USAF-1951 drum scanned?

[SURF]

While I've seen a lot of scans of the USAF-1951 test target from the flatbeds, I'm yet to see any drum scanned. Is there any?

PS. Of cause I'm ready to read that I do not need it and final print is the only thing that counts. :)

Best wishes
Al

[Leigh]

High-quality resolution targets are usually chrome on glass.

They don't bend around a drum. :eek:

- Leigh

[SURF]

I heard they produce targets on Film to. It looks you are right: it is hard to find one on a film.

Below is not USAF-1951. It is a rooler for microscope. On a glass. One step is 0.01 mm and made of white and black. So we see 200 different color steps per mm. That makes resolution 5080 dpi and it is resolved by Eversmart Pro. One can read also the numbers.

68758

[sanking]

If you have a high resolution chrome on glass target, and a small contact printing frame, you could contact print it on high resolution film. For sharpest results use a point light source at 8-10 feet from the frame.

The Eversmart Pro has theoretical optical resolution of 3175 spi X 8200 spi. The higher resolution is in the direction of the movement of the CCD. If you scan at more than 3175 spi there will be some interpolation involved.

Image quality with the Eversmart Pro is on par with a Howtek 4000/4500 scanner. The only down side of the Pro is that you are limited to 8 bit saves. The actual analog to digital conversion is made in high bit (14 bit ) but the file is saved as 8 bit. My work flow with B&W is to scan in RGB, then immediately convert the saved file to 16 bit grayscale before doing any image processing.

Sandy King

[SURF]

If you have a high resolution chrome on glass target, and a small contact printing frame, you could contact print it on high resolution film. For sharpest results use a point light source at 8-10 feet from the frame.

And I finally did it some years ago. I used imagesetter film and managed to contact copy all the fields (up to 7.6).

[Pere Casals]

And I finally did it some years ago. I used imagesetter film and managed to contact copy all the fields (up to 7.6).

This is 5 years later !

7.6 is 228 lppmm, if the drum could see 7.6 then this is more than effective 12000 dpi. Some drums were advertised at even 11000 dpi, but reaching in practice way less effective performance, perhaps 6000.

Not a drum, but X5 reaches 6500 dpi with small format film, much less as scan row is wider. So I guess that a 7.6 target should be fine...


http://www.filmscanner.info/en/HasselbladFlextightX5.html

[sanking]

One can now purchase a high resolution target on film for use on drum scanners.

http://www.silverfast.com/show/resolution-target/en.html

Sandy

[Pali K]

This is a dry mount test on Heidelberg Tango.

http://www.netsoft2k.com/Docs/Media/Pictures/Scans/Tango/Tango%20Res%20Test%20NO%20USM.jpg

[Pere Casals]

This is a dry mount test on Heidelberg Tango.

It looks 6.5 element, so 101.6 lp/mm, 5161 dpi if using "filmscanner" criterion... http://www.filmscanner.info/es/Aufloesung.html

If we consider 6.4 then it is 4598dpi...

[Pali K]

I agree Pere. Tango is around 4500-5500 DPI depending on the condition of the optical path. My Scanmate 11000 which is rated at 11000 DPI also is around 8000 DPI based on a similar test.

[SURF]

The way the universe works is admirable. Ask a question and wait some time. :-)


One can now purchase a high resolution target on film for use on drum scanners.

http://www.silverfast.com/show/resolution-target/en.html

Sandy
I'm always out of phase. The hardest thing was to inspect what was copied: those 7.6 lines are so tiny.


This is 5 years later !

We are not in a hurry here.


This is a dry mount test on Heidelberg Tango.

http://www.netsoft2k.com/Docs/Media/Pictures/Scans/Tango/Tango%20Res%20Test%20NO%20USM.jpg
Thanks! I see 6.5 and it gracefully fades after that. You said Scanmate 11000 too? Which one do you like better to work with?

[Pere Casals]

We are not in a hurry here.


It was not a complain :) just I was awesomed by your commitment, sometimes one wants to test something and it is not easy at one time, doing it long time later just shows how a challenge can persist through the time. This is seen in scientific works that made great leaps forward.



