Regarding scanned film and digital

[ErikKaffehr]

Hi!

Update:

Full size jpegs from the scans and DSLR are available for download here: http://echophoto.dnsalias.net/ekr/images/Scaninfo/

Info: I have some doubts on the published evaluation of the Stouffer Wedge, partly because the image is black on transparent background. I made some more experiments but have no real solid results yet. With high exposures the DMAX of the scanner may matter a lot.



There is an intense discussion going on the forums regarding a test of 8x10" vs. IQ180 back, and a lot of discussion about scanning technique. I have made some tests recently comparing my old Pentax 67 vs. my now three years old Sony Alpha 900. It's clearly not 8x10" vs IQ180. But pixels are pixels and these are hands on comparisons on real life subjects.

Two images shot same day from same tripod positions (well almost, I had L-bracket on the Sony but not on the Pentax, so there is some shift!). The film used was Ektar 100.

Scans were made at 3200PPI on a Minolta Dimage Scan Multi Pro. One of the negatives is quite dusty, the dust shows what edge resolution the scanner is capable of. In both cases the 67 scans are shown at full resolution. The Sony Alpha 900 image was uprezzed to same size with bicubic in CS5. Sharpening I regard to be adequate was done on both images.

The second image I show was carefully focus bracketed. It's the best image of five. There was not a lot of difference between the images.

My observations:

1) The images are pretty close in sharpness. Choice of sharpening may make one or the other resolve better
2) The Sony Alpha image is much cleaner, that is has less noise
3) For some reason the flower image has good resolution in the greens but bad resolution in the reds with film. I have seen the same issue on Velvia. Could be chromatic aberration in the lens, or something else.

I'll probably put the images on a server for download would anyone be interested.

I also added a comparison of film vs. sensor characteristics based on a Stouffer wedge with 41 0.1 density steps. Illumination not really even but same for both images.

Best regards
Erik

[John Rodriguez]

Thanks Erik.  That's about what I've suspected - ~20ish mega-pixels on a DSLR = MF film. You could probably eek out a bit more resolution with a drum scan but it would still be really close.  Do you see the issues with red on the slides themselves? 

[ErikKaffehr]

Hi,

Yes, reds are washed out on Velvia no tonal separation on the slide, on Ektar 100 I don't know, yet.

The Velvia test is here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/16-pentax67velvia-vs-sony-alpha-900

The latest images are here:
http://echophoto.dnsalias.net/ekr/images/Scaninfo/

Best regards
Erik

[Schwarzzeit]

Erik, it's obvious that the red channel is clipped on these roses. How did you do the color neg conversion?

I have seen this from CCD scanners before though it doesn't necessarily have to be the hardware. Sometimes the scanner software clips the highest densities during the negative conversion. In this case it helps to do a raw scan as a positive and then convert manually in PS.

-Dominique

[Stefan.Steib]

Hi Eric

Pretty much on par with my experiences, my actual P45+ will match 4x5" most of the time.
If you pull any tricks and have a very nice slide (velvia 50 or 100) you may have some tiny bit of advantage in Resolution, but only if you do not count in the grain "noise"
which if you remove it will also cut some of the resolution. Whereas if you do a deconvolution sharpening after upres of the P45+  file with software like Genuine fractals you may get better results.
But really - who needs this all of the time ?
We recently sold a Leaf Aptus II 12R with an HCam to a customer who is doing gallery prints with 2x3 meters and maybe larger sometimes.
He was using 4/5" (with 35mm lens) before and is now switching to digital with our TSE -17mm/Hcam.

I had posted some of the images in GetDPI already , but in case someone want´s to see a full res of what this combo did with our Hartblei 80mm take a look here:

http://www.hcam.de/downloads/Hartblei80mm_HCamB1_4degr_tilt_LeafA12.jpg

there is some more stuff, but I think this says it all, even as I did this on my livingroom table(one Halogen lamp-the one I use on that table)  as a fast test before delivery of the system.

