David verus Goliath: or just another vaudeville act

[lowep]

High quality scans from 6x7 color negs shot with a Mamiya 7 should be far superior in terms of detail, dynamic range and noise than low res 12MP RAW files from a Canon 5d DSLR, right?    

This is what I thought until a few days ago when I made the mistake of trying to compare scans from the color negatives (oranges) with RAW files from the DSLR (apples) of the same subject.    My idea was to use the color negatives for big prints and the digital camera as a back-up if anything went wrong with the negatives. However, what I saw – much to my surprise – was that blow-ups of the low-res digital “backup” files seem to be more user-friendly (as source files for working on images in Photoshop to create large prints) than the noisy scans of the color negatives. So I am totally confused.   Maybe I am going nuts?

1)   ORIGINAL NEG VERSUS NEG SCAN: Sharpness of detail and dynamic range of the original 110 Kodak Portra 160ASA color negs (viewed through loupe on lightbox) seems far superior to the 16-bit RAW scans of the color negs from Imacon Precision Flextight II film scanner (viewed on calibrated monitor). Is this normal?
2)   FILE SIZE: Of course file size of the 16-bit Imacon (3F) RAW scans of the MF color negs is much bigger in terms of MB than file size of the 12MP RAW files from the 5D.
3)   NOISE: At 100 percent the scans from MF color negatives are much noisier than the 5D files??? Even when I reduce the size of the scans of the color negatives to the same dpi and output dimensions as the 5D files in Photoshop, noise is still much more extreme in the scans of the color negs than in the 5D files.
4)     NOISE REDUCTION: Using noise reduction plug-ins to eliminate visible noise from the scans of the color negs makes the images so soft they are useless.
5)   INTERPOLATION: When I use interpolation to enlarge the output dimensions of the RAW files from the Canon 5D to about 4 times bigger than their original size (ie up to the same output dimensions as the color negative files at 300dpi), the apparent sharpness of the 5D images does decrease – making the images look slightly softer – but the blown-up 5D images still seem to be quite useable and there is no noise or other problems such as jagged edges, purple fringing etc. Hmmm...  
5)   MANIPULATION: When I start to manipulate curves and levels in Photoshop, I can pull much more detail out of the color negatives (eg clouds) than I can pull out of the files from the digital camera.
     
What suprises me is that the RAW files from the 12MP Canon 5D seem to be vastly less noisier than the scans of the MF color negatives, and how well the RAW negatives from the Canon 5D seem to “stretch” when interpolated.

If this is true, then maybe I need to rethink my strategy of relying on film for big prints and digital as back-up. Before making such a drastic decision (I have a fridge full of unexposed film!) I would like to get feedback from anybody who has done similar comparisons of RAW files from digital camera and scans of color negatives. Hopefully I am completely wrong as this would make my life easier.  

[Czornyj]

High quality scans from 6x7 color negs shot with a Mamiya 7 should be far superior in terms of detail, dynamic range and noise than low res 12MP RAW files from a Canon 5d DSLR, right?    

This is what I thought until a few days ago when I made the mistake of trying to compare scans from the color negatives (oranges) with RAW files from the DSLR (apples) of the same subject.    My idea was to use the color negatives for big prints and the digital camera as a back-up if anything went wrong with the negatives. However, what I saw – much to my surprise – was that blow-ups of the low-res digital “backup” files seem to be more user-friendly (as source files for working on images in Photoshop to create large prints) than the noisy scans of the color negatives. So I am totally confused.   Maybe I am going nuts?