I agree Pere. Tango is around 4500-5500 DPI depending on the condition of the optical path. My Scanmate 11000 which is rated at 11000 DPI also is around 8000 DPI based on a similar test.

Effective 8000 is impressive... Some people were complaining on Nikon 9000 because at around 4000 they had aliasing with grain size, perhaps at 11000 dpi (hardware) some grains can be shaped to preserve grain (or color cloud) structure of the film. I love grain structure... this is a resource (obviously) less used in LF.

Fortunately present computers are more able to work with mastodontic files, but it is important to know when it can be worth...

Anyway the drum best point is PMT single spot reading !

[SergeyT]

This is a dry mount test on Heidelberg Tango.

Pali,
Did you try to scan at the same max scale (3000%?) but different apertures with your Tango?
New Color does not use apertures in a linear way.
For instance at 1300% it uses #2, while at 3000% - #4
SergeyT

[Pali K]

Pali,
Did you try to scan at the same max scale (3000%?) but different apertures with your Tango?
New Color does not use apertures in a linear way.
For instance at 1300% it uses #2, while at 3000% - #4
SergeyT

Hi Sergey, I just scanned it at 11000 DPI and set the Aperture to -5 which forces the smaller aperture. Also, this is was scanned in Negative mode with positive profile which tricks NewColor to turn off all USM so this is purely from the scanner. I did confirm with Karl Hudson that my scanner was resolving what is considered normal DPI from a Tango. I am happy to scan again if you think scanning it differently may improve the performance.

Pali

[sdzsdz]

I also own a Tango and have owned a Scanmate 11000. I was extremely unhappy with the Scanmates dmax performance. I sent it to Danmark and they told me its fine what costed me a couple of hundred bucks. Since two years I own a Tango and I love it. Yes, the Scanmate was better in terms of resolution but in every other aspect the Tango literally blows away the Scanmate. The biggest difference is Dmax and the tonal separation of the files. Files from the Tango contain so much colour and shadow information, its like a dream.

[Pali K]

I also own a Tango and have owned a Scanmate 11000. I was extremely unhappy with the Scanmates dmax performance. I sent it to Danmark and they told me its fine what costed me a couple of hundred bucks. Since two years I own a Tango and I love it. Yes, the Scanmate was better in terms of resolution but in every other aspect the Tango literally blows away the Scanmate. The biggest difference is Dmax and the tonal separation of the files. Files from the Tango contain so much colour and shadow information, its like a dream.

Sebastian, I agree that the Tango is definitely the holy grail of all scanners. Tango has the 12 BIT Log AD converters which I believe gives a little bit more information than what a 16 BIT linear file can store. Scanmate 11000 has 14 BIT linear AD converters so right off the bat, it's not going to be able to match the Tango in all circumstances and the difference really shows when you are working with slide film.

Love your work and Tango scans BTW. Definitely was a big reason in my own decision to eventually go with a Tango.

[SURF]

Digital camera capture of a contact copy (glass target to film):
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172251&d=1511526657

Digital camera capture of a glass target:
http://www.largeformatphotography.info/forum/attachment.php?attachmentid=172252&d=1511527525

[Pali K]

Thanks for sharing these SURF. Very clean captures.

Pali

[SergeyT]

Hi Pali,

I don't know how New Color (NC) reacts on changing the aperture to -5 from the "standard" one, but wonder if it would make a difference scanning that center (4-7) part of the target at Scale = 3000% and apertures set at 0, -1 and -2.
I don't really care if the sharpening is all the way off or not, for as long as the resulting image looks good or details are resolved better.
The NC documentation says that the sharpening is done in NC(unlike the LinoColor), but then why would they move the reduced negative mode sharpening into the Special Scanner Functions section in the NC? Is part of it stiol done in the Scanner hardware then? And if yes, does it help to resolve more details by well defining the edges? There are many questions about Tango and NC that could be answered by experimenting.

