Scan2CAD raster to vector conversion software

NO BULL: Scan Speed Part 2

Contex define large format scanner productivity in "drawings per hour"

... Continued. [Contex Scan Speed - previous] [Contex Scan Speed - next]

How this test was conducted

Many question if Contex's new measure can offer any insight into genuine wide format scanning productivity. In understanding what Contex is claiming, it is useful to look at how it conducted its tests and obtained its data.

Scanners4CAD was not present when Contex conducted its tests.

Contex's tests, like all wide format scanner manufacturer's tests of their products, was conducted in-house by its own staff, not by an independent third party.

[Editor's Note: Contex's press release of April 2009 states that Buyer's Lab International (BLI) "confirm that the Contex SD4490 offers a significant speed advantage without compromising on image quality". However as no qualifying information is given (for example speed advantage over what? How was it tested?) this is a meaningless statement.]

Scanners4CAD had previously discussed the need for a more meaningful way of measuring scan speed with Brian Honeycutt, Contex's Large Format Manager. Honeycutt told us that Contex wanted to provide a new measure of scan speed which he hoped the wide format scanner industry would be challenged to use or at least respond to.

At this time Scanners4CAD were under the impression that these were scanner speed tests. We told Honeycutt we would support Contex in this worthy endeavour.

Contex's scan speeds were made with perfect sheets of pristine drawing paper over the course of 15 minutes by two operators who fed the machine constantly.

As it is impossible to sustain this level of activity for one hour, this 15 minute figure was then multiplied by four to get the hourly rate. This only emphasises that Contex's test was conducted in a laboratory, not the Real World. Statisticians and experienced large format scanner operators may find fault with such easy extrapolation.

EDITOR'S NOTE: Scanners4CAD has egg over its face! Brian Honeycutt asked if we thought that the number of A1 / D-size drawings quoted should be for 15 or 30 minutes or one hour. Honeycutt had correctly identified that perhaps these numbers would be unreal to most people. We disagreed, saying they were impressive as "speed tests" and suggested that the measure be for the number of A0 / E-size drawings scanned in one hour. Ultimately Contex decided on the A1 / D-size, presumably because they were smaller, more manageable and easier to feed into the SD Series at high speed and scanned faster, thereby increasing the impressive number count. Many have faulted Contex's use of a drawings/hour measure, suggesting that a more practical and achievable measure would be drawings/minute. This was partly our fault and we apologise to the industry and to Contex for the poor quality of our advice.

All scans were made at 200 dpi resolution, allowing very fast scan speeds. Honeycutt defends Contex's choice of 200 dpi because it was the lowest reasonable resolution it could use. 200 dpi was chosen because the absurdly high document count will be remembered and because Contex's rivals will be unable to choose a lower resolution for any test they may do and so come up with higher numbers in their supplied data.

In order to maintain the speed and gain the impressive results, no image editing or clean-up tools were applied to the scans during the test and no image checking was done. While this may be acceptable in a test of raw scan speed, it is not in the commercial world where greater respect for image quality is expected. This shows the gap between Contex's competitive productivity test and Real World needs.

Contex's tests were made in Batch mode with Auto-sizing switched Off using only pristine same-size sheets. When scanning many different sizes of drawings, as one does in the Real World, Auto-sizing is often used. This slows the scanning process down. It is generally unlikely that only good quality documents of one size are available. Batch jobs often comprise piles of drawings of different sizes, media and condition. Also, the scans were saved with incremental file names, handy for speed but unhelpful as a description of what is in a document. Again, in the Real World, your needs may be different.

Most independent observers would argue that in the Real World with real drawings and real operators things would not be so easy. We thought Contex's figures were impressive as "speed tests". We did not know at that time that Contex wanted to use this data as a measure of productivity. Perhaps Contex did not know at that time either.

Scanner4CAD advised Contex that it "must be absolutely clear that your speed tests were made using perfectly flat drawings with no creases, curls and dog-ears i.e. that in a real-life working environment you are unlikely to work with such pristine images - and get such speed. You will need to add some caveats, like in the pursuit of excellence and image quality in a real-life situation different drawings may need different settings and individual attention to drawing detail and that all this will consequently take a little longer. If you do this, no-one can accuse you of pulling the wool over anyone's eyes."

Contex chose not to offer this qualification.

This is a double pity, not because we have been ignored, but firstly because such a qualification would have put this measurement of productivity into a Real World context and secondly, it shows that Contex really do believe that SD Series productivity can be measured by vast numbers of unchecked, low resolution scanned images.

We find this and its implications for scanned image quality rather scary!

However, Contex believe - and would have you believe - that the SD Series scanner is incapable of a poor quality scan.

[Contex Scan Speed - previous] [Contex Scan Speed - next]