USAF 1951 test chart in SVG format

[ramon]

Where can I find a scalable vector graphic (SVG file) of USAF 1951 Resolving power target?

Dimensions of USAF 1951 are here: http://www.efg2.com/Lab/ImageProcessing/TestTargets/

SVG software is free (inkscape). Someone with the knowledge (not me) can made this file.

Does anyone here knows how to make this file?

[domaz]

You can get a PostScript version (http://www.photo.net/learn/optics/USAF1951.ps.) of the file from photo.net. That is a vector format- not SVG but still vector. There are many ways to convert PS to SVG, you can Google that one fairly easily.

[Jim Michael]

pm your email address and I'll send you an svg conversion I made from the ps file.

[ramon]

Thank you both, I have found an autocad (DXF,DWG) version: http://www.vinland.com/USAF-1951.html

Now I am searching a high resolution laser engraving service.

(Edmund Optics asks between $120 and $1,250 for a glass target. I Want to know if I can save some cents.)

[Emmanuel BIGLER]

One USAf test target in pdf to download from Japan.

http://www.takinami.com/yoshihiko/photo/lens_test/USAF.pdf
http://www.takinami.com/yoshihiko/photo/lens_test/pdml-procedure_c.html

[Emmanuel BIGLER]

A copy of the test target in pdf (http://www.cijoint.fr/cjlink.php?file=cj201106/cijDHk5XA3.pdf)

Explanations, meaning of group and element numbers in terms of cycles per millimetre (http://www.cijoint.fr/cjlink.php?file=cj201106/cijKOYTjRI.pdf)

[ramon]

Hello Emmanuel,

Thanks, but I don't want to test a lens. I want to test a scanner.

So I do not want a enlarged USAF chart, but instead a reduced chart to test the optical resolution of the scanner.

I contacted with one laser engraving service and the told me that they have a maximum resolution of 1200 dpi (theoretical given by manufacturer of laser engraving system). This is not enough.

I think of two options:

1) print that pdf file and make a reduction with some macro lenses on E6 or BW film.
2) find a printer with high resolution and print on transparency (polyester).

Thanks

[SCHWARZZEIT]

Hi Ramon,

I used the scanner target made by Ivan Danes (Danes Picta (http://www.danes-picta.com/)) to test my scanner. Just look under pre-press. The highest resolution pattern on that target is 250 lp/mm. The price was €67 in 2008.

Keep in mind that testing a scanner with a high contrast target can be used as a reference for comparison but it does not necessarily reflect the results of scanning photographic film. On film the high frequency contrast is much lower than on a test target that is a lithographic contact print. For real world results it may be more interesting to see how your scanner affects your system resolution.

-Dominique

[SteveH]

Thank you both, I have found an autocad (DXF,DWG) version: http://www.vinland.com/USAF-1951.html

Now I am searching a high resolution laser engraving service.

(Edmund Optics asks between $120 and $1,250 for a glass target. I Want to know if I can save some cents.)

Ramon
Silverfast software sells a USAF target that is cheaper but I don't know much about it, I think its 35mm but may still be useful.

Let us know if you find a good option as I would also like a decent scanner test chart..

Steve

[ramon]

Good news !

I have found a test resolution target in the last frame of a microfiche ($3 USD at the auction site)

http://213.251.184.113/largeformatphotography/microfilm_RESOLUTION_TARGET.jpg

With my scanner's (Epson V700) resolution is near 1.4.

How can I translate this number to dpi?

Thanks

[Jim Jones]

Ramon -- Google for NBS 1010a. There seems to be lots of information, although much of it will require interpretation.

[Emmanuel BIGLER]

How can I translate this number to dpi?

Simply check on the actual target what is the actual dimension labeled '150 mm' ; but you can derive it as explaines below.

It might be a true 150 mm ( I hope _not_ ! since it would mean that your scanners resolves only 1.4 cycle / mm ;) )or anything else, this is not very important.

In the jpeg image you've posted, '150 mm' corresponds to 469 pixels, 469 samples. For example, if the image was sampled at 2400 samples per inch = 94.5 samples per mm, the '150 mm' dimension is actually 469/94.5 mm = about 5 mm. So if "150 mm" are in fact 5 mm, one mm is in fact demagnified by 150/5 = 30x.
When you have the actual dimension of this '150 mm' size, you can easily derive what 1.4 means.
The white bars labeled '1.0' feature 1 cycle / mm when the '150 mm' bar measures actually 150 mm. In the image they correspond to 469/150 = 3.1 pixel for one period.
1 cycle/mm means that the (bar+space) size is 1 mm
the 1.4 pattern corresponds to 1.4 cy/mm when the '1.0' pattern corresponds to 1 cy/mm.
in the image 3.1/1.4 yields 2 pixels per perdiod, i.e. the ultimate sampling limit for any digital sampling of a periodic object.
If you know the theoretical number of samples per inch in your scanner, you could easily derive the actual limit.
If this targer has been scanned at 2400 samples per inch, it means 94.5 samples per mm. the optical limit, @2 samples per period, being 94.5/2 = 47.2 cycles/mm, about 1.4 x 30 where "30" is the de-magnification ratio in use when the microfiche was made.

