Measuring film resolution

[Photo Engineer]

Many statements do not include viewing distance for the data cited.

PE

 

[Alan Johnson]

Alan, this formula works for "I get 50% of max contrast", but not for "I can barely see the lines" limit. Given that Erwin's numbers are far below the numbers given by Zeiss, I have reason to believe that he used the "I get 50% of max contrast" limit.

Rudi,
The formula works if you use a contrast of 20-30% for the film (lens=250 aerial resolution).See the MTF curve for TMX here:
http://www.kodak.com/global/en/professional/support/techPubs/f4016/f4016.pdf
IDK what contrast one typically uses looking down a microscope and judging lines, perhaps it is 20-30%.
Sorry, my mistake, looked at TMY curve.You are right, for TMX it is nearer 50%.

 

[Rudeofus]

Alan, 20-30% contrast in that data sheet does not mean, that the dark lines are 20-30% denser than the gaps in between. As your line pattern starts at very low frequencies, you get a certain contrast (2:1, 10:1, 100:1, whatever) and define this to be 100%. As your line patterns go up in spatial frequency, your image contrast at first increases slightly (say hello to Mackie lines), and above some threshold frequency it starts to drop. If contrast drops down to 30% of what you had at spatial frequency 0, that's where you draw the line at 30%. If you started with a 100:1 brightness ratio on your medium, contrast has dropped to 30:1 at this 30% point. That's still very visible!

If you optically inspect line patterns on film, you hardly care about what contrast started with - at some (usually very low) contrast ratio you simply can no longer distinguish individual lines. However, we can safely assume that the brightness ratio of these line pairs must have dropped to a few percent of what it started with.

 

[Nodda Duma]

Many statements do not include viewing distance for the data cited.

PE

They might assume standard accommodation distance of 25cm. That's the only standard viewing distance I know of for testing/optical purposes.

 

[Bill Burk]

Alan, this formula works for "I get 50% of max contrast", but not for "I can barely see the lines" limit. Given that Erwin's numbers are far below the numbers given by Zeiss, I have reason to believe that he used the "I get 50% of max contrast" limit.

I like the way MTF charts take away that ambiguity by showing the function as a curve. It tells a more useful story than a single number.

And this is just a lame attempt at humor, Zeiss tells how to perform such tests and I guess Erwin Puts probably didn't use the Contax RTS with its vacuum back film plane, along with a host of other Zeiss-recommended factors that may have limited critical definition in his tests (unlocked fluid pan head tripod with hands holding camera to absorb shutter slap, mirror lockup etc)... And if that isn't a high enough bar, the lenses Zeiss suggests you should print with are beyond reach of everybody here.

 

[Rudeofus]

Just to put this back into perspective: For pictorial matters I don't care in the least about anything beyond 100 lp/mm. Many low speed films reach this with whatever developer you throw at them. If you follow Henning Serger's numbers, most color films reach above 100 lp/mm easily, too.

But that's not the point. I try to make a resolution target, and if I want to reliably measure 80 lp/mm, that target better be 150+ lp/mm. That resolution target will be contact printed, so enlarger lenses or film flatness issues are irrelevant. Since I don't think that 100 lp/mm are possible with Delta 3200, and since Delta 3200 is the film that most desperately asks for a fine tuned developer, I think I am on the right way here.

 

[Rudeofus]

Why do you say that? Just curious and following along.

The availability of extremely fine grained film and moderately priced medium and large format equipment have pushed the enlargement limits so far up that few will ever hit the grain limit. With pushed HP-5+/Tri-X/Delta 3200 and small format I have had the grain limit hit me hard, so that's what I would look at first. People who have troubles with grainy TMAX 100 do something wrong IMHO, and an improved developer won't be able to help them.

 

[Alan Johnson]

I try to make a resolution target, and if I want to reliably measure 80 lp/mm, that target better be 150+ lp/mm.

Some of Henning Serger's results with microfilm types suggest this may be possible with a good lens, even if not the macro Planar 2/50 that he used.
(there was a url link here which no longer exists)
The resolution of Fuji Velvia from the manufacturers data appears less than you require, it reaches 160 lppm at chart contrast 1000:1 (p7):
http://www.fujifilm.com/products/professional_films/pdf/velvia_50_datasheet.pdf

 

[Rudeofus]

The resolution of Fuji Velvia from the manufacturers data appears less than you require, it reaches 160 lppm at chart contrast 1000:1 (p7):
http://www.fujifilm.com/products/professional_films/pdf/velvia_50_datasheet.pdf

If I contact print my Lasersoft resolution target onto Velvia 50, then I should be able to get these 150 lp/mm and everything should be fine. As I said, the tricky part will be getting the repeated pattern with the correct spacing. I'll keep the "use a blazing sharp macro lens to photograph a printed target" method in mind.

