Pages: 1 2 3 [4]   Go Down

Author Topic: a film to digital print comparison  (Read 23346 times)

Kitty

  • Full Member
  • ***
  • Offline Offline
  • Posts: 126
    • http://
a film to digital print comparison
« Reply #60 on: August 01, 2009, 12:28:10 am »

Quote from: Ernst Dinkla
Not if you take out the default sharpening in the Imacon driver. For that you have to enter a negative number.
Many users are not aware of that "feature". On the Nikon compatible drivers sharpening is an optional choice, not a default one.

The fact that the film is curved during the scan doesn't make it a drum scanner. The reason it is curved in the Imacon is that the film will not pop/deform while mounting and during the scan and the mechanical/optical construction will scan the film along the radius for that curve. The focus should in theory be more consistent that way than in a glassless filmcarrier on a flatbed CCD scanner. A nice solution but not more than that, it still is a CCD scanner. Film wetmounted to glass in a Nikon 8000-9000 and the focusing adjusted for the 4 corners of the frame gives a very sharp scan all over. The wet mounting makes the light transmission uniform, reduces the effect of matte emulsion surfaces and heals damaged film. The drivers for the Nikon, Silverfast, NikonScan and Vuescan, are compatible with the extra Infrared illumination so ICE or ICE alike program features for dust and scratch removal are usable, even with wet mounting. As far as I know wet mounting isn't possible on an Imacon and ICE + hardware infrared wasn't available in older models, but I'm not sure whether the latest versions have it aboard or still not.


met vriendelijke groeten, Ernst Dinkla

Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html

Try scan old photos with texture surface with Imacon. You won't see much reflected texture on paper.
Virtual drum on Imacon did it. There is less reflect with the curve or virtual drum.
I always turn off sharpening when scan.
I think virtual drum also has some effect with film scan too.

I think the big different between direct print from film and scanning and inkjet print is when your negative is overexposure.
You won't see much grain on direct print. The printing tone graduate is still smooth.
While film scan start to see more grain and noise.
Logged

Kitty

  • Full Member
  • ***
  • Offline Offline
  • Posts: 126
    • http://
a film to digital print comparison
« Reply #61 on: August 01, 2009, 12:30:07 am »

duplicated message deleted
« Last Edit: August 01, 2009, 11:31:21 pm by Kitty »
Logged

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
a film to digital print comparison
« Reply #62 on: August 01, 2009, 02:48:58 am »

Hi,

I made 70x100 print from my Sony Alpha 100, APS-C 10 MPixels. This was really intended as an experiment. The image was actually not optimal, because I needed to rotate it and crop. It also went trough some intensive sharpening. The print was than printed on a Lambda at 200 PPI (this is what the lab recommends for large sizes). The outcome was interesting. The image is obviously lacking in detail when looking at short distance, when I back off to about 80-100 cm the image turns pretty sharp and I see no further improvement backing off.

I don't have 20/20 eyesight and need correction glasses, so my findings would probably not apply to person thirty years younger, with 20/20 vision.

Another way to see it is that I often look at pictures as slide shows in 1080P (1920x1080) which is about two Megapixels (2MP!) on a screen 1.5 m wide, I need to sit like 2.5 or 3 meter from the screen, but it seems pretty good at that distance.

On the other hand, I also have a 100x50 cm panoramic stitch on my wall. This was made with the camera in vertical position and it was stitched from seven or nine pictures (with big overlap). Normally you see it at a couple of meters, because thinks standing in the way. If I walk up to that picture and look at very short range it is still critically sharp, but it is a mighty experience because I feel immersed into the picture. The picture essentially covers perhaps 150 degrees of view, so when I look at the picture at normal distance I have a false perspective, looking at the picture close by the perspective is much more real.

There used to be a rule of thumb that correct viewing distance would be focal length times magnification, in the panorama case correct viewing distance would be 50 cm.

A final observation is that sharpening can mask lack of detail as long as we don't have obvious halos. When looking at distance you cannot see detail anyway, but the brain says that because the edges are well defined the picture must be sharp.

So my conclusion is:

1) If you pixel peep all megapixels are needed
2) In many contexts you can do with much less especially if you can keep your viewers from pixel peeping

An observation on the side is that a lot of people I respect would say that it is easy to tell MFDBs apart from DSLR images also in small sizes. I would much like this phenomenon investigated and explained.

