Forgot Password
Pentax Camera Forums Home
 

Reply
Show Printable Version Search this Thread
05-22-2016, 10:50 AM   #31
Veteran Member
artobest's Avatar

Join Date: Jul 2008
Location: Swansea, Wales
Posts: 455
QuoteOriginally posted by bjolester Quote
What you are seeing in the Canoscan scan is digital smearing/noise.
Yes, let's be clear, that is a very poor scan from a very average scanner and in no way a valid basis for any meaningful comparison.

05-22-2016, 11:18 AM   #32
Site Supporter
reh321's Avatar

Join Date: Nov 2014
Location: South Bend, IN, USA
Photos: Gallery | Albums
Posts: 2,728
QuoteOriginally posted by bjolester Quote
Well, if your comparison earlier in this thread is between the Pentax K-1 and a 35mm Ektar 100 shot scanned on a entry level consumer flatbed scanner like the Canoscan 9000 F Mkii, then I would conclude that your comparison is flawed. According to the German film scanner test site "www.filmscanner.info" the Canoscan can only produce a 1700ppi scan of 35mm film, which is the equivalent of about 4mp. You are hardly doing film any justice comparing the Pentax K-1 with a 35mm negative scanned on such a scanner.

Detailed test report flat bed scanner Canon CanoScan 9000F Mark II; evaluation of the image quality of the scanner

The best scanners for 35mm (Minolta Dimage Scan Elite 5400, Nikon 5000, Imacon, drum scanners) can produce scans that could capture much of the information held on a 35mm negative or slide, and yield scans around 24mp with a colour slide film like Provia 100F. Now a comparison between the very best fullframe DSLRs and a professionally scanned 35mm would make a relevant point of departure for an interesting comparison. But wait, there has already been published such comparisons. Here is one of the most interesting ones:

https://www.onlandscape.co.uk/2014/12/36-megapixels-vs-6x7-velvia/
Back before I was willing to buy a digital camera, I sent some Kodachrome 25 slides in to a professional to be scanned; he produced 6MP images. I then set my slide projector and screen up in our "library" {aka "family room", aka "den"} where our computer was at the time, and compared the slides to the scans. I had intentionally picked slides for their detail {some were railroad cars with lettering on them and some were taken at an annual New Year party thrown by a friend of ours who was a librarian, so lots of reading material was visible in the pictures}. Every detail I could find on a slide was also present on the scan. At the time I worded my conclusion as "at the level of equipment I'm willing to pay for, 6MP is as good as Kodachrome 25, which was generally recognized as the finest grain slide film available, so when digital cameras reach 6MP, I'll be willing to switch". I had switched to Canon by then, but when I switched back to Pentax by buying a K-30, I put the lens {Pentax-A 50mm f/1.7} used to take those pictures on it, and I was shocked at how much more detail I was getting from it now. I am firmly convinced that every digital camera sold today provides much more detail than we ever got from film, which is partly why I use to term "needle sharp" {instead of the usual "razor sharp"} to express how addicted we have become to sharpness.
05-22-2016, 11:27 AM   #33
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,189
QuoteOriginally posted by funktionsfrei Quote
When a scanner with an effective resolution of 1700 ppi already shows the film's grain, that film can't have that mystical super-ultra-mega resolutions analogue addicts dream about.
It is more complicated than that* and one of the reasons why film/digital comparisons of pictorial photographs are best done as optical prints from the negative compared to high quality ink-jet prints from digital. For resolution targets, direct evaluation of the negative using a microscope is the gold standard** vs. 1:1 (pixel to pixel) on-screen for digital.

In my opinion, people should stop playing number games regarding film resolution. Both mediums are valid for photographic capture and both offer high quality options for display at reasonable viewing size.


Steve

* Simply put, the scanner may introduce diffraction-related artifact that obscures available detail.

** This is how Zeiss and other high-end lens makers test lens resolution...with film. The film is very high contrast and very fine grain copy and easily out-resolves the lenses.
05-22-2016, 11:30 AM   #34
Site Supporter
reh321's Avatar

Join Date: Nov 2014
Location: South Bend, IN, USA
Photos: Gallery | Albums
Posts: 2,728
QuoteOriginally posted by HoustonBob Quote
I have to ask this question: "Why stop with a K3?" If you crop a K3 sensor to the size of a Pentax Q sensor the Q is much more demanding of lenses than the K3. Everything you have said argues more for using the Q than the K3.