Summary in 3 lines
- find the actual size of the '150 mm' bar and derive the de-magnification ratio of the microfiche
- multiply the number : 1.0, 1.25, 1.4 etc that you read on the target by this ratio, e.g. 30, and you -
get the resolution limit in cycles/mm.
- to convert cycles/mm into samples per inch : 1 cycle/mm = 25.4 cycles per inch ; 2 samples are needed per cycle. hence 47.2 cycles/mm are 47.2 x 25.4 x 2 samples per inch = about 2400.

[ramon]

Hello Emmanuel,

Thanks for your detailed explanation.

The image I previously posted was scanned at 12800 dpi.

But size was also scaled to keep the file under 1 Mb, so pixels do not correspond with original size.

I have made some numbers and your calculus is very close to the actual enlargement (32.4x).

I obtained that the resolution of my scanner is near 2308 dpi.

This also proves why I cannot get any improvement scanning at more that 2400 dpi.

Scanner interpolation does a good job, but this also could be done postprocessing the image.

Next step I want to do to improve image is trying to find the best focus position.
I want to attach the film test target to a flat acryllic sheet rised in each corner with nylon washers.

Thanks

[Emmanuel BIGLER]

Hello Ramon
I'm was not sure that my explanations were clear, so I'm glad that you could work out the experiment.
Actually resolving above 2400 ppp (1200 cycles per inch) is something not so easy to reach for amateur-grade affordable flatbed scanners, but using a microfiche with a test target is a neat idea that will not cost and arm and leg !

[Leigh]

Please understand that the test target was already degraded by the process used to create the microfiche.

An accurate evaluation of the scanner would require scanning an original target, preferably chrome on glass.

- Leigh

[ic-racer]

Good news !

I have found a test resolution target in the last frame of a microfiche ($3 USD at the auction site)

http://213.251.184.113/largeformatphotography/microfilm_RESOLUTION_TARGET.jpg

With my scanner's (Epson V700) resolution is near 1.4.

How can I translate this number to dpi?

Thanks

?
Looks like you have someone else's resolution test. You need the chart they photographed with the microfilm, not the microfilm.

[Bill Burk]

Please understand that the test target was already degraded by the process used to create the microfiche.

An accurate evaluation of the scanner would require scanning an original target, preferably chrome on glass.

- Leigh

Yes, the target is already degraded once it is photographed (as this example photograph on microfiche). And the numbers can probably never be truly meaningful. But as a low-cost target to find the approximate resolution of a system, this seems like a neat trick.

There probably aren't many film targets that are higher resolution or finer grain than microfiche.

[Leigh]

Hi Bill,

Certainly, as a quick check, such as for focus, and an inexpensive target, the microfiche would be quite satisfactory.

There are much finer targets available, but they're expensive.

- Leigh

[Struan Gray]

One option for hi-res, high-contrast targets would be a semiconductor lithography mask. They tend to be chrome on glass (quartz if you pay more) and 5" square is a standard size that would work well with most photographic applications. They are robust enough to be used as a master for contact printing or placing on a scanner platten directly.

There are online companies who will allow you to upload a file and they will send you the mask by mail soon thereafter. But you'll need to use compatible design software to create the design file, and they won't hand hold you much, if at all. But mask-making is often an in-house service at university engineering departments and high-tech startup incubators and you may be able to find someone local who will help you with the design and produce the actual mask for a cheaper price. It's easy to over-specify things like the line quality and end up paying too much, so, again, it's good if you can find a real person you can talk to - you'll get what you want, rather than what you ask for.

I used to get 5" masks for around €100, but that was an internal price and I did the design work. It's a commodity job though, so commercial suppliers were not much more expensive.

[Jim Jones]

I've used an ultramicrofiche Bible on a 2" square chip https://patriotpostshop.com/products/166 as inexpensive target for quick projection testing of 35mm camera lenses. The detail is much finer than a consumer flatbed can resolve. These chips can be placed on a flatbed scanner with one edge raised a few mm to determine optimum height of film.

[ramon]

One option for hi-res, high-contrast targets would be a semiconductor lithography mask. They tend to be chrome on glass (quartz if you pay more)

(...)

I used to get 5" masks for around €100, but that was an internal price and I did the design work. It's a commodity job though, so commercial suppliers were not much more expensive.

Hi Struan,

Please can you share with me some commercial photomask suppliers?

I already have the design files, provided by Bertho Boman. And also I have access to CAD tools and can create custom targets (for 5" squares).