Regards

Stefan

[ErikKaffehr]

Hi,

Your customer gets value for his money!

Best regards
Erik

Hi Eric

Pretty much on par with my experiences, my actual P45+ will match 4x5" most of the time.
If you pull any tricks and have a very nice slide (velvia 50 or 100) you may have some tiny bit of advantage in Resolution, but only if you do not count in the grain "noise"
which if you remove it will also cut some of the resolution. Whereas if you do a deconvolution sharpening after upres of the P45+  file with software like Genuine fractals you may get better results.
But really - who needs this all of the time ?
We recently sold a Leaf Aptus II 12R with an HCam to a customer who is doing gallery prints with 2x3 meters and maybe larger sometimes.
He was using 4/5" (with 35mm lens) before and is now switching to digital with our TSE -17mm/Hcam.

I had posted some of the images in GetDPI already , but in case someone want´s to see a full res of what this combo did with our Hartblei 80mm take a look here:

http://www.hcam.de/downloads/Hartblei80mm_HCamB1_4degr_tilt_LeafA12.jpg

there is some more stuff, but I think this says it all, even as I did this on my livingroom table(one Halogen lamp-the one I use on that table)  as a fast test before delivery of the system.

Regards

Stefan

[Anders_HK]

That's about what I've suspected - ~20ish mega-pixels on a DSLR = MF film.

In all respect but the above is plain nonsense. In particular when film has been scanned to not highest quality on a Minolta instead of using a quality drum scan. It is sort of using a point and shoot instead of a 20MP dslr...

Had 20MP ZD and 28MP Leaf Aptus 65... both failed quality 6x7 drum scans...

Pretty much on par with my experiences, my actual P45+ will match 4x5" most of the time.
If you pull any tricks and have a very nice slide (velvia 50 or 100) you may have some tiny bit of advantage in Resolution, but only if you do not count in the grain "noise"

In the end digital and film are simply different medias, and matter of preference which prefer. Film is quality. The grain in film is among the qualities.

Having shot 6x7, 617, 4x5 and also 35mm quality slide film, the rendering of a quality slide, correctly exposed within a reasonable DR never failed to amaze me... yet, per my own impression it is first with now  80MP Leaf Afi-II 12 that have made me give up on film. The resolution is not what made me give up film, but the sheer rendering and ability with this back compared to prior generations for a resulting image to better represent a scene per how I experienced it; the quality of colors, range, shadows and very fine gradation of colors... simply brings me in a sense more, but tad different than slids. In the end... I guess simply because shooting both side by side a number of years, my mind also has turned to think more in terms of digital media, and since it is difficult to buy slide film, Quickloads no longer made, difficult to find a quality lab, difficult on flight travels... and difficult to find a quality lab for drum scan and at reasonable price nowadays... simply... because the world has been led to believe that all digital as latest is better...

In the end film and digital are mere different materials... and, a quality slide is magic... but...

When compare please do in quality photographic shoot, and use a quality drum scan! Else you have as in above reduced film to a point and shoot...

 

[ErikKaffehr]

Hi,

I'm in arrangements in scanning one or two of these images on high end drum, at 6000 PPI. The recent comparison was done on Ektar 100. I plan on scanning two images, either the two from this test, or one from this test and a Velvia. Problem with Velvia is that I have no focus bracketed shots, which was done in the Ektar test.

The scan is done at highest quality on the Minolta, it resolves 3200 PPI on MF, and 4800 PPI on 135. An observation is that both images can be improved with proper sharpening. So which image is "sharper" depends on sharpening applied. The digital image has little noise so it takes sharpening better. The film images actually seem to resolve better on high contrast detail.

It will be interesting to compare 6000 PPI drum-scan to 3200 PPI scan.

Stefan Seib has rich experience of both large format photography and drum scanning, by the way. On the other hand he is the developer of Hartblei H-cam, so he has a horse in the digital race.

