Something common to test fine resolution on my scanner?

Hi everyone,

 

My first post! I know scanners are not Photo.net's focus per se, but this forum seems really active, and I couldn't find anything related and very active on the net...

 

I bought a new Epson V600 scanner to digitize almost a thousand old family slides. The scanner's specs say it has a hardware resolution up to 6400dpi, and up to 19200dpi with interpolation.

 

In testing on slides, zoomed up close on the scans produced, I found that the best scans were at resolutions of 4800 and 6400dpi, but I couldn't really tell the difference between them. (There are some very subtle differences, but whether they matter, I don't know.) Anything higher showed clear signs of interpolation, so that's out. Of course, scans at 4800dpi are a little faster, and the resulting pics are smaller in size.

 

I'm trying to make a decision about what to use for the thousand slides, before I start. Realistically, it will probably never be done again. So I'd like to get the best scans I can, once and for all.

 

These old slides (typical family pictures) aren't the best thing for testing viable resolution. It's hard to tell what might even be a true hard edge or whatever.

 

So I flailed around on the net and found, e.g., the Silverfast resolution target (USAF 1951). But I'd rather not pay $69 plus $10 for shipping on something I'll basically only use once. I'm not even sure it's what I really need here.

 

With cameras, it's a little different ... you can always get farther away from, e.g., a printed target, etc. Not true with a scanner ... it has to actually be something of super-fine resolution (clear edges for something 6400 dpi or better, for my purposes).

 

I tried to print, e.g., some test charts I found on my 1200 dpi printer. But as expected, 1200dpi wasn't good enough to show any difference.

 

Okay, all of this leads to my question...

 

Surely there's some sort of thing found naturally in most homes that could be used as some sort of resolution test? Hair ... magnetic shavings ... something? Of course, it won't provide any sort of absolute metric ... but it ought to let me see if there's really a difference between 4800 and 6400dpi on my scanner.

 

Any thoughts, anyone?

 

Thanks! And thanks for having your forum.

 

Mike

Atlanta GA USA

This is a commercially available, high resolution, high contrast target for testing scanners, among other things.

 

LaserSoft ImagingSilverFast Resolution Target

 

That said, there is o difference between 4800 and 6400 lpi on a flatbed scanner, which is unlikely to exceed 2200 lpi. That's still higher than the 1200 lpi (or less) you can produce on a laserjet printer.

I have V600 and scan at 2400. 48 bit. The other things you have to deal with is are you going to let the scanner adjust exposure automatically or do it in post. Also, you have to eliminate a lot of the dust using ICE. Much slower scans. Also, you need a lot of time to adjust after the scans and eliminate any dust left over. There's always fine tuning even if you let the scanner auto adjust.

 

I would try different ways of scanning with a few pictures. When you determine which works best for you, then do the rest. Scanning can get frustrating so be prepared.

I mis-typed my response. It should read, "There is no difference ...", because 6400 and 4800 ppi are grossly overstated, simply puffing up the size of files without adding any information.

 

A simple resolution test would be to scan an image of a black strip and magnify the results, digitally, to see how many pixels it takes to go from black to white. Reduce the resolution until the transition occurs over a single pixel width. You can't do that with film, other than with a few technical emulsions, because edges are never sharp when you look that closely, for example with a 10x magnifier.

Scanning over 2400 spi on a flatbed is pretty much a waste of time. For slides, I would get a Plustek or something similar.

Somewhere I have a 2" square resolution test glass plate made on a practically grainless 'Lippmann' emulsion.

 

I went through the process of testing a few scanners with it some years back.

 

The dedicated 'bread-tin' size scanners all came out of it well, returning an optical resolution close to, or at the pixel-per-inch theoretical resolution. I can't say the same for any of the flatbeds. The best of them approached 3200 ppi, no more.

 

Ed's idea of scanning a sharp edge and counting the number of pixels diffusion will give you a very approximate idea, but flare and diffraction across the opaque/transparent border might well give a false impression, as would any focussing error. You really need a proper resolution plate to get a true reading.