1)   ORIGINAL NEG VERSUS NEG SCAN: Sharpness of detail and dynamic range of the original 110 Kodak Portra 160ASA color negs (viewed through loupe on lightbox) seems far superior to the 16-bit RAW scans of the color negs from Imacon Precision Flextight II film scanner (viewed on calibrated monitor). Is this normal?
2)   FILE SIZE: Of course file size of the 16-bit Imacon (3F) RAW scans of the MF color negs is much bigger in terms of MB than file size of the 12MP RAW files from the 5D.
3)   NOISE: At 100 percent the scans from MF color negatives are much noisier than the 5D files??? Even when I reduce the size of the scans of the color negatives to the same dpi and output dimensions as the 5D files in Photoshop, noise is still much more extreme in the scans of the color negs than in the 5D files.
4)     NOISE REDUCTION: Using noise reduction plug-ins to eliminate visible noise from the scans of the color negs makes the images so soft they are useless.
5)   INTERPOLATION: When I use interpolation to enlarge the output dimensions of the RAW files from the Canon 5D to about 4 times bigger than their original size (ie up to the same output dimensions as the color negative files at 300dpi), the apparent sharpness of the 5D images does decrease – making the images look slightly softer – but the blown-up 5D images still seem to be quite useable and there is no noise or other problems such as jagged edges, purple fringing etc. Hmmm...  
5)   MANIPULATION: When I start to manipulate curves and levels in Photoshop, I can pull much more detail out of the color negatives (eg clouds) than I can pull out of the files from the digital camera.
     
What suprises me is that the RAW files from the 12MP Canon 5D seem to be vastly less noisier than the scans of the MF color negatives, and how well the RAW negatives from the Canon 5D seem to “stretch” when interpolated.

If this is true, then maybe I need to rethink my strategy of relying on film for big prints and digital as back-up. Before making such a drastic decision (I have a fridge full of unexposed film!) I would like to get feedback from anybody who has done similar comparisons of RAW files from digital camera and scans of color negatives. Hopefully I am completely wrong as this would make my life easier.

I use M7 system + Nikon 9000ED and in terms of resolution and sharpness it clearly outperforms any 35mm 12MP DSLR RAW I saw:
50mm @f/11, Ektachrome VS

@100%

On the other hand - DSLR RAW files are more flexible, clean, and color is easy to control. I never use negatives, but also on Velvias, Provias, Ektachromes hi-res scans grain is visible.

[Ben Rubinstein]

The Imacons never used to be much good with neg film.

[lowep]

Hmmm... your scan at 100% is much less noisier than mine at 100%. I will attempt to load 2 sample images from one of my 110 Kodak Portra 160NC color negs for comparison. First sample is whole frame and second is 100% crop. Maybe my problem is my ageing  Imacon Flextight Precision II scanner?  I just changed the light tubes but it doesn't seem to make any difference. What scanner/scanner resolution/capture software did you use for the image you posted?

[Czornyj]

Hmmm... your scan at 100% is much less noisier than mine at 100%, as you can see by clicking on this link to 2 sample images from one 6x7 color neg. Alternatively refer to the attachments. Maybe my problem is my ageing  Imacon Flextight Precision II scanner?  I just changed the light tubes but it doesn't seem to make any difference. What scanner/scanner resolution/capture software did you use for the image you posted?

I used Nikon Coolscan LS-9000ED, NikonScan, image scanned @ 3000dpi.

[Czornyj]

Here's an example of 4000dpi scan, Fuji RDPIII:

[lowep]

Impressive detail... did you notice those people on the rocks when you made the photo?

How big would this print be if you printed it at 300dpi without interpolation and how visible would the noise be?

One reason why I am suddenly becoming so sensitive about noise is I have noticed that when I view the 16-bit TIFFs created from RAW files from the Canon 5d at 100% they look ok.  But files from film scans don't -- even though the 100 percent clip from Czornyj's image of the sea has amazing detail it also has a lot of noise. But maybe this is not important?

What is the relevance/relationship of 100 percent viewing of files in Photoshop to actual print output? Previously I never thought about this, as I thought that viewing 300dpi files of film scans at no more than 50% zoom on the screen would give me sufficient guidance to be able to predict how the final print would turn out. Now I have (by comparison) "little" digital files from the 5d that I want to blow up as big as possible, so I am looking more carefully at what I see when I zoom in to 100%. Am I making a mistake by doing this? Obvious solution would be to do some test prints comparing files from digital and film sourced files but I don't have easy access to a printer at the moment. So (yet) another question: what is the relationship of the 100% zoom that can be viewed in Photoshop (if any) to the maxim size printout that can be produced from any given file?

[Czornyj]

Impressive detail... did you notice those people on the rocks when you made the photo?

How big would this print be if you printed it at 300dpi without interpolation and how visible would the noise be?