Best regards
Erik

In all respect but the above is plain nonsense. In particular when film has been scanned to not highest quality on a Minolta instead of using a quality drum scan. It is sort of using a point and shoot instead of a 20MP dslr...

Had 20MP ZD and 28MP Leaf Aptus 65... both failed quality 6x7 drum scans...

In the end digital and film are simply different medias, and matter of preference which prefer. Film is quality. The grain in film is among the qualities.

Having shot 6x7, 617, 4x5 and also 35mm quality slide film, the rendering of a quality slide, correctly exposed within a reasonable DR never failed to amaze me... yet, per my own impression it is first with now  80MP Leaf Afi-II 12 that have made me give up on film. The resolution is not what made me give up film, but the sheer rendering and ability with this back compared to prior generations for a resulting image to better represent a scene per how I experienced it; the quality of colors, range, shadows and very fine gradation of colors... simply brings me in a sense more, but tad different than slids. In the end... I guess simply because shooting both side by side a number of years, my mind also has turned to think more in terms of digital media, and since it is difficult to buy slide film, Quickloads no longer made, difficult to find a quality lab, difficult on flight travels... and difficult to find a quality lab for drum scan and at reasonable price nowadays... simply... because the world has been led to believe that all digital as latest is better...

In the end film and digital are mere different materials... and, a quality slide is magic... but...

When compare please do in quality photographic shoot, and use a quality drum scan! Else you have as in above reduced film to a point and shoot...

 

[ErikKaffehr]

Hi,

I intend to do a 6000 PPI drum scan on the image att a lab in Germany...

Regarding the roses, there may be another explanation, the film used, Kodak Ektar 100 looses resolution faster in reds than blues and greens. Se enclosed figure from Kodak data sheet.

Best regards
Erik

Erik, it's obvious that the red channel is clipped on these roses. How did you do the color neg conversion?

I have seen this from CCD scanners before though it doesn't necessarily have to be the hardware. Sometimes the scanner software clips the highest densities during the negative conversion. In this case it helps to do a raw scan as a positive and then convert manually in PS.

-Dominique

[Stefan.Steib]

@Anders

"The grain in film is among the qualities".........

;-)


Well this is one of the statements that make me smile.
The problem is: The grain is there I have no choice. With a digital file and lets say DXO I can choose any time to have that "grainy look" but I don´t have to........

regards

Stefan

[rickk]

Thanks to Erik for another comparison on this long-term topic.
Although I've never quite managed to duplicate exactly the same scene (in a testing sense) with a DSLR (Sony a900) and 120 film (Mamiya 7 or Fuji 690),
I have occasionally used both film and digital on the same general subject a few minutes apart.
Qualitatively, the results have been mixed, and so, I continue to keep the film cameras around.
The Sony gets vastly more use, just because of convenience and lack of additional expense, and is a totally competent tool.
Nevertheless, for reasons I haven't been able to really identify or quantify, sometimes I just like the "look" of the print from the scanned film better.
Perhaps the film camera gets on the tripod when lighting conditions are optimum, and my impressions are biased as a result.

Regards,     Rick

[Eric Myrvaagnes]

In forty-plus years of working with film I never found grain to be a desirable quality in any of my photos, so I never missed it at all when I switched to digital.

Now if I could only get rid of some of that residual noise...  Oh! But wait! Maybe I should consider noise to be a desirable "quality." And then, how about the "qualities" of burnt out highlights, or unintentional blur, or ...

What a world of possible "qualities" is opening up!

 

[LesPalenik]

That's about what I've suspected - ~20ish mega-pixels on a DSLR = MF film.

That's about right. However, there is a difference between 4.5x6, 6x7, and 6x9 - 6x9 having twice the resolution of 4.5x6, so 6x9 camera may still have an edge.
At the same time, the stitching programs are getting better with each new release, so you can now create large files even with a smaller camera.

