Digitising field drawings

[Steve H]

In the office at the moment we have 2-3 diggers busy digitising field drawings onto AutoCAD.
Despite there being two unused digitising tablets in the office, they are all scanning the drawings then over drawing the scanned images improted as rasters.
I was taught that using a digitising tablet was both more accurate and faster.
What do people think?
What are the advantages and disadvantages of the two methods?
Thanks for your comments:face-approve:

[kevin wooldridge]

I don't have any problem with the methodology (scan and trace). However whats the point of AutoCAD...? Straight into GIS for me and then export in a CAD friendly format if needed later....(but before Chiz comes back on me, I would say use a GIS progrmme that is more user-friendly than QGIS (ArcGIS or even ArcViewfor example)..

[BAJR]

For me... scan, trace in corel and then georef into GIS. Digitisng tablets... ah me... I remember my first one... and now I use a pen tablet and a scanner much faster. I can draw as fast as with a pen.

It does come down to what you are digitisng and what software and hardware you have

Don't mention QGIS to Chiz

[barkingdigger]

Nothing wrong with CAD - it all depends on what you have (as BAJR said) and what you intend to do with the data. And these days it's easy enough to port linework between CAD and GIS, with a little forethought.

When digitising in CAD I prefer the "scan & trace" method these days, having got the T-shirt from years of calibrating tablets for digitising off the permatrace. You can zoom in on the scan for better accuracy, and of course it all takes up less room on the desk! In skilled hands there's very little difference in the time required, in my humble opinion...

[ShadowJack]

I've used both methods in CAD and GIS over the years and personally, although I have to to use the scan method at the moment, I prefer digitising into CAD using a tablet. I've just found that CAD gives me more flexibility than GIS (Especially when comparing AutoCAD Map 3D to Mapinfo) and using tablets is quicker and rules out file management issues when trying to deal with the storage of hundreds/thousands of plan/section scans. But, at the end of the day, it simply comes down to what you are comfortable with and the combination of software/hardware at your disposal.

The only word of caution I have is to be aware of scan distortion. All our scanners, A4/A3, together with the commercial large format scan services we use, create a distorted image - usually more pronounced in the horizontal than the vertical, which although is negligible for individual sections/ trench plans, can have a significant effect on the accuracy for large area excavation plans if you are scanning in and digitising a large number of adjoining plans. The distortion tends to result in the grid on an A3 plan sheet being out by up to 5mm once it's scanned.

As you can image - if you have a hundred scans for an open area excavation and each scan is distorted by 5mm - you might begin to wonder why the digitised plans don't seem to match the GPS data .... I correct for this by setting up an accurate grid in CAD and then rubbersheeting the scans to the grid (I use AutoCAD Map 3D specifically for this reason).

I first came across the problem when a project manager wanted me to figure out why they were not only having problems tying in survey data to a phase of quarry excavation plans - but also why they were having problems tying in that phase to previous phases .... which also seemed to be 'a bit iffy'!

The problem with scan distortion is also the reason why it is never, EVER a good idea to either scan an object directly, or scan your pencil drawing for inking up digitally or printing out at a larger scale for hand-inking. I know of at least one large corpus of Anglo-Saxon material for a cemetery site which had to be redrawn at a late stage when the specialist realised the final drawings didn't match the dimensions he'd recorded - simply because the illustrator, to save time, scanned the 1:1 pencil drafts into the computer, re-sized them to the desired scale, and then traced over the resultant print-outs.

How about mentioning gvSIG?

[Steve H]

I completely agree with Shadowjack. He rehearses the arguements a few of us have been having with the 'scanners' very well.
As an aside, the people who favour scanning use a really quite bizare method of importing all the scanned drawings onto the CAD plan and digitizing them, but never getting rid of the scans. This of course means that a big CAD plan can take an age to open and sometimes it becomes impossible to open or do anything with.

[ShadowJack]

It can be a common problem when dealing with large area excavations or very complex multilayer sites. A workaround I sometimes use is to import and geo-reference all the scans into one standalone CAD file, turning off or unloading each scan as it's imported and overlaying this with a planning grid identifying the location of each scan by it's plan number. Once they've all been imported (and turned off) I then import this into a new CAD file as a reference file and just turn on the individual scans as and when they're needed. Means once the site is digitised I can detach the scan reference without losing the geo-referencing for the images and I can have an excavation plan of the site without the clutter of the image references hogging up memory.