 

[Ashfaque]

I apologise beforehand, because most likely my post is not useful and utter non-sense.

I've been messing around with some electric stuffs (for light meter). These days these PCBs can contains hundreds of lines, if not thousands in a tiny place. I noticed that you can design your own PCB with extremely fine threads. If it is possible to print a PCB with resolution target formula (of 1951 USAF Targets and any other), along with the actual colour, perhaps it would cost lot less. If I understand correctly, it would cost even less if printing more than 3-5. I have no idea how to design PCB stuffs yet though. So far, all I know are some softwares - like EAGLE 3.7, fritzing, etc., that are free for personal use.

Bests,

Ashfaque

 

[Photo Engineer]

At one time, you could buy complete PCB kits at Radio Shack. I made a few way back when and you use either a photographic process or you can draw the lines with a resistive "ink". The photo process makes far finer lines of course, but I never pushed it.

PE

 

[Rudeofus]

Ashfaque, there are two issues with this approach: first of all, PCBs are made by chemical etching, which means that traces can't be finer than they are thick. With typical 35 µm thick copper layers you have line pairs which are at least 70 µm wide, and that gives you a whopping 14 lp/mm. There are speciality circuit boards with thinner copper, but 100 lp/mm seems off limit. Second, the FR-4 material typically used for circuit boards is not transparent, and I know of no circuit board material that is. There are flex prints which are somewhat transparent, but these are usually deep orange/brown. If you've ever tried to B&W print a C-41 negative, you know that this is not what you want.

This whole PCB thing does bring up an interesting idea, though: PCBs are made lithographically, which means they start with a set of transparent B&W masks. These masks aren't overly expensive, I don't know about their limit resolution, though.

 

[Ashfaque]

Thanks you both very much. I learned some new things today.

Bests,
Ashfaque

 

[sfaber17]

I have no idea how to design PCB stuffs yet though. So far, all I know are some softwares - like EAGLE 3.7, fritzing, etc., that are free for personal use.

I would use the open source Kicad for the PCB CAD software. In the past I made PCB boards with free PADS DOS version and laser printed the layout, took a picture with 35mm Kodalith film and printed to a 4x5 Kodalith film, then contact printed that to another one to get the actual size negative used to make the board. The board was coated with a spray, dried and exposed for 2 min or so under a sun lamp then developed with xylene to remove the unexposed resist. Then used hot ferric chloride soln. to etch. Ammonium persulphate was less noxious to use, but slower and didn't quite work well enough. Of course lately I noticed Kodalith was no more, but I think there is some substitute still available from Freestyle. I don't think I'd try this again though since there are a lot of places to get prototype boards made fairly cheaply. The usual design rules require spacing and line widths of around 5-7 mils, so the resolution isn't in the range of the film tests that were discussed. I recently was designing a surface mount board that used chip capacitors. I ordered some without realizing they were something like 10 mils wide and too small to be practical with a manually populated board, so I had to re-order and make sure the design only used the larger caps.

 

[Bill Burk]

...Ammonium persulphate was less noxious to use, but slower and didn't quite work well enough...

Interesting, I have some of that here.

I got it to make up some superproportional reducer - which I will be using in an experiment to try to re-create a vintage film look.

In the process I might affect resolution... And so all these efforts to make a resolution test can help me quantify and objectively describe the amount of image destruction.

So making up a resolution chart is not a totally worthless venture.

 

[Rudeofus]

I just checked with a local PCB maker. The finest feature they can make on photo mask is around 50 µm wide (2 mil), which means 10 lp/mm if you make a resolution target with that technique, even less than I estimated previously.

Does anyone know of another process where inexpensive fine resolution masks are needed on a regular basis?

 

[Bill Burk]

I think you should be looking at chip makers photolithography, they work with much finer resolution than PCB.

But in that line of work, inexpensive is not one of the words they use in that department.

 

[Athiril]

Might be able to make a very high resolution film recorder (monochromatic) using this, I'll you know if I do it. The laser unit should be cable of sub-micron levels, which is beyond what we'd need.