 

A razor-blade edge or cut piece of aluminium kitchen foil would give you a sharp enough edge for Ed's rough test.

 

Personally, I'd just do something easy and practical; like see how sharply the grain of a medium-speed B&W film is defined in the scan. A film like FP4+ will give a grain that should be easily discernible in a sharp scanner.

 

Whatever. I wouldn't expect anything better than a true optical resolution of 3200 ppi or lppmm from any flatbed, no matter how many ppi it outputs digitally.

 

P.S. I suspect that Siverfast target will just be a mass-produced and quite poor contact copy of a resolution chart on fine-grain film, not on a glass Lippmann plate. I might be wrong, but I certainly wouldn't risk $99 to find out. Maybe someone reading this has bought one?

 

"For slides, I would get a Plustek or something similar."

 

- The Pacific Image/PrimeFilm scanners are better than Plustek's IMO.

Edited November 18, 2018 by rodeo_joe|1

....and up to 19200dpi with interpolation.

 

- LOL.

19200 ppi is equal to a feature size of 1.3 microns, and due to diffraction, that's not even physically possible using white light.

 

It would be a total waste of time and digital storage scanning at that bogus 'resolution'.

Hi Mike

Good question yet I am going to take ad different approach to your question. That is, since these are family pictures, what resolution do you need for usage purposes.

 

4K screens (pretty good viewing) have a resolution of 3840 x 2160

8x10 prints at 300 dpi are 3000 x 2400 dpi

 

For most all direct viewing uses 4800 dpi should be sufficient.

 

If you are going to crop into a small subset of the image you might need more

 

Not you also need to decide if you want 8 bit dept sanning or 16 bit scannng and which storage format. If you are going to do significant post process that 16 bit an TIFF would be good. If this is just for quick viwing of images than 8 bit and storing as JPG should do

 

If the number of keeper keeper images is really small, you could always scan at lower resolution 4800 dpi or even 200 dpi and if you ever (maybe unlikely) where you need higher resolution, you pull up the scant few slides that need it and resacn at 4800 or maybe 6400 dpi. I bet you find that it would be rare you would need a higher resolution scan for family photos which are for jogging your memories rather than ones needed the finest resolution .

 

BTW - If you have the best low grain film, high contrast subjects using the best possible lenses on the camera for those family pictures you can see a diffeence between 4800 and 6400 dpi by pixel peeping yet do you need it for your viewing pleasure even if all those other conditions were met.

 

Just another way of looking at it.

 

Hope this helps.

take a picture of a brick building at infinity focus. scan it n see if you can see the details of the bricks?

Hi everyone,

 

Sorry for the delay - I tested more based on your replies above back then, but then I got sidetracked...

 

First, for John Wheeler - I am trying to get the best resolution I can for posterity. I imagine disk space will get ever cheaper in the future, but no one may ever scan these pictures again. So, never mind current disk space or even current intent IF I am getting worthwhile information at higher resolution. (Are you in Atlanta, by chance? I know a JW there.)

 

Rodeo Joe, I suspected something like that might be the case - people cranking out mass-produced "master" target images that aren't actually any such thing. So I didn't want to drop $79 when, even if it was for real, I'd only need it for tests of only a few hours. That's why I made this thread. Still, it would be great if there were reliable high-resolution scan targets for cheap. Now I know that there probably aren't.

 

On to the meat of the matter -

 

Ed Ingold,

 

I ran with your idea, and put a bunch of... well basically, thin trash that might have sharp edges ... on my scanner:

• Some plain old paper - part of a NetFlix envelope
  
• A common plastic tan grocery bag (in the southeast U.S.). The idea was, it's so thin, hopefully it will have a hard edge
  
• A cut-out of the nutrition label from a bag of chips. (Washed, of course.)
  
