April 13, 2012 - The Tide is High

...but I'm holding on.

The Challenge is to read the tidal height using this graph:

The vertical axis scale is:

0 - 5m

and the Horizontal axis is:

31-Dec-2011 00:00:00 to 3-Jan-2012 00:00:00.

Specific time:

1-Jan-2012 13:51:30

..............................

For those who did not see my previous Question you can use the attached user component to change the date and time into a more usable format.

..............................

I know of two very different methodologies. But as ever with GH I would like to see more :)

Attachments:

DateTimeCode.ghuser, 43 KB
TidalChart.png, 6 KB

Replies to This Discussion

Permalink Reply by Brian Harms on April 13, 2012 at 11:54am

This seems to work pretty well. Not super accurate just because of the .png's resolution.

Attachments:

wave reader.gh, 57 KB

Permalink Reply by Brian Harms on April 13, 2012 at 1:19pm

Ah, noticed a mistake. Here is my revised reader.

Andrew, I got a (slightly) different value than you (3.18m). I wonder if I made a mistake or if it's just the method that I went about reading the png (color hue).

I guess the thickness of the line is counted toward the total, maybe that's it.

Attachments:

wave reader_revised.gh, 103 KB

Permalink Reply by Brian Harms on April 13, 2012 at 4:19pm

Here's an alternate method. Same value though (3.18)

9 components not including panels

Attachments:

wave reader_alternate method.gh, 99 KB

Permalink Reply by Andrew Heumann on April 13, 2012 at 1:00pm

Well it won't win any beauty contests...

I'll go on the record with my specific-time solution at 2.87m.

Attachments:

Tidal_Chart.gh, 238 KB

Permalink Reply by Andrew Heumann on April 13, 2012 at 4:51pm

When I do the job manually in rhino, drawing lines on top of the image, dividing, measuring, I wind up with something very close to my value.

If you look at your "Find similar item" component, the value it's finding is actually a white rather than a blue pixel... perhaps that has something to do with the discrepancy? Converting your colors to points before passing them through returns 2.9 in the final measurements.

Permalink Reply by Brian Harms on April 13, 2012 at 5:40pm

Ah gotcha. I thought the find similar component would take each of the three components of color into account, looks like it doesn't. Good call with the point conversion.

So thanks to Andrew, here's a (much more accurate) version with 11 components not including panels.

Attachments:

wave reader_alternate method points.gh, 99 KB

Permalink Reply by Andrew Heumann on April 13, 2012 at 9:55pm

nice, very elegant and clean solution.

Permalink Reply by Danny Boyes on April 16, 2012 at 4:32am

Very nice indeed, I think Brian you are the winner, with a meritorious assist for Andrew for pointing out the follies of your method along the way. :)

For completeness I'll try and post my definition at lunchtime today if I get a chance.

RSS