Scanning shadows problem

[Boris Mirkov]

I am using VueScan and Canon CanoScan 9000f and alongside all my troubles with scanning I have this shadow problem on every scan that has deep shadows, they all look like this. A lot of time is needed for this to be corrected in PS so I need to know what is it that I am doing wrong so I can get as clear scan as possible. I'm attaching a screenshot of my VueScan settings and the unedited scan.

Thank you in advance for the help.

Boris

 

[L Gebhardt]

It looks like you are underexposing your negatives. There doesn't appear to be any density difference between the shadow areas and the film borders (assuming the black area on the right of the detail scan is film border). If you set the black point based on the film border it will hide the noise in the shadows, while also mostly obscuring the minimal details you have there.

 

[Boris Mirkov]

OK, so I tried that one a moment ago, it seems that VueScan has a mind of its own and does not work in team with mine.
Scanned one frame with blacks set to 0% and the next one to maximum of 11% and literally with no changes...

Thanks for the help.

Boris

 

[pschwart]

I am using VueScan and Canon CanoScan 9000f and alongside all my troubles with scanning I have this shadow problem on every scan that has deep shadows, they all look like this. A lot of time is needed for this to be corrected in PS so I need to know what is it that I am doing wrong so I can get as clear scan as possible. I'm attaching a screenshot of my VueScan settings and the unedited scan.

Thank you in advance for the help.

Boris

You need to set the white and black points and possibly modify the tonal curve. Look at Image->Graph b/w
and Image->Graph curve.

 

[Lamar]

Also you can go to the color tab and use manual then tweak the black point slider so that the areas you want black appear black. White point works the same way for white. If you have the graph b/w showing you will see the pointer movement on the tone curve histogram corresponding to your selection.

 

[L Gebhardt]

While you can certainly make the settings in Vuescan, I prefer to do all the adjustments in Photoshop. It's a lot easier to use a levels or curves layer to adjust the black and white points. Simply being able to visualize the clipping with the option key as you adjust the levels is very powerful. It saves needing to rescan because something got clipped. But if you've mastered the Vuescan interface it can certainly be done there as well. Neither necessarily gives better results.

 

[Lamar]

But getting it as close to the final desired image from the VueScan raw scan data should result in a better starting point for editing. I can usually eyeball it in the preview with the histogram then scan to a raw and an output file. I check the output for blown highlights or lost shadows in my editing software then adjust WP/BP in VueScan and use the save button to create the new image while the raw data is still in memory. No re-scan necessary. I only get about 5% that need adjustment and a re-save.

While you can certainly make the settings in Vuescan, I prefer to do all the adjustments in Photoshop. It's a lot easier to use a levels or curves layer to adjust the black and white points. Simply being able to visualize the clipping with the option key as you adjust the levels is very powerful. It saves needing to rescan because something got clipped. But if you've mastered the Vuescan interface it can certainly be done there as well. Neither necessarily gives better results.

 

[L Gebhardt]

But getting it as close to the final desired image from the VueScan raw scan data should result in a better starting point for editing. I can usually eyeball it in the preview with the histogram then scan to a raw and an output file. I check the output for blown highlights or lost shadows in my editing software then adjust WP/BP in VueScan and use the save button to create the new image while the raw data is still in memory. No re-scan necessary. I only get about 5% that need adjustment and a re-save.

On most scanners it shouldn't matter as long as the file is saved as 16 bits. Some drum scanners can adjust the gain on the amplifiers based on the scan settings. In that case getting close in the scanning software will show an improvement. But for CCD scanners they capture the same set of data no matter what the scan settings. Then the scanner software manipulates the data. In theory you should get better results if you do all the work in Photoshop with adjustment layers, since no data is discarded until the end. But that does require you to know how to duplicate what Vuescan is doing in Photoshop, which may not be easy.

 

[Lamar]

If you save and work in tiffs in your work flow it makes sense. I save a VueScan RAW tiff and a 100% quality jpg on the scan. The RAW is so I don't have to rescan if at some point later I wish to rework the image, I delete them after a month or so. For me the jpg is my working image making for a speedier work flow and smaller files with 8 bits per channel giving no discernible difference in my final image output as long as I get it close in VueScan. I have tried many different variations of workflows and have found this to work best for me but I understand the argument and did exactly what you describe for a while.