Thank you,
SergeyT

[timparkin]

Hi Pali,

I don't know how New Color (NC) reacts on changing the aperture to -5 from the "standard" one, but wonder if it would make a difference scanning that center (4-7) part of the target at Scale = 3000% and apertures set at 0, -1 and -2.
I don't really care if the sharpening is all the way off or not, for as long as the resulting image looks good or details are resolved better.
The NC documentation says that the sharpening is done in NC(unlike the LinoColor), but then why would they move the reduced negative mode sharpening into the Special Scanner Functions section in the NC? Is part of it stiol done in the Scanner hardware then? And if yes, does it help to resolve more details by well defining the edges? There are many questions about Tango and NC that could be answered by experimenting.

Thank you,
SergeyT

I tested with a whole range of apertures and resolutions and I reckon the best resolution of my system is just under 5500dpi. I compared it with my Screen Cezanne which resolved about the same but with a little less softening (but the Cezanne had a whole host of other problems). I made a few test targets on film as part of my "Big camer comparison" and came up with a 5400dpi so am happy the Tango lives around that figure.

Interestingly, I worked with an owner of an Aztek Premiere and we tested resolution vs aperture size and as the aperture got smaller, resolution got better until 10 micron and then the grain/noise started hiding detail. Supposedly the Tango/Primescan is 10 micron aperture which would make sense in my tests. The Aztek Premiere seemed to manage about 5500-6000dpi in the comparisons we made.

Tim

[Pere Casals]

resolution got better until 10 micron and then the grain/noise started hiding detail.

Interesting that 10 micron value... this is just 50lp/mm spatial frequence, and this also is the top resolving capability of sharp films like TMX in low contrast conditions, so it arises the question that it could be due to aliasing with grain size of the Tabular layer for BW. (Also there is the low speed cubic layer under that resolves much more, 200, I think).

It would be interesting to know if optimal aperture for CMS 20 it is the same or if 3um is better for it.

Another interesting question it would be knowing best aperture for color negative films that have larger color clouds (reengineering made to perform well in those digital minilabs of the Frontier era) compared to Velvia/Provia.

Just speculating if a matching grain/cloud size vs aperture can worsen things...

[SergeyT]

Thank you Tim.

According to the Tango docs it has a variety of apertures, with automatically selectable by NC ranging from 185.5 micron (#20) down to 11.7 (#2). Apparently the 10 micron would be the #1 and it is selectable only when the Aperture values -X is dialed-in in the NC UI.

SergeyT

[interneg]

Interesting that 10 micron value... this is just 50lp/mm spatial frequence

It's also the point at which a heck of a lot of 1990s transparencies hit 30% MTF response (including all the Kodachromes & Agfa Scala!) & the point that Kodak Tri-X of that era & a whole lot of C41 films hit 50% response. 30% response for transparency & 50% for negatives at 50 lp/mm seems to have been regarded as the threshold for a film to be perceived as having first rate 'sharpness'.

From the 80s/90s onwards, a more holistic approach based around the MTF behaviour of films seems to have been of greater research interest than notions of ever-finer grain size (already achieved once controlled crystal growth & tabular grains were a reality) or high contrast resolution tests & I don't doubt that Linotype/Hell & subsequently Heidelberg were very much up-to-date on the imaging research in this direction & that most of the optical decision making was driven by MTF theory.

[Pere Casals]

most of the optical decision making was driven by MTF theory.

OK, but perhaps at one point optimization was made to get good MTF response of the scanned image, rather than "on negative" to obtain an optimal optic print.

This is what I concluded, just my opinion:

When digital minilabs were everywhere they had clear advantage aganist optical minilabs. Digital minilabs allowed easy image enhacement, but C-41 films had to be reengineered to deliver larger clouds to be easy to scan and to prevent cloud aliasing. (softer clouds but "solved" with some digital sharpening)

Digitalization of minilabs was big business, a common machine was $250000 and they sold that at world scale...

Chromogenic BW (BW400CN, XP2...) also was ideal for digital minilabs.

Perhaps because that some films changed to be scanned better, this is Portra, Ektar, Fuji 160, chromogenic BW and all consumer films, while BW and slides remained without those changes.

It may be because this that today BW and slides are more difficult to scan than Porta...

Another film kind that may have moved to larger clouds is stock Vision 3, as they have to deliver a DCI.