I am living in Taiwan. Six months ago I found a photomask house that asked $150 for a 5". But resolution was very limited (10 microns) so I didn't ordered. (10 microns is only around 50 LPPM).

I know of commercial USAF 1951 charts in glass (2"x2") at $120 USD with high resolution 228 lppm. But I'd would like to make some 5"x5" target to test some LF lenses at 1:1.

[Nathan Potter]

I'll chime in here in response to Struans' comments. There are a lot of chrome mask shops around the world. Many do state of the art masks which you do not want or need - and they would be too expensive anyhow. I think the least expensive for you that would be adequate is a Pattern Generated image on chrome on soda lime glass. Typically these are .125 or .150 inch thick for a 5 by 5 inch size plate.

The resolution obtainable from an optical pattern generator can be as fine as about 2 µm features (250 lp/mm). That is easily enough to serve your needs. If you can stand 5 µm minimum linewidth you can reduce the price a bit more.

These type of masks are made by exposing a thin layer of photoresist to the light from the scanning pattern generator. The resist is usually a photosensitive novolak resin, very thin, over the chrome layer, and is then developed in a strong base solution to form the pattern in the resist. Then the chrome is etched in an acid bath to form the image. Much finer patterns can be made by electron beam exposure using the appropriate short wavelength radiation.

The common file needed for the pattern generator is GDSII. But a number of other file formats can be converted to GDSII. If you supply other than GDSII files I would recommend you get a blowback file from the manufacturer to assure that the file conversion was not corrupted.

Three mask houses I use:
www.compugraphics.com in Los Gatos CA.
www.photomask.com in Austin TX.
www.advancerepro.com in Andover MA.

Advancerepro (Advanced Reproductions) often has the better price but if you want 2 µm feature size I think you will be looking at $US 150 to 200 or so.

It would be wise to design a repetative pattern to cover the whole mask (or say 4.5 X 4.5 inch of pattern). Then you could assess plane of best focus using the wedge technique.

Check out the above mask facilities on Google.

Nate Potter, Austin TX.

[Struan Gray]

Hi Ramon

I myself never used a commercial house, but I know several who have ordered from www.photomask.com and were happy. I always got my masks from an in-house facility, or from the semi-commercial clean room labs attached to several of the scandinavian research universities.

I left the university a year or so ago, but at that point the in-house team were gently pushing routine users like myself to use a commercial supplier, since prices had dropped to the point that it was no longer worth their staff's time making our kind of stuff. I will contact my friends on the clean room lab staff and find out who they use for purely commercial orders - I'm pretty sure it was an outfit in Germany, which would save you VAT/customs hassles and postage.


Nate: thanks for the info. I'm sure there are others like yourself with far more experience making and using masks than I have, perhaps they'll chime in too. We always seemed to be able to get masks on a friend-of-a-friend basis so I never had to take the purely commercial route.

[Nathan Potter]

Ramon, the go to mask house in Taiwan is the Taiwan Mask Corp. I don't know their mask making equipment capability but is likely near the state of the art for the advanced Taiwan semiconductor fabs.
You need someone with relatively antiquated equipment (optical pattern generators) to keep the costs down. In the USA there are a few left who find enough business with startup companies and universities that they can maintain their business.

BTW I have occasionally seen lithography masks for sale on ebay but of course they will generally not be suitable for or designed for resolution tests.

Also note that it is not trivial to obtain useful resolution data using a digitizing device on an analog mask structure.

Nate Potter, Austin TX.

[ramon]

Thanks a lot to all of you for your help. Maybe I will resume the idea of ordering a photomask.

Here is some initial drawing I made six months ago. Don't know if anyone else has interest in making some design, modification, contribution, or maybe want also to order a mask.

http://213.251.184.113/largeformatphotography/USAF-1951_4x4_Inches.dwg

http://213.251.184.113/largeformatphotography/USAF-1951_4x4_Inches.pdf

This drawing contains group numbers -2, -1, 0, 1, 2, 3, 4, 5, 6, 7 and 8.
So it resolves from 0.25 lpmm to 456 lpmm (1.1 micron size).

It needs to modified. Because text witdh is around 0.75 microns (very expensive). Maybe I will modify it to resolve only until group number 7 (2 microns, cheaper) or delete text on smallest group.

[Emmanuel BIGLER]

http://213.251.184.113/largeformatphotography/USAF-1951_4x4_Inches.pdf

Thanks, Ramon, for sharing those files with us.
In this PDF, there is a problem, small targets are completely clogged and un-readble. Like if it had been plotted with a broad pen, wider than the fine lines themselves. But this might not be a problem since pdf is not the native format of pattern generators, may be the .DWG file is OK.

This problem does not occur in the other pdf file mentioned above,
http://www.takinami.com/yoshihiko/photo/lens_test/USAF.pdf
where you can zoom up 6400% down to group 7 and still be able to resolve all finest groups.