Best regards
Erik


Quote from: Dan Wells
Someone on here quoted really good 7x-8x enlargements from scanning on the Coolscans. I have pushed a little bit beyond that comfortably - about 10x  (so 10x15 inches from 35mm, and 20 inches from the 6cm dimension on 120 - 16x20 from 645, 20x20 from 6x6, as much as 20x30 from 6x9). How much beyond that (for a really good print of a high detail subject) will an Imacon go? A drum scanner? I can't imagine getting 20x, because that was widely considered impossible in the chemical darkroom (that would be a 20x30 from 35mm, and any print that size from 35mm I've ever seen was a grainy mess, no matter how it was produced). Is 15x realistic? A good darkroom worker could ALMOST do that, IF they started with a really good negative or chrome - that's where the special developers like Acufine and Microdol came in - people trying to make detailed 16x24 prints from 35mm. Of course there are low-detail subjects (or places where grain is acceptable) that print significantly larger, but that's true of digital as well.
        The best digital sensors today will go something like 25x on the physical size of the sensor (I use the D3x as an example, because it's what I use, but I'm sure that there are other sensors that are in the same range). That gives us 24x36 from full-frame 35 (I do that all the time, and it looks great), 33x44 inches from the smaller medium-format systems and 36x48 from the P45+ size sensor. A P65+ with its oversize sensor SHOULD print right around 40x50, which is, coincidentally, also a 10x enlargement from 4x5.
     There are a couple of problems with this... First is depth of field - I find myself REALLY thinking hard about DOF on big prints (using a bigger camera, one often has tilt to help deal with that - of course this is a problem with non-tilting medium format as well). My next lens purchase will certainly be a PC-E Nikkor (I'm trying to decide on focal length now). The second issue is when do you not actually need any more resolution for a bigger print, because of viewing distance? Ansel Adams raised this question in The Print in his chapter on very large prints. My 24x36's look good from a foot away - could I make a 40x60 that held up at 3 or 4 feet?  If so, who'd ever look at a print that size from any closer than that? Prints that big are often displayed in ways where you can't GET closer than that...
« Last Edit: August 01, 2009, 03:12:59 am by ErikKaffehr »
Logged
Erik Kaffehr
 

ErikKaffehr

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 11311
    • Echophoto
a film to digital print comparison
« Reply #63 on: August 01, 2009, 03:11:04 am »

Hi,

When digital came around I also made the math and it was like this:

Normal lenses and technology produces about 50 lp/mm. This was essentially based on Modern Photography tests. There were plenty of lenses resolving higher but 50 lp/mm is quite respectable if we don't have focus bracketing and so on.
You need two pixels to resolve a line pair that is 3600x2400 that's about 8.5 MPixels. Based on this I guessed that break even would be about 6 MP, this was about when the Mavica came out.

To this we need add that the photographic process essentially looses "sharpness" in each step. In digital photography we can improve sharpness at each step essentially compensating for the loss of edge contrast. The image on film is seldom the final result, you can look at a slide trough a loupe or even a low power microscope but it is not a very pleasant experience. If we enlarge we loose sharpness due to the enlarging lens (that is not infinitely sharp) and also the photographic paper has some MTF characteristic.

A small remark on the side. I have a projector for 6x7 slides made by Götschman, it has a quite good Schneider lens. Even with my slides carefully mounted in glass (GP-slide mounts) the slide is not flat enough to achive perfect focus on screen. When I scan my 6x7 slides I see detail (and color fringing) that neither a 15x loupe or my Götschman projector can resolve.