[interneg]

When digital minilabs were everywhere they had clear advantage aganist optical minilabs. Digital minilabs allowed easy image enhacement, but C-41 films had to be reengineered to deliver larger clouds to be easy to scan and to prevent cloud aliasing. (softer clouds but "solved" with some digital sharpening)

...Another film kind that may have moved to larger clouds is stock Vision 3, as they have to deliver a DCI.

I think you've got this backwards - bigger clouds equals more aliasing, not less. The solution is age old - finer grain, smaller dye clouds! The variable aperture on some drum scanners is intended to allow a reduction of aliasing by increasing the aperture size to match the clouds/ grain.

From an entirely subjective perspective, having scanned quite a lot of various generations of the 160 speed Kodak C41 films in the last few years from 90s Vericolor to current Portra160 (before anyone asks, all generally was processed close to time of exposure & well stored subsequently - mostly 120 & scanned at the same resolution & without sharpening of any sort) I can offer a few observations. Sharpness-wise, the current Portra 160 is definitely the most immediately perceivably 'sharp' out of the box - as the MTF chart suggests. Resolution-wise, from Vericolor it increases somewhat through the VC/NC Portra era - the NC3 I've scanned is perhaps a fraction higher resolving than the current Portra 160. Grain/ dye cloud aliasing again is worst in the Vericolor, drastically better in the Portras & the current generation seems about the same as the last of the NC's.

The one change in the current generation of Vision3 & C41 films that does seem different to earlier generations is the red component having an MTF response closer to transparency films (to emphasise 'kodak' reds & general colour scheme?). Contra to the received wisdom about ease of scanning, BW & transparencies are easier, even now - as long as you have a scanner with competent Dmax & reasonable manual controls - colour neg is relatively demanding of the operator's abilities to get correct colour. That said, photo engineers generally say that colour neg delivers more accurate colour than transparencies - and if you get Fuji 160s/NS to invert correctly, you see what they mean - it can be boringly, almost digitally 'accurate'!

The current generation of Vision3 50D seems a bit of a joker in the pack with an MTF that seems almost transparency-like - the others seem more 'normal' in their behaviour - unlike Vision2 which attempted to seemingly offer a different film for every possible application. All this is quite astonishing, considering that before the mid-late 1980s, there was pretty much one colour negative film for professional cinema work, latterly rated at 100T, & pushable by a stop or so...

It's also important to remember that ECN-2 films are designed to print to another film, not paper & that DI stages (ie scanning etc) have been commonplace since the 1990s, so claims about Vision3 being optimised for scanning are moot. Indeed, Kodak made negative films specifically for scanning at several points in the last couple of decades - & withdrew them because they weren't popular enough!

[Pere Casals]

I think you've got this backwards - bigger clouds equals more aliasing, not less.

You share a very interesting experience.

Anyway I'd like to discuss the statement I quoted...

My view is the counter: larger clouds equals to less aliasing. IMHO the benefit of larger clouds is cloud overlaping.

This is my reasoning: If you have an smooth subject (like skin or blue sky) you can have two situations:

1) Non overlaping small (and more color intense) clouds, then you will have aliasing with small aperture at the drum (or at the nikon 9000) because you can read on a cloud or between clouds. This delivers an undesirable grainny look.

2) Larger overlaping clouds, aliasing problem related to cloud size disapears, you can scan with higher dpi sensor/optics or smaller drum aperture with lower aliasing.


So what I introduce in the debate is what happens from the way clouds overlap...

If clouds are not overlaping we have two kind of noises: luminance noise and chroma noise. If clouds from different layers do not overlap well then the scanner may read a lot of color noise, as color signal also aliases.

This is just an opinion...

You are right, if we view the MTF graphs from Velvia 100 and Fuji 160 we see comparable graphs... but the shape of the graph is quite different, much more curved in the 160 case, it is an interesting difference...

[Twyse]

While I've seen a lot of scans of the USAF-1951 test target from the flatbeds, I'm yet to see any drum scanned. Is there any?


Was just doing a search for the USAF target and came across this forum.....since I've recently jumped back into 4x5, thought I might as well sign up.

I have a Screen 1045ai drum scanner in my basement office and recently purchased the Lasersoft version of the USAF 1951 film target from B&H Photo. My initial results are rather dissapointing...no where near the 8000ppi max resolution of the scanner, closer to half that. I'm trying some other techniques such as manual focus on the target and adding a bit of scanner sharpening.....but I have my suspicion that the target itself might be lacking.