I don't think it helps greatly to scan film over 3,200dpi. Now, the quality of scanner really matters. The grain starts showing up at that resolution, and as some other posters mentioned, not everybody considers grain as a desirable feature.

Having said that, I find that older 6x9 and 617 film cameras are still a practical option, especially in dynamic situations when stitching wouldn't work.

[allenmacaulay]

I don't think it helps greatly to scan film over 3,200dpi. Now, the quality of scanner really matters. The grain starts showing up at that resolution, and as some other posters mentioned, not everybody considers grain as a desirable feature.

Fortunately we don't have to speculate since someone else has done the work for us.

http://www.clarkvision.com/articles/scandetail/index.html#testarea2

Velvia scanned at 2700 to 6000 dpi.  Judge for yourself.  Also drum scans compared to conventional film scanners.  It's all there.

[ErikKaffehr]

Hi,

Just a few small points.

1) The test was done on 6x7 frames, and those are 55x69 mm, 6x8 and 6x9 are signifiacntly smaller than indicated by cm size, 6x9 may be closer to 55x80 than to 55x90.

2) The test was done on a Pentax 67 with Pentax 90/2.8 lens. This is not a top performing lens. Mamiya lenses are probably better, what I have seen. Mostly thinking about the "old photodo tests" which indicated that the Mamiya rangefinder lenses were especially good. On the other hand, at f/8 and f/11 (the apertures used) most lenses are pretty close to diffraction limits.

3) Both test images were focus bracketed, 5 exposures within "depth of field according to scale", on the statue image all images were scanned and the best chosen. This was not done on the image on the roses.

Best regards
Erik

That's about right. However, there is a difference between 4.5x6, 6x7, and 6x9 - 6x9 having twice the resolution of 4.5x6, so 6x9 camera may still have an edge.
At the same time, the stitching programs are getting better with each new release, so you can now create large files even with a smaller camera.

I don't think it helps greatly to scan film over 3,200dpi. Now, the quality of scanner really matters. The grain starts showing up at that resolution, and as some other posters mentioned, not everybody considers grain as a desirable feature.

Having said that, I find that older 6x9 and 617 film cameras are still a practical option, especially in dynamic situations when stitching wouldn't work.

[LesPalenik]

Fortunately we don't have to speculate since someone else has done the work for us.

http://www.clarkvision.com/articles/scandetail/index.html#testarea2

Thank you for providing the link, it is a well written and useful article.

I should correct myself. As I said, it depends on the quality of the scanner.
A wet drum scan looks good even at 4,000 dpi. 
I use an older Imacon Flextight scanner with a magnetic holder (no fluid, no glass), and find that even at 3,200dpi, I'm getting too much grain for my liking.   

[Anders_HK]

@Anders

"The grain in film is among the qualities".........

;-)


Well this is one of the statements that make me smile.
The problem is: The grain is there I have no choice. With a digital file and lets say DXO I can choose any time to have that "grainy look" but I don´t have to........

regards

Stefan

@ Stefan,

Smiles are well . While I too prefer digital now, that does not mean that film is inferior and I much value seeing a quality film image no less than a digital image.

1)Stefan Seib has rich experience of both large format photography and drum scanning, by the way.

2) I'm in arrangements in scanning one or two of these images on high end drum, at 6000 PPI.

@ Erik,
 
1) You know I have been around enough to know who you both are . As a photographer we should make our own choices also based on our own educated eye.

2) If you wish to compare both 6x7 to your Sony, may I suggest that you do make pictures of your preferred photography, with quality subjects and quality light? This is similar to what I did using ZD and Mamiya 7 in MFDB vs. Slide Film Challenge in late 2007... http://www.luminous-landscape.com/forum/index.php?topic=20970.0 . My best guess is that the ZD at base ISO have higher image quality than the Sony, pending on subject and due difference in also the formats. However, for film requires a quality made drum scan. My Mamiya system was also SHARP and well CALIBRATED, thus sharp focus.