[barkingdigger]

Interesting points! Scan distortion is something to watch out for, but like anything technical it can be alleviated with appropriate effort. Trouble is, if you buy in a bulk-scanning service it can be hard to implement effective Q-control. And if you use affordable flatbed scanners you do need to carefully choose to reduce the error (5mm on an A3 sounds a bit extreme...). Coming from an "urban" background where the average A3 sheet rarely holds more than a 5x5m patch, with grid points in all four corners (a hanging offense to omit these on MY projects!) the odd mm of distortion on the scan only blends in with the thick pencil lines and the slight uncertainties normally encountered when drawing features in a trench. Thus any error is trapped within each 5x5m square, and doesn't grow across the site. Planning at smaller scales (1:100?) to fit whole fields on a sheet magnifies the error, but then it also magnifies all the drawing errors too. And I never got on with those back-breaking A1 boards...

As for scans in CAD, AutoCAD only incorporates a rectangular "frame" and associated image path within the CAD drawing file - the actual raster image lives outside in its own file. The thing that really bloats the CAD file is all the digitised linework, since each line segment is defined separately in the file and needs to be "drawn" whenever the file is opened or zoomed or panned. But once again we return to Q-control and a firm sense of purpose. If the CAD drawing is the "master" dataset for a project it will have lots of extraneous stuff that will slow it down, but the "deliverable" products ought to be much smaller & more manageable exported sub-sets of the data, only enough to do the job for each intended drawing product. If your CAD system is somehow sucking in the whole raster image into the CAD file then you need to look at procedures & settings. For digitising (and rapid opening-times for CAD files) I also convert bloaty TIFFs to JPEGs - these are "working" copies so don't need to be archivally stable, and the visual effect of JPEG degradation is minimal.

And yes, I prefer doing "drawing" tasks like digitising in CAD rather than GIS, simply because CAD was developed as a drafting tool while GIS is more limited on the drawing-tools front.

[ShadowJack]

Completely agree - quality control is the key to making these systems work. Over the years I've developed different templates, layers, codelists, procedures and way too many manuals to cover topic like these!

Isolated features within discrete planning areas on a single sheet rarely create any problems, which is why I don't worry about digitising trench plans or sections drawings - its only where you have large 'organic' linears which traverse several sheets that you tend to notice the problem.

The problem I find is that even though the CAD file only contains the frame and path to an image, on opening - the software still has to load that pathway and, if like me, you have to work on either university or local authority network (we're just about to migrate from XP professional to Windows 7!)...... that is the Achilles heel. If I leave the xref manager open in CAD and accidentally let my mouse move over an image xref - the entire programme hangs for a couple of seconds - even though the image isn't loaded, while it tries to retrieve info from the system that I might want to know.

So, you have to adapt and find workarounds.

[kevin wooldridge]

... I prefer doing "drawing" tasks like digitising in CAD rather than GIS, simply because CAD was developed as a drafting tool while GIS is more limited on the drawing-tools front.

Thats fair enough. My personal preference is for GIS over CAD however....mainly on the grounds that I believe that 'illustration' is the by product/end result of analysis and synthesis of data. I think that CAD at too early a stage in that process is indiscriminate. Everything gets digitised irrespective of value or sigificance. Whereas digitising into GIS, whilst perhaps less aesthetic, does allow the totality of the data to be considered (the 'image' is just the front end to any number of attribute values), analysed and synthesised ..........after which the relevant image can be exported to CAD and 'tidied up' for the final illustration if needed. In these days of watching every penny and getting full value for the archaeological buck that seems the efficient way to go....

My GIS by the way will georeference and overlay scanned raster image data without the need for any further digitisation at least at the primary data stage. As we routinely scan all plan and profile data as a security consideration that seems the ideal manner in which to minimise costs with no loss of capability. At the same time the raster data can be integrated with the site survey data from the TST that requires no further digitisation.

Is that capability possible in CAD?