Best regards
Erik


Quote from: Gemmtech
Of course it is referenced from Ken Rockwell's site, he's the epitome of why the internet is NOT a good place for reference material since 99% of it is false, you have to look for accurate information.  Regarding these "tests" and the use of test charts and camera A vs. camera B, they are just silly nonsense.  I'm assuming most people started using digital as I did, shooting the identical scene with film and their new digital camera and was wondering why the digital file produced a better print.  When the Canon D30 came out and people (including MR) were claiming it to be as good as 35mm film, I balked at such a silly notion, "Just do the math, it will take 40mp to equal 35mm film" so I said.  But then I started doing my own tests and the resolution numbers kept growing and I kept testing (Always against a Nikon F5) and I came to the conclusion many cameras ago that not only was a digital print better (resolution, color, dr, noise etc.) it was exponentially faster and easier to produce a better print.  I truly believe that this is old news and probably not even worth discussing anymore.   And as always, the bottom line is whatever works best for you and whichever method you prefer, that's what you should use.
Logged
Erik Kaffehr
 

Ernst Dinkla

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4005
a film to digital print comparison
« Reply #64 on: August 01, 2009, 04:52:17 am »

Quote from: Kitty
Try scan old photos with texture surface with Imacon. You won't see much reflected texture on paper.
Virtual drum on Imacon did it. There is less reflect with the curve or virtual drum.
I always turn off sharpening when scan.
I think virtual drum also has some effect with film scan too.

I think the big different between direct print from film and scanning and inkjet print is when your negative is overexposure.
You won't see much grain on direct print. The printing tone graduate is still smooth.
While film scan start to see more grain and noise.

I thought we discussed film scanning here.

On flatbeds the usual lamp arrangements are one lamp or two lamps along the optical path/slit, both versions have advantages depending on whether you want texture or no texture. The reason I kept my Epson 3200 next to the V700. Both very able to scan A4 reflective originals.

In the Imacon software turning sharpening off is setting a number like -120 in that menu entry, zero isn't turning it off.

I think the curved mounting was meant for film in the first place.

???


met vriendelijke groeten, Ernst Dinkla

Dinkla Gallery Canvas Wrap Actions for Photoshop
http://www.pigment-print.com/dinklacanvaswraps/index.html

Logged

NikoJorj

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1082
    • http://nikojorj.free.fr/
a film to digital print comparison
« Reply #65 on: August 01, 2009, 10:33:25 am »

Quote from: ErikKaffehr
A final observation is that sharpening can mask lack of detail as long as we don't have obvious halos.
IMHO, this is just the point where the "image content" variable comes in the equation.

Take an image with not much useful detail past a certain resolution, eg a portrait or a proxyphoto with shallow depth of field and where all image information stays at relatively low frequency : all you need is a good interpolation/sharpening workflow to keep the lines fine and straight without halos, and you can stretch it really big. The viewer doesn't really expect to see details past a given frequency.

On the other hand, an image like a classic landscape, where there should be more and more details when you get closer (the ants on the trunks in the distance...), will clearly show a lack of detail when pixels are enlarged too far, even with the best upsizing method.

Of course the above examples are much exaggerated, eg there are many portraits that need fine details (at least in the hair) and simple landscapes that enlarge very well - I think it's just a matter of how much detail the viewer expects with a given subject.
And there are other content-related factors, as said the cobblestone-grain and out of focus blur of some images just saying "this is very hot news" in many minds is another good example.
Logged
Nicolas from Grenoble
A small gallery

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
a film to digital print comparison
« Reply #66 on: August 01, 2009, 10:08:12 pm »

Quote from: NikoJorj
IMHO, this is just the point where the "image content" variable comes in the equation.

Take an image with not much useful detail past a certain resolution, eg a portrait or a proxyphoto with shallow depth of field and where all image information stays at relatively low frequency : all you need is a good interpolation/sharpening workflow to keep the lines fine and straight without halos, and you can stretch it really big. The viewer doesn't really expect to see details past a given frequency.

On the other hand, an image like a classic landscape, where there should be more and more details when you get closer (the ants on the trunks in the distance...), will clearly show a lack of detail when pixels are enlarged too far, even with the best upsizing method.

Of course the above examples are much exaggerated, eg there are many portraits that need fine details (at least in the hair) and simple landscapes that enlarge very well - I think it's just a matter of how much detail the viewer expects with a given subject.
And there are other content-related factors, as said the cobblestone-grain and out of focus blur of some images just saying "this is very hot news" in many minds is another good example.


Excellent points, Nicolas. Holds true for color and tone reproduction as well. Image-specific content is a huge influencer on the viewer's impression of overall image quality.

Kind regards,

Mark
http://www.aardenburg-imaging.com
Logged
Pages: 1 2 3 [4]   Go Up