One reason why I am suddenly becoming so sensitive about noise is I have noticed that when I view the 16-bit TIFFs created from RAW files from the Canon 5d at 100% they look ok.  But files from film scans don't -- even though the 100 percent clip from Czornyj's image of the sea has amazing detail it also has a lot of noise. But maybe this is not important?

What is the relevance/relationship of 100 percent viewing of files in Photoshop to actual print output? Previously I never thought about this, as I thought that viewing 300dpi files of film scans at no more than 50% zoom on the screen would give me sufficient guidance to be able to predict how the final print would turn out. Now I have (by comparison) "little" digital files from the 5d that I want to blow up as big as possible, so I am looking more carefully at what I see when I zoom in to 100%. Am I making a mistake by doing this? Obvious solution would be to do some test prints comparing files from digital and film sourced files but I don't have easy access to a printer at the moment. So (yet) another question: what is the relationship of the 100% zoom that can be viewed in Photoshop (if any) to the maxim size printout that can be produced from any given file?

This is a 96MP file - if I'd like to make a print of the actual size of that 100% crop to see that noise, I'd end up with 220x280cm (86x110") print - so I don't really care, as I don't make such a large print very often  .
It's rather not likely that the noise would be visible at 300dpi (73x93cm - 28x36"):

[lowep]

This is a 96MP file - if I'd like to make a print of the actual size of that 100% crop to see that noise, I'd end up with 220x280cm (86x110") print - so I don't really care, as I don't make such a large print very often  . It's rather not likely that the noise would be visible at 300dpi (73x93cm - 28x36")

Meanwhile I found this very detailed piece about exactly this question that together with the chance to consider your images Czorny has helped me to figure out why my film scans look so noisy - at 100 percent zoom they are blown up so far beyond the snowline of their resolving limit that nothing can grow there apart from noise - plus what I need to do, so thanks for the input and advice!

[keithrsmith]

This is really interesting.

A question:  if the digital file is uprezzed and treated in the way Schewe suggests ( adding noise, sharpening etc) does this change the equation where file gains due to real or pseudo detail at large enlargements?

keith

[Anthony R]

I can't wait to celebrate the day/year when photographic discussions go back to talking about photography and not THIS subject in all its various forms. By all means, carry on...(wasted keystrokes and breath).

[lightstand]

I can't wait to celebrate the day/year when photographic discussions go back to talking about photography and not THIS subject in all its various forms. By all means, carry on...(wasted keystrokes and breath).

were those the same days when everyone used HC-110 and not D-76? :-) Sorry couldn't resist but I definitely hear you.

[JeffKohn]

I can't wait to celebrate the day/year when photographic discussions go back to talking about photography and not THIS subject in all its various forms. By all means, carry on...(wasted keystrokes and breath).

Such a time has never existed in the history of photography, and probably never will... http://theonlinephotographer.typepad.com/t...rest-facts.html

What I don't get about people who complain about tech talk, is that they seem to be operating under the assumption that craft/technology and artistry are somehow diametrically opposed. Why can't people be interested in both?

[Czornyj]

What I don't get about people who complain about tech talk, is that they seem to be operating under the assumption that craft/technology and artistry are somehow diametrically opposed. Why can't people be interested in both?

I suppose they can - otherwise what would they do in Digital Image Processing subforum?

[ErikKaffehr]

Hi,

I made once a 70x100 enlargment from scanned 6x7 slide, probably Velvia. The print was absolutely smooth although the scan was rather grainy in actual pixels on screen. I spent a couple of hours optimizing it in Photoshop. It turned out "perfect", razor sharp an no grain. I interpolated to 200 PPI at 70x100 cm and printed on Durst Lambda.

I also made a 70x100 enlargement the same way from a 10 MP APS-C camera. It does not has the resolution of the 6x7 but also no grain.

Best regards
Erik

This is a 96MP file - if I'd like to make a print of the actual size of that 100% crop to see that noise, I'd end up with 220x280cm (86x110") print - so I don't really care, as I don't make such a large print very often  .
It's rather not likely that the noise would be visible at 300dpi (73x93cm - 28x36"):