This thread was about evaluating film lenses for use on the K1. It didn't have anything to do with the K3 or any other digital camera.
I got my Q-7 partly as a "birding" camera, naively thinking I could pair it up with any of the old zoom lenses I had left from old film cameras. Every one was basically inadequate, in CA as well as in detail. Only when I bought a new Sigma 70-300mm APO lens was I able to get "satisfactory" detail, but I am certain that a lens actually designed for the Q-mount would do somewhat better yet.



05-22-2016, 11:35 AM   #35
Pentaxian
aurele's Avatar

Join Date: Jan 2011
Location: Paris, France
Photos: Albums
Posts: 2,906
QuoteOriginally posted by funktionsfrei Quote
When the scan with whatever effective resolution already shows the film's grain, it won't be useful to increase the scanner resolution.
With the Canoscann it's digital noise, not film grain that you see !
05-22-2016, 02:44 PM   #36
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,189
QuoteOriginally posted by funktionsfrei Quote
When a scanner with an effective resolution of 1700 ppi already shows the film's grain, that film can't have that mystical super-ultra-mega resolutions analogue addicts dream about.
QuoteOriginally posted by stevebrot Quote
It is more complicated than that*
OK, here is a link to a PDF that explains fairly well what film grain is, how it relates to "fundamental particles" (the stuff the image is made of), how grain relates to detail capture, and the implications for scanning. The article was authored about nine years ago, but is still helpful. Some of the information confirms assertions by HoustonBob, while some does not. I would not consider the numbers in the tables to be authoritative, particularly in regards to lens resolution. The same is probably true for many of the films (e.g. the resolution for Panatomic-X (FX) is overstated by about 50% while that for Technical Pan is understated by about half...I shot both for years).

http://cool.conservation-us.org/coolaic/sg/emg/library/pdf/vitale/2007-04-vi...resolution.pdf

The range of (apparent) size for film grain (~0.5 μm to ~30 μm) overlaps the resolution range of most scanners*, but what is recorded is not an image of grain per se. Rather, it is how the scanner's sensor array records the interaction of light with the "fundamental particles" suspended in the emulsion layer. How much of that record appears as grain in the scanned image depends on the hardware, settings, and the operator skill.


Steve

* The finest grain films are recorded as essentially grainless by even the best consumer-grade scanners whereas some materials have obvious grain at fairly low resolution.
05-22-2016, 03:17 PM   #37
Site Supporter




Join Date: Feb 2013
Location: Houston TX
Photos: Gallery
Posts: 209
Original Poster
QuoteOriginally posted by bjolester Quote
Well, if your comparison earlier in this thread is between the Pentax K-1 and a 35mm Ektar 100 shot scanned on a entry level consumer flatbed scanner like the Canoscan 9000 F Mkii, then I would conclude that your comparison is flawed. According to the German film scanner test site "www.filmscanner.info" the Canoscan can only produce a 1700ppi scan of 35mm film, which is the equivalent of about 4mp. You are hardly doing film any justice comparing the Pentax K-1 with a 35mm negative scanned on such a scanner.

Detailed test report flat bed scanner Canon CanoScan 9000F Mark II; evaluation of the image quality of the scanner

The best scanners for 35mm (Minolta Dimage Scan Elite 5400, Nikon 5000, Imacon, drum scanners) can produce scans that could capture much of the information held on a 35mm negative or slide, and yield scans around 24mp with a colour slide film like Provia 100F. Now a comparison between the very best fullframe DSLRs and a professionally scanned 35mm would make a relevant point of departure for an interesting comparison. But wait, there has already been published such comparisons. Here is one of the most interesting ones:

https://www.onlandscape.co.uk/2014/12/36-megapixels-vs-6x7-velvia/

There are other tests of the Canoscan 9000F Mk II which disagree substantially with that test. Who are we to believe? I have seen tests directly comparing the Canoscan with the best scanners that Nikon and Minolta made; the Canoscan came out on top - resolving much more than either of those dedicated scanners could. I guess it depends on whose test you want to believe. Maybe one reviewer got a defective Canoscan while the others had defective Nikon and Minolta scanners.

