much higher results should be obtainable in practice than your quoted 40 lppm.

 

[138S]

IMO this illustrates that if a good lens is held steady [or image stabilization is used] much higher results should be obtainable in practice than your quoted 40 lppm.

Alan, Real photography is not shooting flat targets on a wall, not at all...

If you want to see the 120lp/mm film recorded detail on the print then you have to enlarge x20 or x40 to end in 6 or 3 lp/mm on the print, if you don't enlarge that then you won't see that detail.

At this say (x30) enlargement regular pictorial film does not offer much image quality anyway, instead if you enlarge (say) x10 to have image quality then you can see only 30 to 60 lp/mm of what recorded in the film.

Also it may be difficult to capture from the scene all what film can record...

Many subjects are mostly made of continuous shades of gray, still you may have edges in the scene, OK, but when we want project a dot on film that has 6 stops more exposure than another dot that lays 0.01mm far (50lp/mm) we have several problems... In a regular scene almost nothing is in perfect focus but in the Depth of Field, casting a circle of confusion for each dot. Also you may shot handheld, for example...

Then... what happens if you enlarge a lot and you take advantage of 75 lp/mm detail to be shown in the print ? It happens that you also show degradated contrast and noise ! When we downsize the print to not see the flaws then also we don't see the ultimate resolving power.

When we reach the frequency limit (80 or 127lp/mm) contrast is at extintion, near zero, this shows a detail that is it not worth to show in the picture. Still here YMMV, it is a personal choice WHAT image quality we want on the print.

Please check again this scanner comparison https://www.largeformatphotography....Epson-Flatbed-Eversmart-Flatbed-Drum-Scanners.

The Epson delivers the same practical result (even if pixel peeping) than an Scanmate 11000 drum and two high end Creo:

https://www.largeformatphotography....rum-Scanners&p=1479178&viewfull=1#post1479178

The Creo is resolving 5100 dpi effective, twice what the Epson, but the Epson doesn't take a beating from a machine impressively superior: https://www.largeformatphotography....rum-Scanners&p=1478451&viewfull=1#post1478451

The Epson resolves 55 lp/mm in the hor axis and 48lp/mm in the vertical one, but an scanner resolving just the double is not capable to show better results. Do you know why ? This is easy, the shot has not much more than what the Epson can take. And see how bad the image is when you show the 50lp/mm detail...

Still there are situations were you may craft a contrasty edge on film, if we have a powerful silhouette, in perfect focus, shot on tripod and at best aperture, but this is exceptional.


IMO, for all these resons, while we may cature way more, in practice we usually take advantage of those lower (say) 50 lp/mm, but not much beyond.

 

[Alan Johnson]

There is no reason why the low contrast resolution ,quoted in post 123, should suddenly drop off from 88 lppm to 50 just because no test chart is included in the picture.
You have disagreed with the results I have referred to from the commercial scanner dealers Scandig.com in post 117.
You have disagreed with the results from the expert who did the tests for Amateur Photographer for many years , post 123.
Maybe they will be glad you have pointed out the error of their ways.