Which DPI To Use When Scanning Film

Picture North Carolina wrote in post #11638804
I understand what is being said about flatbed optical resolution exaggerations. That being said and accepted, would it not be safe to say that scanning a neg at up to the manufacturer's stated optical resolution would be safe and produce the best quality?

Not according to Sascha Steinhoff:

http://www.amazon.com …93395230X/ref=o​ss_product

Select 2400 dpi. Anything above that will waste disk space with no additional benefit. Alternatively, invest in a dedicated film scanner.

Tony-S wrote in post #11639062
Not according to Sascha Steinhoff:

http://www.amazon.com …93395230X/ref=o​ss_product

Select 2400 dpi. Anything above that will waste disk space with no additional benefit. Alternatively, invest in a dedicated film scanner.

... or save a lot of time spent feeding a scanner and use a scanning service like ScanCafe.

Bob_A wrote in post #11647593
... or save a lot of time spent feeding a scanner and use a scanning service like ScanCafe.

And hope they get it right.

3600 dpi is a good starting point. However, 5000, can be required to full in the spaces where the histogram looks like a comb. Whatever the true resolution, the larger figure certainly renders fine detail better.

corkneyfonz wrote in post #11648616
3600 dpi is a good starting point.

Not with a flat bed scanner.

Tony-S wrote in post #11647681
And hope they get it right.

True, but it's a decent choice for someone that knows little about scanning negs.

I have my own Nikon Super Coolscan 5000ED, so have no need for ScanCafe, who use the same equipment. I'd definitely try them or a similar service over scanning with a flatbed scanner.

As I have been archiving my old pictures this thread got me curious as to my process. I've been scanning all my pics at 4800dpi tiff files and wondered if it was overkill. So tonight I ran a small experiment. I scanned one negative from 600 dpi to 6400dpi as a tiff. They were scanned with no processing and no changes between scans except for the resolution.

Then I loaded them into photoshop, selected one area where it was easy to see the detail and zoomed in each pic so that the same area was visible on the screen.

I then took a screen shot and copped out a 1000x1000 pixel area. So what your seeing is the same area at the different resolutions.

At 600 DPI I was zoomed in so much that the pixelization was very evident. The change from 600 to 1200 was very noticeable even on the whole image. After 1200 DPI there was no visible change when the whole image is viewed on the screen but when zoomed in it was very noticeable.

From my little experiment there was a noticeable difference in detail between scans up to 4800dpi. The jump from 4800 to 6400 was much less noticeable.

very interesting scanning experiment... What scanner are you using?

So, it's been a long time since I've done any scanning -- how big can you print from a negative/transparency scanner before things turn ugly?

I have had similar results scanning 35mm slides on my Epson V7000 scanner. The scans look real good from 1600 to perhaps 3800-4800 maximum. After that I can't tell any difference. But then I am not an expert by any means. I have been using VueScan instead of SilverFast and the Epson software, and saving the images to "dng" format, which in turn I can easily process with CS5. Also, I set VueScan so that it takes one light pass of infrared dust removal (whatever that means). What I like about VueScan is that the User's Manual has a lot of information relating to scanning for the best image quality, and this makes it easier for me.

tonylong wrote in post #11650523
So, it's been a long time since I've done any scanning -- how big can you print from a negative/transparency scanner before things turn ugly?

As a point of comparison, the 5D2's sensor resolution is 3,900 spi (sensels/inch). This means that a 13x enlargement (12.28 x 18.42 in.) is 300 ppi without resampling. A true 4800 spi (samples/inch) scan would become 300 ppi at 16x, so 15.2 x22.8 in. for optimum quality. Of course we all know that sub-optimum can still be darn good, but that starts to involve subjective factors.

tzalman wrote in post #11651267
As a point of comparison, the 5D2's sensor resolution is 3,900 spi (sensels/inch). This means that a 13x enlargement (12.28 x 18.42 in.) is 300 ppi without resampling. A true 4800 spi (samples/inch) scan would become 300 ppi at 16x, so 15.2 x22.8 in. for optimum quality. Of course we all know that sub-optimum can still be darn good, but that starts to involve subjective factors.

So, do you think scanners can really deliver that kind of "real" resolution from negatives and slides?

KurtGoss wrote in post #11650454
very interesting scanning experiment... What scanner are you using?

Epson V500

gjl711 wrote in post #11650190
As I have been archiving my old pictures this thread got me curious as to my process. I've been scanning all my pics at 4800dpi tiff files and wondered if it was overkill. So tonight I ran a small experiment. I scanned one negative from 600 dpi to 6400dpi as a tiff. They were scanned with no processing and no changes between scans except for the resolution.

All this does is show the pixels. It does not determine the resolution. For that, you need to buy a USAF-1951 resolution test target to determine the actual resolution. Virtually all flat bed scanner manufacturers (including Epson - specifications tab, note 1) use the ISO 14773 standard for advertising scan resolution, instead of the more useful ISO 16667 standard. Steinhoff calls this the "resolution swindle". The Epson Perfection V500 tops out around 1800 dpi (rated by Epson at 6400 optical) and the V700 tops out around 1900 dpi.

tonylong wrote in post #11651375
So, do you think scanners can really deliver that kind of "real" resolution from negatives and slides?

No, not the flat bed scanners (if you're talking about a 35mm slide or negative).