Starting my focus calibration of V850 using resolution target

[Steven Ruttenberg]

from Stouffers. Should be interesting. I am starting with what I had come up with by scratching an exposed negative with a razor blade and straight edge. Will see what this yields as my base. Not only for focus, but for resolving power. All files are wet scanned using Kami scan fluid and optical mylar on ANR glass from Betterscanning and the better scanning variable height holder. My determined height initially is .125mm or 3.048mm.

[Steven Ruttenberg]

Here is the initial scan. Setting is 4200dpi, scanned as a slide at 48bit RGB and outpout as a linear raw tif file. I am hesitant to mess with the focus settings as you can read the entire scale, which shows this is better than 20lpmm or 500lpi. I applied zero sharpening, no adjustments whatsoever, etc. Exported as a jpg at 100% with bicubic only (not auto, etc) Next I will try as bw negative scan 48bit rgb then output to 16bit grey. Next I will vary the scan resolution in increments of 300dpi down, then up from 4200. When I find the best setting from that, I will then adjust focus. Finally, I will then scan the Stouffer 31 step wedge at .1 increments in steps. It is the TP 4x5 31-step transparency.

If anyone would like the full size file to look at (full tif file) let me know and I will put on dropbox. It is a 435mb size file.


[IMG]20181120_1-T_Stouffer_Resolution_Target_4200dpi_48bitrgb_48bitrgb_gray_BW_Negative_003 by Steven Ruttenberg, on Flickr[/IMG]

[Steven Ruttenberg]

The full size file tif and jpg look better before upload, but still you can get my point. This is a fine scanner.

[Steven Ruttenberg]

This is gonna take a while. One thing i have notices is that the slide setting looks a bit better than the b/w negative setting when scanning as 48bitrgb whether output as 16bit gray or 48bit rgb, the two versions of b/w files look identically the same, but slight worse than the slide version.

[Steven Ruttenberg]

Okay, I finished all scans from 2100dpi thru 6300dpi. Scans at each dpi were 48 bit color slide output 48 bit color, 48 bit color bw output 48 bit color, 48 bit color bw output 16 bit gray, finally 16 bit gray output 16 bit gray. Definitely gets better the higher the resolution, but need a finer resolution chart to see for sure. Also, my focus setting is good, but I need to tweak it. First I need to find the resolution that appears the best out of all of them. Then tweak at that dpi. Definitely though don't want anything below about 3600dpi more like 4000 plus. It appears that things are much better at 6300dpi for what I scanned, although 4200 looks pretty good as well which is currently my scan dpi.

Link to all of the files for those interested in looking at them. I hope the titles are not too confusing and self explanatory.

Feel free to download the files if you want to play with them if flickr will let you. I can send the full size tif file for anyone who wants to play with those.

https://www.flickr.com/photos/157376...7673732821157/

[Pere Casals]

Amazing test !!

One thing, all scans look made with the bars vertically arranged. Probably if arranged horizontally it would be noticed a lower performance, it would be interesting to also post that to see the difference.

[Ted Baker]

One thing i have notices is that the slide setting looks a bit better than the b/w negative setting when scanning as 48bitrgb whether output as 16bit gray or 48bit rgb, the two versions of b/w files look identically the same, but slight worse than the slide version.

This is an example of how contrast effects your results. The BW version is not just the pixel data of the slide upside down. i.e. it is not 1-x

I am finding it hard to interpret what you have here, other than one looks better than other

If your happy to make the original tif available I would like to have a go at using ImageJ to do a slanted edge MTF test. This should give a resolution at each contrast level. They need to be raw scans, or scanned as a slide with no contrast adjustment, otherwise the data doesn't make sense.

https://imagej.nih.gov/ij/plugins/se-mtf/index.html

[Steven Ruttenberg]

Possibly could be, but they are at an angle so that should pick up some effects degradation if any. Once I set focus at chosen dpi, I may do a horizontal scan at that final setting for comparison. The flip sjde is, whatever is best from this direction by default is best horizontally. So for information purposes I will do horizontal scan.

[Steven Ruttenberg]

This is an example of how contrast effects your results. The BW version is not just the pixel data of the slide upside down. i.e. it is not 1-x

I am finding it hard to interpret what you here here, other than one looks better than other

If the your happy to make the original tif available I would like to have a go at using ImageJ to do a slanted edge MTF test. This should give a resolution at each contrast level.

https://imagej.nih.gov/ij/plugins/se-mtf/index.html

Well, what looks better than the other is part of what I did. I am looking for what "looks" best then uzing that dpisetti g adjust focus to get best result. This wasn't intended as a resolution test per se, but a focus test. However, I am surprised at the results and how they definitely improve through 6300dpi, although, past about 4k or 5k dpi it is about quality vs resolving power with better quality at higher dpi.

I will put files into drop box today sometime and post up link. Will be interesting to see what mtf is although in the end I trust my eyes on what looks best .

[Bernard_L]

Kudos to the OP for doing actual tests, but scratching my head. The scanner claims resolutions of 4000+dpi and you test it with a target that maxes at 500lpi. What am I missing?
And, your focus increments, as in 2700, 3000, 3300dpi, etc... are not hardware settings; they represent the result of some software data massaging; so, even though you perform no explicit sharpening, the perceived quality is more a judgement on the quality of the hidden re-sampling software.