With the V700 and fluid mounting, I get about 2200 spi. I have found no difference with the 6400 setting and lens. The key is obtaining correct focus.
With the V700 and fluid mounting, I get about 2200 spi. I have found no difference with the 6400 setting and lens. The key is obtaining correct focus.
To David's point, you can get a modest but helpful improvement in resolution, by simply raising the subject off the glass by a small amount.
This will also prevent you from seeing Newton's Rings, when the film comes in contact with the glass. Unfortunately, the effect can be seen, even when we place the film emulsion-side-down.
If you search this forum, you will find that in order to accomplish this, some people have devised a variety of do-it-yourself methods, while others use the custom "Variable Height" holders made by www.betterscanning.com.
I'm talking about the setting above which the optical lens in the scanner is not resolving anymore detail, but honestly I have never tried the 6400 setting. I'm developing a bunch of new 4x5 film tonight from a recent trip so I'll scan a couple of 6400 and compare to my method of scanning.Originally Posted by Keith S. Walklet
I also found this on the Epson site which seems to suggest there is something extra at work, the "Micro Step Drive™ technology", when scanning at 6400.
Hardware Resolution
* 4800 x 9600 dpi, 6400 x 9600 dpi with Micro Step Drive™ technology
Optical resolution is determined by how many pixels the scaner can sample. A flatbed scanner with a 1200 sensors per inch can sample 1200 spi in one direction, the width of the CCD. To sample in the other direction the CCD is moved along the page, and more samples may be taken in this direction depending on the stepping motor. For example,if the scanner takes 2400 samples per inch in the direction of the movement of the head, it would be called a 1200X2400 spi scanner. Strictly speaking either of these figures could be called optical resolution, though for most scanners the first, or lowest, number is the only one that matters.
Interpolated resolution is based on how many pixels the the scanner can guess at, by inserting new imaginary pixels in between the real ones, and guessing at what light reading should be there based on surrounding pixel readings.
Effective Resolution is not a term used by scanner makers, but by people who actually test the real resolution of the scanner. The fact that the optical resolution is 1200 spi does not mean that the optics of the scanner is able to discriminate at this level. Effective resolution can be measured by scanning a target of known resolution, magnifying the image on screen and determining visually the real discrimination.
In general the effective resolution of consumer flatbeds is only about 30-40% of the stated optical resolution. Professional flatbeds do much better, ranging from about 60% with some models to 90% or more with the very best, and drum scanners in theory resolve at close to 100% of stated optical resolution.
Sandy King
Last edited by sanking; 27-Mar-2009 at 10:41.
I don't understand what Epson means by the term "micro step drive" technology.
However, when discussing the optical resolution of scanners it is often stated that the larger number is of no importance. For example, when talking about a scanner with stated optical resolution of 4800 X 9600 spi most people say that the larger figure is not important, that only the smaller figure applies. This is, practically speaking, true in most cases, but not always. In fact, the larger figure is based on the number of samples that are taken in one direction (the direction of the movement of the CCD head), and is quite real.
If you look at the image files of the targets that I recently posted in another thread on this forum, you will see that the resolution of the bars is much greater in one direction with the EverSmart Pro scan than in the other. This is due to the fact that the scanner samples 3175 spi wide, but 8025 spi in the direction of movement of the head. As you can clearly see the resolution in the direction of the head movement is quite real, and is not interpolated. So, If effective resolution based is on both horizontal and vertical bars the resolution is about 60 l/mm, but based just on the vertical bars it is over 90 l/mm.
I am going to re-post that image here so you don't have to go back to the other thread to see my point.
BTW, the 4990 also shows more resolution in the movement of the head, but not nearly as much as the EverSmart due to optical limitations.
Sandy
Last edited by sanking; 27-Mar-2009 at 10:41.
Flatbed CCD scanners (opposed to CIS scanners which have full width CMOS elements with micro lens but are not good for transparency) are using tiny sensors, very small and usually plastic lens and set of mirrors to project 8" scanline on to tiny CCD. Most of CCD's are re-branded Philips elements or similar:
http://www.kodak.com/global/en/busin...?pq-path=14426
Take Epson apartand you'll see set of mirrors that shrinks it to "flatbed" height and not table high machine or Imacon like tower. As almost everything related to photography depends on good glass - plastic piece is just not good enough.
That said, most of our published pics are scanned on mediocre flatbeds.
Jeremy, I know I'm often on my own with the opinion that it is worthwhile to scan at the higher resolutions, but the difference is very apparent in my sample 20x24 prints from a 35mm transparency, especially when final sharpening is applied.
I was just sharing the comparison prints made from my old 4870 scans at 1200, 2400, 4800 and 9600spi with a friend today, in fact.
His observation, "Wow, sharper, sharper, sharper, then what happened to that? Its so soft in comparison to the others, like it just falls off a cliff." which was in reference to the 9600spi sample, which on the 4870 is an interpolated setting.
Have to laugh, Sandy. My astigmatism makes lines going both directions look the same. A great illustration, regardless.
I've been scanning 4X5 B/W on my V700 with Ed Hamrick's Vuescan. I have it set to scan at the max 6400 but a setting in "Output" downsamples by whatever factor you choose. I have mine set to downsample to 3200. (That extra 1,000 or so dpi above 2,400 is probably just wishful thinking...) Ed makes a pretty good case for this approach on his website. In any case, it's a solid piece of software. I'm getting better results than I did w/ Silverfast a few years ago and saving a huge bundle every year by not having to pay for upgrades.
Jeremy, to follow up my post on the incremental improvements by boosting the scanning resolution, I've attached some recent tests. These were all scanned on Leigh Perry's collaborative scan comparison transparency with Silverfast, 48bit HDR, wet-mounted on 3/16" plex directly on the platten. The file on the left was scanned at 2400 and left that way, the others were downsampled from their respective resolutions, and all three files were sharpened 400-1.0.
I usually am scanning at 6400spi. The file sizes at 6400spi are manageable (I can scan a 4x5 in two passes). I did find that the detail continues to improve with the 9800spi and 12800spi settings, but the file sizes are unwieldy.
Keith,
Don't you have any problem with the fluid making its way over the bed glass and into the inside of the scanner with this method of fluid mounting?
Sandy King
Posting Permissions
Reply With Quote
Bookmarks