As I said it is very easy to shut me up, and get an apology from me for everything I've said in this thread. Post film scans of real world color subjects from 35mm color film that out resolve a K-1. I told you how to do that in an earlier post. Don't quote some review or bogus tests done by somebody with an agenda. I would love for my color film to suddenly out resolve a K-1 and to be able to buy equipment that could do that. However I don't see the information on the negatives or on slides I have looked at with a microscope so it is very hard for me to believe it is going to happen, I have been wrong often in the past so I could be wrong here. Don't talk Do. I posted my scans - post yours.
05-22-2016, 05:25 PM   #38
Site Supporter




Join Date: Feb 2013
Location: Houston TX
Photos: Gallery
Posts: 209
Original Poster
I just scanned a slide using my Canoscan. I set the resolution to 9600 dpi which yielded a 100 MB tiff of approximately 35 megapixels 7204 x 4838. I then used the Gimp to blow up the picture until I could easily see the individual pixels on the scan. The review

Detailed test report flat bed scanner Canon CanoScan 9000F Mark II; evaluation of the image quality of the scanner

made a claim that high resolution scans with the scanner group pixels into blocks of 4 identical pixels. That claim is clearly false. Looking at the individual pixels on the slide scan it was obvious that they were just that; individual pixels which were produced by doing an optical scan at the claimed density. Objects in the scanned picture were clearly delineated by varying individual pixels - with none of the claimed blocking being visible. This is exactly the behavior I would expect from an actual 9600 dpi optical scan; which causes me to believe that the Canoscan is doing what its specifications say it does. The Canoscan is not producing false high res scans - it appears to be doing its job correctly.

I can clearly see film grain in the high res scans it produces.

05-22-2016, 06:34 PM   #39
Marketplace Reseller
dcshooter's Avatar

Join Date: Jan 2011
Location: Washington DC
Photos: Gallery | Albums
Posts: 4,112
Out of morbid curiosity, I decided to pop back into this thread to see how this conversation was going, in a staring-at-a-car accident kind of way, and I was absolutely unsurprised to find HoustonBob being shown to be wrong on every point (great explanations, BTW, Steve).

Among other obsevations:
- As others have stated it's absolutely ridiculous to make any comparisons based upon a scan from a consumer-grade flat bed scanner. Even the best wet drum scans are generally considered by actual experts to lose 20% of the actual resolution of the film. A large part of this is due to the fundamental difference in the way resolution is produced by randomly shaped film grains, composed of randomly sized silver halide crystals, which are distributed randomly across a film surface vs. equally sized pixels in a square grid, produced either via a scan or through a bayer arrayed sensor. Even the very best flatbeds for scanning negatives lose even more than half of the film's resolution due to diffraction/refraction from the glass, unevenness of the medium surface, etc.

- You also seem to not understand the difference between resolution vs. perceived sharpness. Even with your crappy scan and ludicrously uncontrolled comparison, a large part of what you are putting forth as increased resolution has to do with lack of grain in the digital image and greater edge contrast (acutance) rather than actual better resolution. Unprocessed digital naturally has more edge contrast than does unprocessed scanned film. It's a property of the medium. But edge contrast is the one parameter of perceived contrast that can be adjusted after the fact (and is what pretty much all software sharpening algorithms manipulate), so in reality your "fair" comparison actually works to make film look worse than it really is.

- lp/mm measurements are made at standardized contrast ration precisely because they help control for the subjectivity (both in terms of medium performance and the eyesight/perception of the measurer) in lp/mm tests by eliminating one of the variables that can affect perceived sharpness in a finished image.

- There are many films that have exceeded 100 lp/mm at 1000:1 (i.e. black white bars) contrast. Your assertion that lenses were "designed down" to just above 50lp/mm as a result is preposterous.

- Turning this all back around to your original proposition, your proposed testing "methodology" includes so many uncontrolled variables (in scene, composition, shake, etc. etc. etc.) that it is completely meaningless if you are just trying to see if a lens is "sharp enough" for the K-3. This is precisely why resolution tests use set scenes/charts, controlled for ISO and/or film type in order to get oranges to oranges comparisons of sharpness.

- Nearly all prime Pentax lenses ever made have plenty of resolution to be pixel sharp on a K-1 at the center of the image under the right conditions (shutter speed, aperture, etc). Edges and corners are a different story. So are different aperture values. And many lenses that aren't sharp in all circumstances will still take great pics on the K-1 by virtue of other rendering characteristics, not to mention the ability of the person pushing the shutter button. Simply grabbing a pic taken with arbitrary settings and pixel peeping, then rendering judgment on whether a lens is good enough for the K-1 is just about the most myopic - pun intended - way to go about doing so.

Last edited by dcshooter; 05-22-2016 at 08:35 PM.
05-22-2016, 06:34 PM   #40
Lens Buying Addict
monochrome's Avatar

Join Date: Mar 2008
Location: Kirkwood (St. Louis) MO
Photos: Gallery | Albums
Posts: 19,462
We need gaweidert in this thread. He did technical work for Kodak for decades. Digitalis would also be helpful, or Chris Platt, RockvilleBob or Phil (who has already stopped in).

