IQ180 and 8x10 assessment

[timparkin]

Tim,

I was pretty much disappointed when I compared nominal dpi numbers of a scanner with actual line pairs resolved ..(and these translated into dpi's) .. high dpi numbers mean nothing unless one has run something trivial as a USAF 1951 chart test .. quoted 6400 dpi may turn out to be just 1800 dpi .. if at all.. I learned that file size is nothing .. a decent true 1800 dpi scan may be better than a quoted 6400 dpi scan with inflated file size on a different scanner... running simple lp tests like on 8x10 allow to determine system performance .. and to optimize .. or live with it.

joerg

Hi Joerg - I don't think anyone is saying that scanning at 12,000dpi means 12,000 dpi of content (look at Epson scanning resolutions for confirmation of that). I also don't think a USAF 1951 resolution target is trivial to set up or read either (and how do you read resolution figures for digital? at 90 degrees or 45 degrees, with monochromatic light, green light or red light? How do you interpret the sharpening of the Bayer array and it's effect on contrast?

[Jim Michael]

The instructions for the USAF test chart are pretty clear and have you set an array of targets in various orientations (90 and 45 degree orientations included) at a set distance dependent on focal length. You then locate the least resolvable lines and look up the resolution. Nothing difficult at all, unless you start changing the distance and size of the targets in which case some math is involved. The benefit of using the test charts is being able to quantify the results and compare to the theoretical numbers.

[Jim Michael]

Re the Bayer related issues, I think the approach there is to shoot a zone plate (sinusoidal) to determine the true resolution of the sensor.

[Brian Ellis]

Hi Joerg - I don't think anyone is saying that scanning at 12,000dpi means 12,000 dpi of content (look at Epson scanning resolutions for confirmation of that). I also don't think a USAF 1951 resolution target is trivial to set up or read either (and how do you read resolution figures for digital? at 90 degrees or 45 degrees, with monochromatic light, green light or red light? How do you interpret the sharpening of the Bayer array and it's effect on contrast?

A USAF 1951 resolution target also isn't inexpensive. Prices start around $150 and can go up to more than $1,000. The ones you can download or buy for a lot less are just copies of the target.

[Jim Michael]

A USAF 1951 resolution target also isn't inexpensive. Prices start around $150 and can go up to more than $1,000. The ones you can download or buy for a lot less are just copies of the target.

Link to a USAF 1951 chart PDF on this page. No reason the USAF chart hasn't been done in PS, it's a pretty simple construct.

Oh, and I should have an SVG around somewhere, if there's any interest.

[Nathan Potter]

It may have been in the scanner comparison thread, but someone posted some results from a well tuned V7xx series scanner that were pretty outstanding. The calibration methodology for these scans should be published with the results since it has a bearing on the quality of the output.

Count me in as a contributor.

[IMG]
EPSONcont-web-1 by hypolimnas, on Flickr[/IMG]

Jim, I think you may be referring to a plot I made recently for the V750 Pro. Plot is above from Flickr. The plot was made from the resolution target after importation into PS and with the full density range from 0 to 256. In other words my interest was in finding the MTF curve for a scan where the linearity and full 0 to 256 density range was preserved during and after the scan.

Since the resolution mask has no intermediate densities (it is either clear or opaque)
it is possible to simply clip the white and black points to provide a high contrast image that would indicate a high resolution (I could get to 2600 SPI at above 50% contrast) but that sacrifices all middle density values and is not useful for many of us; and me.

I will stress again that the LL article is almost entirely about resolution between the IQ180 with snappy modern lenses and 8X10 format with high quality LF lenses. Zubers addenda recently, in answer to criticism about the 8X10 limitations, only describes a "richness of detail" benefit in using the IQ180. Dunno what that means precisely. One mans richness is another mans fault. Still wondering how to extract quantitative data out of "richness of detail".

I understand completely why a subjective judgement is useful at the end of the comparison. One could even tabulate and rate certain attributes of prints such as clarity, contrast, color saturation, etc. Then I think there would be no consensus as to which is best - if best is what we would be after. IMHO all the examiners will bring a different perspective and persuasion to bear on the images and yield no consensus, leaving us where we are now - in argument with endless discussion.

Nate Potter, Austin TX.

[Ken Lee]

Thanks for making that curve about the Epson. It makes me want to get a drum scanner

It also shows why different people find different rates of enlargement to be acceptable: 3x, 5x, etc. They're merely stating their tolerance for low contrast, so to speak.

[Jim Michael]

No I hadn't seen that curve Nate. Thanks for do that. I'll have to go search the archives but what I was referring to may have been a scan of the reference target that was being used for the scan comparisons. There was a description of the methodology used in setting the scanner up IIRC.

[Jim Michael]

I think this is the post to which I was referring. See #32.

[John NYC]

I think this is the post to which I was referring. See #32.

Yes. Not to start this all over again (there have been many threads where we all disagree on the degree of the problem), but the advertised dpi of scanners is -- how should I say this? -- optimistic. Cough.