One of later posts in that thread concluded "The film scans have a real wonderful quality to them. While looking at these last 100% crops, I find that I'm really drawn to them and the ZD images that looked so great before look lifeless in comparison." Thus there is more than pixels and grain that plays in ...

[ErikKaffehr]

Hi!

My plan is right now to make two 6000 PPI drum scans, one from one of mine Ektar test shots and another from a Velvia which I happen to consider very sharp and that doesn't scan really well on my scanner, because of excessive density range. When focusing the enlarger I have seen some detail (I believe) which I did not see in my scan. Interesting to see if it will show up on the 6000PPI scan

I will make all data files available for download.

The Ektar test shots are bracketed for focus.

Best regards
Erik

@ Erik,
 
1) You know I have been around enough to know who you both are . As a photographer we should make our own choices also based on our own educated eye.

2) If you wish to compare both 6x7 to your Sony, may I suggest that you do make pictures of your preferred photography, with quality subjects and quality light? This is similar to what I did using ZD and Mamiya 7 in MFDB vs. Slide Film Challenge in late 2007... http://www.luminous-landscape.com/forum/index.php?topic=20970.0 . My best guess is that the ZD at base ISO have higher image quality than the Sony, pending on subject and due difference in also the formats. However, for film requires a quality made drum scan. My Mamiya system was also SHARP and well CALIBRATED, thus sharp focus.

One of later posts in that thread concluded "The film scans have a real wonderful quality to them. While looking at these last 100% crops, I find that I'm really drawn to them and the ZD images that looked so great before look lifeless in comparison." Thus there is more than pixels and grain that plays in ...

[ErikKaffehr]

Hi,

Download: http://echophoto.dnsalias.net/ekr/images/PublishedTests/index.html

Update2: I discovered a scaling issue and updated the images (castle facede + flowers). Note pixelation on the small flowers. Check this article for explanation: http://www.landscapegb.com/2010/11/the-mysterious-case-of-the-missing-berries-and-other-stories/ (Dominique referred me to this article, thanks a lot!)

Update:  I recently got two films scanned at 6096 PPI by High-End Scans ( www.high-end-scans.de ) in Germany. The scans were made from the Ektar shot at Nynäs and from an old Velvia slide which I think is both sharp and a real challenge to scan. Here are actual pixels.

Top Velvia scanned on Minolt Dimage Scan Multi Pro at 3200 PPI and upprezzed to 16911 pixels width using "bicubic smoother" and resharpened.

Bottom Velvia scanned at 6096 PPI at High-End Scans and slightly sharpened.

Two crops, center crop (maximum sharpness) and off center crop with lens aberration. Radial chromatic aberration was reduced in post.

I also included a comparison of the Nynäs shot, in this case all images resampled to correspond to the 3200 PPI scan. Comparing at 6096 PPI will be done later:

Top left: Sony Alpha 900 uprezzed to 8400 height, bicubic smoother and sharpened

Top right: 60096 PPI High-End scan downsampled to 8400 height, bicubic sharper

Bottom: 3200 PPI scan with my CCD scanner

Some comments:

Quite obvious that film "struggles" on the red flowers for some reason. Very obvious that the 6000 PPI has better detail than the 3200 PPI scan even when downsampled to 3200 PPI. Except for the roses the film seems to outperform 24MP digital. More aggressive sharpening could be used on digital, but I have seen it causing artifacts. Also, finding the correct sharpening for different images is not really that easy.

The results are much dependent on sharpening. Excessive sharpening of digital images is possible but may result in pixelated edges and other artifacts.

Files for download will be posted later this week.


Best regards
Erik

[LesPalenik]

Erik,

thank you for showing the samples from both scanners.
But I'd venture to say that the improved quality of the High-end scans is much more likely due to a higher quality of the drum scanner compared to the Minolta scanner, rather than to the actual scanning resolution.

If possible, it would be useful to add to your comparison a third scanning method - a 3200dpi drum scan (same resolution as the Minolta desktop scanner, but different technology).

Les