Has anyone ever put the Lasersoft film target under a microscope and actually VERIFIED that the target can actually resolve up to the stated 11,000+ppi? If you have, what power microscope would I need to assess the target I purchased?

The reason I'm suspicious is that a few years ago I was scanning some 4x5 CMS 20 film for a friend of mine and we did a test to see what resolution would be the maximum needed. In our test of his girlfriend wearing a textured sweater, we observed subtle detail improvements all the way to the scanner's maximum of 8000ppi. I also question whether a low-contrast target might reveal more resoluton than the ultra-high contrast of the USAF target.

Regards,
Terry

[calebarchie]

AFAIK the 1045ai has a maximum interpolated resolution of 8000dpi but optical resolution is closer to 4000dpi, doubt there are problems with the chart especially if new (they are supposed to be certified).

[Pere Casals]

Lasersoft version of the USAF 1951 film target

You can bet that the Laser Soft target is good. But check it.



VERIFIED that the target can actually resolve up to the stated 11,000+ppi? If you have, what power microscope would I need to assess the target I purchased?

10,000 ppi is 400 p/mm so 200 lp/mm, at 400x you should observe it with total confort.


you may get something new like the $334 AmScope T490B at Amazon, or something better, or you may even use a cheap hand microscope

203820

You may also purchase a used Pro Leitz or Nikon microscope.


You may also do that: use an enlarger with a good lens like the Nikkor-EL-50 and project the slide on the easel, then place on the easel a DSLR without the lens, projecting the slide on the sensor, see this:

https://www.largeformatphotography.info/forum/showthread.php?151265-9000-dpi-effective-DSLR-scanner

It shows that Group 7 can be perfectly check in that way.

Now I've a good 1951 G9 slide, but first I purchased a very cheap defective one (in the higher elements), I wanted only G5 to rate a V850 scanner, this is a shot of that defective slide projected with the enlarger:

https://www.largeformatphotography.info/forum/attachment.php?attachmentid=189049&d=1553121347





I also question whether a low-contrast target might reveal more resoluton than the ultra-high contrast of the USAF target.


It would revel the same, or less, depending on your sight.

[Twyse]

AFAIK the 1045ai has a maximum interpolated resolution of 8000dpi but optical resolution is closer to 4000dpi, doubt there are problems with the chart especially if new (they are supposed to be certified).

Where did you get your information that 8000 ppi is interpolated?

[Pere Casals]

Where did you get your information that 8000 ppi is interpolated?

https://www.filmscanner.info/en/FilmscannerKlassen.html

See the "Drum scanner with photo multiplier technique" section. They say that the 1045ai can resolve around 6000dpi effective, but you have to scan at 12,000dpi nominal.

If you scan at 8000 then you get 4000 effective, which is an overkill anyway, usually, at least for LF or for regular films and lenses.

[calebarchie]

Where did you get your information that 8000 ppi is interpolated?

My bad I got mixed up with the 1015ai, looks like your scanner may need a thorough service.

[brbo]

Howtek 4500 (you need to click on images to see full res scans)...

4000dpi, 6m aperture:

https://live.staticflickr.com/65535/51488689362_f4a8780de1_c.jpg (https://live.staticflickr.com/65535/51488689362_d528724bc8_o.jpg)

4000dpi, 13m aperture:

https://live.staticflickr.com/65535/51488688362_edbd04beeb_c.jpg (https://live.staticflickr.com/65535/51488688362_ec30c6e423_o.jpg)


2000dpi, 13m aperture:

https://live.staticflickr.com/65535/51489475751_914c7b5b4f_c.jpg (https://live.staticflickr.com/65535/51489475751_36427b35f0_o.jpg)

You can clearly see the benefits of oversampling in the 4000dpi vs 2000dpi with 13m aperture. 13m aperture should, in theory, only allow 2000dpi, but since the readings at 4000dpi setting are taken at twice the rate of the 2000dpi you can get higher actual resolution. I guess that is how the Tangos get their smooth and yet detailed scans, deliberately selecting larger than optimal aperture.