Digital Photography Now - www.dpnow.com  
 
advertisements

26th October 2004
Project: make your own digital camera resolution test chart
by Ian Burley
1263: Project: make your own digital camera resolution test chart

Introduction


We find out if you really can make your own professional digital camera resolution test chart for free


ISO_12233-reschart

We recently ran a news story about a Cornell University researcher, Stephen H. Westin, who created his own version of the ISO 12233 electronic camera resolution test target and made it available for free download. The ISO 12233 target is a standard benchmark tool used by a lot of magazines, including dpnow, for testing cameras. Manufacturers also use it. You can buy a high quality commercially produced ISO 12233 target from suppliers like Edmund Optics, which is where we obtained ours. It comes in two versions, the standard size, priced $159 and a large size, for no less than $875. This is beyond the reach of most who might simply want to do some experimentation at home.

There was immense interest in our news article. So the question is, can you really use the Westin target by printing it out at home on a typical ink-jet printer? We decided to find out by printing our own target using Westin's vector pdf drawing and comparing with our own commercially produced target.

Westin has created his own version of the ISO 12233 test target by referring to the official ISO 12233 specifications. His vector drawing has been made available as a vector pdf file. Being a vector drawing, it can be scaled to any size before printing, so it can make best use of the available printer resolution.

First attempt
We imported the file into Photoshop at a selected print resolution of 300ppi, the recommended target resolution for photo printing. But on-screen it was clear that this was way too low as edge-aliasing was very evident.

ISO_12233-reschart-aliased
Without anti-aliasing (300ppi)
ISO_12233-reschart-antialiased
With anti-aliasing (300ppi)

There is an anti-aliasing option when importing vector drawings into Photoshop and while this improved things, the mid-high resolution test lines remained poorly defined. There certainly wasn't any point in printing what we had produced.

Second attempt
So we decided to import the test target drawing at a much higher resolution. We tried 600ppi, 1200ppi and even 2000ppi. At the latter resolution, we almost reached the Photoshop limit of 32,000 pixels for the width of an image. Even at this resolution, with anti-aliasing on, the edges on curved detail sometimes looked less than perfectly smooth. But we could see the high resolution test lines clearly, so we decide to make a print using our Epson Stylus 2100 photo printer.

Epson Archival Matte paper was selected to avoid reflections from the lighting and this paper doesn't suffer from bronzing. Our Edmunds Optical target is a photographic print onto semi-gloss paper. The target is just less than 400mm wide, too big for A4 sheets – another reason for using the Epson 2100 as it's an A3 wide format printer. Custom settings were used to switch off bi-directional printing and to use the printer's maximum print resolution – 1440dpi in Archival Matte mode.

Boy, what a disappointment! The printer was clearly incapable of reproducing anywhere near the full detail of the image file. The higher resolution test lines were merged and other test lines were sharply serrated. Maybe this wasn't going to work after all.

Success at last?
Plan C focussed on selectively enlarging a portion of the target drawing to print it as large as possible. The portion we finally printed is shown below:

ISO_12233-reschart-portion

It occupies most of a sheet of A3 and represents a 250% increase in size, or a 625% increase in area. Even at this size, the printed detail of the chart didn't look as smooth and as clean as we had hoped, but this was really the biggest we could usefully go with our Epson 2100, so it was now time for the reckoning.

To test this morsel of a chart, we calculated the positions of the alignment points and mounted the chart towards the top right corner of our big DXO Analyzer test target, which is basically a large piece of plate glass. Once satisfied that the alignment points were correct, we used two cameras – a 5MP Casio Exilim EX-Z55 and a 3MP Olympus Mju:300 (Stylus in the US) to take test shots and then repeated the exercise with the Edmunds target.

As the two targets are very different sizes, the cameras needed to be repositioned each time. We tried to maintain exposure and zoom settings, but there are slight differences, though we managed to match the recorded sizes to within one percent of each other. And the results – well, we were rather pleased. To see them, please go to page 2.

Results and Conclusions

The first thing that's noticeable about the respective (Westin and Edmunds Optical) targets is that the text size used in the scales are different – Westin's are smaller. There are some other minor detail differences, but apart from that the two targets are essentially the same design, as you would expect from two accurate interpretations of the same ISO specification.

We have reproduced cropped selections of the results below to minimize download delay. PNG lossless image format has been used to avoid the limitations of GIF and JPEG that would compromise detail image quality. The colour differences are down to the different tungsten light balance presets in each camera.

Olympus Mju:300

Horizontal resolution
A.

Olympus-Mju310-Z55-realISOa

B.

Olympus-Mju310-Z55-replicaISOb

Can you tell which target is which? Sample B, with the smaller scale legend, is the Westin target. The three megapixel Olympus doesn't stretch the target. Both samples show that the test lines merge between 10 and 11 on the scale (1000 to 1100 lines per target height).

Casio Exilim Zoom EX-Z55

Horizontal resolution
A (Edmunds Optics)

Casio-Z55- realISOb

B (Westin)

Casio-Z55-replicaISOb

With the higher resolution Casio Z55, differences between the two targets stand out a little more, but the result is still pretty good for the Westin target. Both indicate a measurement of around 1500 lines, though the Westin target looks a bit less refined, with a harder moiré pattern towards the resolution limit.

Vertical resolution

res-vertical


Again with the Casio Z55, the Edmund Optics target is on the left and the Westin on the right. Much the same conclusion for the horizontal results applies here, though the Westin target, if anything, shows a slightly increased resolution result.

Conclusion
I doubt that an optics expert would be totally satisfied with our testing methodology or the absolute quality of the results we obtained with our 'version' of the Westin target. But we're not suggesting those that require high quality, professional level, test data replicate our method. If you want a professional result, the best advice is to buy a professionally produced target like the Edmunds Optics one.

However, we hope we have demonstrated how you can get close to professional results using an ordinary printer, that you get a feeling for the limitations of printed resolution and maybe have a bit of fun and learn a few things on the way.


 
advertisements
©2001-2014 Digital Photography Now, All Rights Reserved.