Tweet
By Sharon Brunson on Wednesday, May 02, 2001 - 05:52 pm:
Doug, you mentioned elsewhere in the forum that you use multiple scans that can reach up to 75mb. What software do you use and would you mind sharing a little more about your scanning techniques? Thanks, Sharon
By Doug Nelson (doug) on Thursday, May 03, 2001 - 10:20 am:
My own scanning setup is woefully undernourished, so I've learned to make do.
First, a scanner test:
Scan something with a lot of fine, precise detail (a dollar bill works fine) into your favorite editing software. Use the highest-quality settings available (real, not interpolated), with no processing of any kind. Save.
Now, scan again, changing nothing.
Copy the second scan, and paste it into the first as a new layer. Save.
Switch the blend mode of the new layer to 'difference' (this is a Photoshop setting, other software probably has a similar mode. It simply highlights any differences between layers).
Examine in highly magnified mode. Can you see anything? It should be completely black, with nothing else visible at all.
If this has been successful, your scanner has the repeatability necessary for multi-layer scanning.
The basic fault of most scanners, particularly less expensive scanners, is that they introduce noise into the darker areas. This manifests as random lighter dots. They aren't the same from scan to scan, so if we switch the new layer's mode from 'difference' to 'darken', the darkest value of each scan will be combined, eliminating any noise spots that aren't accidentally aligned.
Flatten, scan again, paste again, mode to darken again, flatten again. repeat. (I usually do four, but this is overkill).
This is the most elementary area of multi-scan. Some higher-end scanners, particularly film scanners, can do this all automatically. There's a lot of room for error, but again, if you're stuck with what you have you make do.
Now you can experiment. If your software allows it, you can do scans at different lightness levels, expanding the range beyond what most scanners can perform in a single pass. For example, in the tintype challenge, you could make a layer scanned to optimize for the stained area.
These layers should be left separate.
I have to stress, this whole procedure is a bad mistake if your scanner shows even the slightest bit of misalignment during the 'difference' test (someone please post what this mode is called in other software editors).
Always scan at the highest bit level your scanner allows (many offer 16-bit nowadays). Since Photoshop's support for high-bit images is pathetic, use whatever tools your scanner manufacturer supplies to work with the highbit image before moving to the editor's tools, and don't convert to 8-bit until you absolutely must. This is particularly important in B/W images.
I'll let people experiment awhile before I inflict any more
Again, translation into PaintShop Pro, Painter, Photopaint, etc., terms would be appreciated.
By Sharon Brunson on Thursday, May 03, 2001 - 11:57 am:
Wow, I appreciate the detailed information. Can't wait to try it out. Thanks, Sharon
By Sharon Brunson on Thursday, May 03, 2001 - 02:14 pm:
Okay, I tried the difference test. I don't have a dollar here so I used a picture. At 300dpi the scan was completely dark but at 600dpi and 1200dpi the difference test looked kind of like a negative - everything had a semi-light outline. I completed your instructions on the 300dpi picture and it turned out great. How do the different resolutions change this? I am using VueScan software. And I have an Epson Perfection 636U scanner. Thanks, Sharon
By LindaJ on Tuesday, May 08, 2001 - 09:50 pm:
Sharon, I have Vuescan (Version 7) also. I notice that one can make multiple passes with it. I have used the multiple pass feature and it has helped my scans. Have you tried the feature?
I wonder if multiple scans with Vuescan will accomplish the same as the above procedure? My scanner is an Epson 1200U.
Linda
By Doug Nelson (doug) on Tuesday, May 08, 2001 - 10:05 pm:
Sharon:
looks like your scanner passes at low-rez, but fails at higher resolutions. Any gains you might see would be lost by giving up the higher resolution. Try Vuescan multipass, it may offer an alternative method.
If you try this with misaligned images you'll actually multiply noise instead of eliminating it.