In my experience there are a couple dozen film era lenses from K through FA that test 90 - 110 lp/mm @f/5.8 -f/8 in the center and 80's at the edges. That means the m42 versions of the K lenses should also have been equally high-resolving.

These lenses were not designed to meet any supposed limitations of color film resolution.

Last edited by monochrome; 05-22-2016 at 06:43 PM.
05-22-2016, 06:37 PM   #41
FOX
Senior Member




Join Date: Jan 2013
Location: Ottawa
Photos: Gallery
Posts: 132
A lot of film scanners have not true dpi but interpolated, if you test them with a test chart you will not resolve 9600dpi. Even cheap 100 dollar scan can deliver a blown 9600 tiff. My CanoScan mark I for slides is shitty compared to my OpticFilm 120 at the same 9600 dpi.
Also focus is hard to do on most, even when you would think they would leave a depth margin for film width/build errors.
05-22-2016, 07:58 PM   #42
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,189
QuoteOriginally posted by HoustonBob Quote
There are other tests of the Canoscan 9000F Mk II which disagree substantially with that test. Who are we to believe?
Please believe me when I say that I am typing this comment with sympathetic fingers. Please also read my personal experience on down before discounting the following comment on Scandig.

Who to believe? I would tend to believe the reviewer that actually uses a resolution test slide (USAF-1951 in a 35mm slide mount) and publishes their test methods. Scandig is sort of a standard resource on the Web for scanner reviews and widely regarded as authoritative. How they test resolution is detailed at the link below:
Filmscanner: Determination/Measurement Resolution (dpi, line pairs); Flat bed scanner USAF 1951 Test Target Test-Chart
...along with instructions on how to repeat the tests yourself using their target available at market rate:*
USAF-1951 test chart - high resolution target for specifying the resolution of film scanners, resolution measurement - ScanDig GmbH
FWIW, I don't believe that Scandig has tested any consumer flatbed scanner at better than about 2400 dpi, regardless of brand or reputation.

OTOH, you own the Canoscan and are in a position to do head-to-head testing if you can locate another scanner, say a Nikon 5000 ED, to compare against. That would be a reasonable test. Your scanner ought to blow the Nikon out of the water. What does Scandig know? That was my feeling several years ago.

Now my sad story. I don't own a Canoscan, but I do own an Epson V700, a fairly expensive flatbed scanner that Epson and others (several reviews) claim has a true 6400 native dpi. I also own a Nikon 5000 ED which Nikon claimed has a true 4000 dpi. I bought the Epson specifically for 120 roll film and 4x5 sheet, but was hoping that I could use it for bulk scanning 35mm stuff, a task that is much easier on the Epson than on the Nikon. I did a few test scans on the Epson with disappointing results. I then set up a head-to-head with the Nikon with both set at 2400 dpi, figuring that both scanners should have plenty of head room. After the first set of images, I felt more than a little "subdued".

I documented my experience on this thread from 2010:
https://www.pentaxforums.com/forums/107-film-processing-scanning-darkroom/103...on-images.html
After several pages of discussion during which it occurred to me that negative flatness may have been an issue, I decided to try again using a very flat TMax negative having very high detail. The second set of comparison scans showed pretty equivalent performance at 2400 dpi.

Full image:


TMax 100 processed in FG-7 (1+15), Tamron 28/2.5 (02B)


Full-resolution comparisons:




I then repeated at 4000 dpi on the Nikon and 4800 dpi on the Epson downsampled to 4000 dpi at scan time.




Hmmmm...perhaps Scandig was right when they pegged the V700 at 2300 dpi! There was no contest. From that date forward, when asked about the V700, I have cautioned that about 2400 dpi is about the best one can expect and that I don't use it for fine work with 35mm. (Love it for MF and LF though.)

I know there are reviews for the Canoscan that claim pretty impressive performance, but based on the sample you posted, I don't believe the scan is representative of the film and is what I would expect from your scanner and not a whole lot worse than what I would expect from my Epson.


Steve

* I believe that Edmund Optics has a similar target on 2"x2" glass for about the same amount of money.

Last edited by stevebrot; 05-22-2016 at 09:34 PM.
05-22-2016, 08:28 PM   #43
Site Supporter
reh321's Avatar

Join Date: Nov 2014
Location: South Bend, IN, USA
Photos: Gallery | Albums
Posts: 2,728
QuoteOriginally posted by stevebrot Quote
...FWIW, I don't believe that Scandig has tested any consumer flatbed scanner at better than about 2400 dpi, regardless of brand or reputation.