On most scanners it shouldn't matter as long as the file is saved as 16 bits. Some drum scanners can adjust the gain on the amplifiers based on the scan settings. In that case getting close in the scanning software will show an improvement. But for CCD scanners they capture the same set of data no matter what the scan settings. Then the scanner software manipulates the data. In theory you should get better results if you do all the work in Photoshop with adjustment layers, since no data is discarded until the end. But that does require you to know how to duplicate what Vuescan is doing in Photoshop, which may not be easy.

 

[glhs116]

On most scanners it shouldn't matter as long as the file is saved as 16 bits. Some drum scanners can adjust the gain on the amplifiers based on the scan settings. In that case getting close in the scanning software will show an improvement. But for CCD scanners they capture the same set of data no matter what the scan settings. Then the scanner software manipulates the data. In theory you should get better results if you do all the work in Photoshop with adjustment layers, since no data is discarded until the end. But that does require you to know how to duplicate what Vuescan is doing in Photoshop, which may not be easy.

This is a common point of confusion. The CCD gain cannot be adjusted per channel but the exposure time can. It is either a separate lamp per colour like the Coolscan or a separate strip of elements behind different colour filters like most flatbeds. This is one of the chief advantages that scanners hold over DSLR scanning especially for colour neg.

 

[Doyle Thomas]

most of the time a good color scan should contain both a pure black and a pure white. a BW scan should contain one or the other. The reason for this is to provide the eye with reference points. The more you can do at the time of scan the better.

 

[L Gebhardt]

This is a common point of confusion. The CCD gain cannot be adjusted per channel but the exposure time can. It is either a separate lamp per colour like the Coolscan or a separate strip of elements behind different colour filters like most flatbeds. This is one of the chief advantages that scanners hold over DSLR scanning especially for colour neg.

Thanks for the clarification on the exposure time. I knew about the Nikon's color ability, but had forgotten. As far as I know my 4870 and CanoScan FS4000US do not, or it's not accessible through the software. I'll see if VueScan can access that feature for either of them.

As far as scanning color negatives with a DSLR you can use a color enlarger head to adjust the light color. This gives you a nice neutral image to invert and lets you expose to the right. Done this way I get cleaner results than with the two above scanners (but I still find some issues where a color isn't quite right).

 

[glhs116]

Thanks for the clarification on the exposure time. I knew about the Nikon's color ability, but had forgotten. As far as I know my 4870 and CanoScan FS4000US do not, or it's not accessible through the software. I'll see if VueScan can access that feature for either of them.

Nikon Scan puts the controls directly at your finger tips. Epson Scan does it based on your per channel histogram adjustments. Even HP Scan will do it if you use the little colour wheel thingy to adjust the white balance when scanning neg as pos. All scanners have to do this to scan colour neg but most don't give you a direct way to control it when scanning negs as positive. With up to three stops exposure difference between red and blue on colour neg the cheaper scanners wouldn't be able to scan color neg at all without doing this.

The colour head is a great solution if you have one. I had to use colour filters. Amazing how many people think RAW can somehow invalidate the laws of physics. I gave up on DSLR scanning because without a bellows or copy stand it is too hard to keep the film plane perpendicular to the required accuracy.

 

[Hatchetman]

I have this scanner and it is not great by any stretch. But I don't think I've gotten such poor results unless the negative was seriously flawed. I turn off all the scanner settings (auto-sharpening, etc) and do the processing myself in Photoshop.

 

[chuck94022]

I had this scanner in the past and never saw results like this. (I currently have an Epson flatbed.)

Looking at the pattern on the negative strip (first image) I'm inclined to suspect a negative processing issue.

Have you looked carefully at the negative on a light table with a loupe?

 

[Alan Klein]

Doesn't that depend on which scanner? Some flat bed speeds cannot be changed, I believe? (V600 that I use for example)

On flatbeds where you can change the speed, how does that work? Do you have any examples you can point us too that shows the differences between multiple scans and single scans/deifferent speed scans. I've heard a lot about this but never saw any examples that were meaningful in their end result.