By Sharon Brunson on Wednesday, May 09, 2001 - 11:40 am:
Hi Linda, I have tried the multipass in Vuescan. I did notice that Doug's method produced better results but I'm going to take his advice and stick with Vuescan's multipass since my scanner won't handle the higher resolutions in Doug's method. I bought my mother the Epson 1200U and I really liked it. Sharon
Doug, you mentioned elsewhere in the forum that you use multiple scans that can reach up to 75mb. What software do you use and would you mind sharing a little more about your scanning techniques? Thanks, Sharon
By Doug Nelson (doug) on Thursday, May 03, 2001 - 10:20 am:
My own scanning setup is woefully undernourished, so I've learned to make do.
First, a scanner test:
Scan something with a lot of fine, precise detail (a dollar bill works fine) into your favorite editing software. Use the highest-quality settings available (real, not interpolated), with no processing of any kind. Save.
Now, scan again, changing nothing.
Copy the second scan, and paste it into the first as a new layer. Save.
Switch the blend mode of the new layer to 'difference' (this is a Photoshop setting, other software probably has a similar mode. It simply highlights any differences between layers).
Examine in highly magnified mode. Can you see anything? It should be completely black, with nothing else visible at all.
If this has been successful, your scanner has the repeatability necessary for multi-layer scanning.
The basic fault of most scanners, particularly less expensive scanners, is that they introduce noise into the darker areas. This manifests as random lighter dots. They aren't the same from scan to scan, so if we switch the new layer's mode from 'difference' to 'darken', the darkest value of each scan will be combined, eliminating any noise spots that aren't accidentally aligned.
Flatten, scan again, paste again, mode to darken again, flatten again. repeat. (I usually do four, but this is overkill).
This is the most elementary area of multi-scan. Some higher-end scanners, particularly film scanners, can do this all automatically. There's a lot of room for error, but again, if you're stuck with what you have you make do.
Now you can experiment. If your software allows it, you can do scans at different lightness levels, expanding the range beyond what most scanners can perform in a single pass. For example, in the tintype challenge, you could make a layer scanned to optimize for the stained area.
These layers should be left separate.
I have to stress, this whole procedure is a bad mistake if your scanner shows even the slightest bit of misalignment during the 'difference' test (someone please post what this mode is called in other software editors).
Always scan at the highest bit level your scanner allows (many offer 16-bit nowadays). Since Photoshop's support for high-bit images is pathetic, use whatever tools your scanner manufacturer supplies to work with the highbit image before moving to the editor's tools, and don't convert to 8-bit until you absolutely must. This is particularly important in B/W images.
I'll let people experiment awhile before I inflict any more
Again, translation into PaintShop Pro, Painter, Photopaint, etc., terms would be appreciated.
By Sharon Brunson on Thursday, May 03, 2001 - 11:57 am:
Wow, I appreciate the detailed information. Can't wait to try it out. Thanks, Sharon
By Sharon Brunson on Thursday, May 03, 2001 - 02:14 pm:
Okay, I tried the difference test. I don't have a dollar here so I used a picture. At 300dpi the scan was completely dark but at 600dpi and 1200dpi the difference test looked kind of like a negative - everything had a semi-light outline. I completed your instructions on the 300dpi picture and it turned out great. How do the different resolutions change this? I am using VueScan software. And I have an Epson Perfection 636U scanner. Thanks, Sharon
By LindaJ on Tuesday, May 08, 2001 - 09:50 pm:
Sharon, I have Vuescan (Version 7) also. I notice that one can make multiple passes with it. I have used the multiple pass feature and it has helped my scans. Have you tried the feature?
I wonder if multiple scans with Vuescan will accomplish the same as the above procedure? My scanner is an Epson 1200U.
Linda
By Doug Nelson (doug) on Tuesday, May 08, 2001 - 10:05 pm:
Sharon:
looks like your scanner passes at low-rez, but fails at higher resolutions. Any gains you might see would be lost by giving up the higher resolution. Try Vuescan multipass, it may offer an alternative method.
If you try this with misaligned images you'll actually multiply noise instead of eliminating it.
By Sharon Brunson on Wednesday, May 09, 2001 - 11:40 am:
Hi Linda, I have tried the multipass in Vuescan. I did notice that Doug's method produced better results but I'm going to take his advice and stick with Vuescan's multipass since my scanner won't handle the higher resolutions in Doug's method. I bought my mother the Epson 1200U and I really liked it. Sharon
Ding!
Comment