...After several pages of discussion during which it occurred to me that negative flatness may have been an issue, I decided to try again using a very flat TMax negative having very high detail. The second set of comparison scans showed pretty equivalent performance at 2400 dpi...
I own three scanners, an Epson flatbed, a Plustek 8200, and an old Nikon LS-2000. My personal experience is that the Epson lags way behind the other two. I believe that the issue is partly simply construction; the Epson has a plate of glass in the optic path, while the other two have nothing but air, and we all know that extra glass in the path reduces sharpness.
05-22-2016, 08:51 PM   #44
Marketplace Reseller
dcshooter's Avatar

Join Date: Jan 2011
Location: Washington DC
Photos: Gallery | Albums
Posts: 4,112
QuoteOriginally posted by funktionsfrei Quote
Sharper outlines of the individual silver halegonide crystals that make the film's grain?
Also, it's important to realize that the "grains" you see in developed films are not individual crystals but rather are an observation of the film's RMS granularity, which is basically the signal to noise ratio of the film. The areas of discrete color are formed by clumps of crystals of various sizes/shapes, and the macroscopic grain size is a function partially independent of the size of individual grains and the film's overall resolving power.

Just because you can get to the point where you see differentiation of individual "grains" on a scan does not mean you have reached its maximum resolution. The sharpness of the outlines of those grains actually does matter, since in a film image, the shape of the grains carries a good deal of the total resolution information!
05-22-2016, 08:56 PM   #45
Site Supporter
stevebrot's Avatar

Join Date: Mar 2007
Location: Vancouver (USA)
Photos: Gallery | Albums
Posts: 26,189
QuoteOriginally posted by HoustonBob Quote
Post film scans of real world color subjects from 35mm color film that out resolve a K-1.
We both know that is unlikely to happen, not because it can't be done, but because it is not worth the effort. Back to the original question regarding legacy lens performance on film and presumably on digital, I can provide a fairly ancient (1978 from Modern Photography) article that details a test of a half-dozen lenses that made 100+ lp/mm against readily available films of the time. All were capable of coming within 0.5 lp/mm of or exceeding the pre-Bayer interpolation K-1 figure of 102.5 lp/mm. I would expect that those same lenses should be in the same range in 2016.

All sizes | How Sharp Can You Get? | Flickr - Photo Sharing!
All sizes | How Sharp Can You Get - continued | Flickr - Photo Sharing!

(Many thanks the @Nesster for his scans. Originally shared as part of a previous thread on resolution in 35mm film photography)


Steve

(...the Summicron 50/2, don't know which version, made 89 lp/mm with KII...)

Last edited by stevebrot; 05-22-2016 at 09:09 PM.
Reply

Bookmarks
  • Submit Thread to Facebook Facebook
  • Submit Thread to Twitter Twitter
  • Submit Thread to Digg Digg
Tags - Make this thread easier to find by adding keywords to it!
4k, color, detail, dslr, film, format, full frame, full-frame, grain, k-1, k-1 lens rating system, k1, lens, lenses, medium, pentax, pentax k-1, picture, print, resolution, result, scanner
Thread Tools Search this Thread
Search this Thread:

Advanced Search


Similar Threads
Thread Thread Starter Forum Replies Last Post
FF lenses for the K-1 that can be bought new Class A Pentax K-1 52 09-22-2016 12:20 AM
Bargain lenses for K-1? loco Pentax K-1 55 05-12-2016 08:56 AM
Lenses recommendation for the K-1 FtYoU Pentax K-1 19 05-04-2016 02:51 PM
Limited Lenses with the Pentax K-1 johnmb Pentax K-1 24 04-24-2016 04:22 AM
Which of the lenses from my K-50 are worth using with film? bladerunner6 Pentax Film SLR Discussion 12 12-22-2014 02:28 PM



All times are GMT -7. The time now is 08:53 PM. | See also: NikonForums.com, part of our network of photo forums!
  • Red (Default)
  • Green
  • Gray
  • Dark
  • Dark Yellow
  • Dark Blue
  • Old Red
  • Old Green
  • Old Gray
  • Dial-Up Style
Hello! It's great to see you back on the forum! Have you considered joining the community?
register
Creating a FREE ACCOUNT takes under a minute, removes ads, and lets you post! [Dismiss]
Top