• A nutribar faux foil wrapper (I think it's actually super-thin plastic with metallic spray)
  

I scanned at various resolutions, all the way from 300 to 6400 dpi and above.

 

First, the grocery bag seemed fruitful. I even saw teeny tiny hairlets from a nylon mitten I'd been wearing! But the microscopic hairs were too curly and had no real edge to hold onto, for testing resolution...

 

The NetFlix paper, including a little tear, was interesting to see. But not ultimately useful.

 

Next, I thought the nutribar foil's edge would be super-precise ... but really, it was just an edge. Not sure how much I can do with that.

 

But the thing that ultimately seemed to have the most coherence was actually the rounded edge of the nutrition labelling's ink (a corner of the printed frame around it), on the chip bag. Why? I suppose because, when you are super-magnifying things and they may be subject to all kinds of whatever, you're not sure if a straight line is still in focus ... but a curve gives your mind more to work with.

 

On to the actual question -

 

There was a clear point where, from very low rez to ~4800dpi, it became increasingly less blocky (sort of like pixelation). Then above 6400dpi, it started to become softer, out of focus. I know anything above 6400 dpi must be interpolated (the scanner's specs say so). It was sort of like a U-shaped curve... first very blocky, then more precise, then ultimately... going to less in focussed or blurry. As one moved from 300 to 2400 to 4800 to 6400 and beyond.

 

So:

 

1) Some of you have said 2400dpi is the max on a "flatbed" unless I get a "Plustek". Have you actually used/tested an Epson V600 or similar scanner rated at 6400 dpi? Can you show me side-by-side scans from a Plustek vs something like a V600 at a stated 6400 dpi? I don't believe you until you can demonstrate it. Thanks.

 

2) I interpret what I saw as a U-shaped curve, going from blocky to clear to blurry, where the true resolution is either 4800 or 6400 dpi. Does this make sense?

 

Thanks for all your input! This is so helpful.

 

Who needs a fake $80 target image when you've got a bag of chips anyway?

Edited December 2, 2018 by mikef7

Hi Mike

No I am not in Atlanta.

There are quite a few things that go into the best image from scanning so thought I would attach a link to this article about a person that did resolution testing of his Epson V700. My conclusion is that at the higher settings you are likely not going to get better resolution yet there is little downside to the resolution if you do scan at the highest setting (other than scan time and file size). Good luck with your project.

Optimum Resolution & Sharpening Settings for Epson Scanners

Short of a proper scanning target, an old-fashioned razor blade (unused) is going to give you the best information about scanner performance. If the razor-blade is positioned at a slight angle, you should see a 'staircase' of pixels down its edge. The spread and contrast of pixels is directly related to the lens resolution, with a small pixel spread and high contrast indicating a high-quality lens with good resolution.

 

This method doesn't give you a numerical figure for resolution, just a qualitative 'feel' for how good the system is. It also doesn't show you whether any lack of definition is due to a focussing error or a basically poor lens. However, colour fringes are a definite indicator of an inferior lens.

Scanning Targets are no better than actually scanning what you have, IMHO.

Flatbed scanners are notoriously "over-spec'ed" -- Scanner Features and Definitions - Resolution

My personal flatbed scanners promising 12000 or whatever ppi are nowhere near as good as my various 9000ppi actual film scanners (Nikon Super Coolscan 9000 and Canon 4000).

 

Some of my posts on this topic:

Huge Scanning Job Finished (yet again!) rescan

2015: A scanning Odyssey - Nikon Super Coolscan LS-9000 ED Nikon

CanoScan 9000F vs. CanoScan FS 4000US Canon vs Canon

Nikon Coolscan LS9000 ED, ICE, and CanoScan F4000US - Part 2 Canon vs Nikon

Thoughts on Theory and Practice of Scanning (Archival/Forensic) forensic scanning

 

There is also a iimitation of how much resolution the original film image has.

Practically 9000 ppi resolves film grain/texture on all but a few very specialized scientific films, meaning they've gone about a far as you can go.