Raster or Vector charts or BOTH?

[danielamartindm]

For planning purposes, I use raster charts with an eye toward water depths, obstructions, magnetic courses, reefs, spoil, leg distances, landmarks, Plan B's, etc. Rasters are great because ALL of the data is right there. For underway navigation, I prefer vector charts. I love their uncluttered screens (courtesy of layered data), and their depth contour lines instead of a zillion little raster numbers. If I need those under way for some reason (as in a narrow, shallow channel), I can always zoom in to see them.

[CaptTom]

Originally Posted by Dockhead

Vector charts are created on the fly by a computer, and according to the exact zoom level you need, so are not limited to a limited set of views as are raster charts, but the views are mechanistically created and always lack important detail when zoomed out. This makes them better for pilotage and for use in chart plotters. But they are dreadful for passage planning, with regularly fatal results.

Think about how the raster charts were originally digitized to vector. Someone looked (electronically) at the paper chart and clicked on some (clearly not all) of the place names, to fix their location in the raster data. But where is that place? The cartographer knew, and found a good spot next to the island, cove, port, harbor, bay or town where the name would fit, and where it was clear to the user what it was marking.

But when the person doing the digitizing clicked on the name, they just fixed it wherever it had appeared on the chart, not on the actual place. Zoom in or out, and the place name looks like it's next to something else. With a different background, the spot where the name now falls may not a good spot any more. I see this all the time. The name of one island appears three islands over.

But many place names never show up at any zoom level, even though they're on the raster chart. Oddly enough, the ones that do make it through the process are often irrelevant, like some upland pond or village which has no connection to the navigable waterway.

Not to mention all the shoreside structure that's not on the raster charts. buildings, hills, towers and steeples, built-up areas, highways, marshlands, etc. Ideally, everything visible from the water should be on the chart, and that's what the cartographers used to shoot for. Apparently, not any more!

The whole process smacks of outsourced, low-bid, contract labor. They filled their quota, and didn't really care about the end user. Now that the digitizing project is over, the locations that have been lost or erroneously placed will never be recovered or fixed.

[Dockhead]

Originally Posted by CaptTom

Think about how the raster charts were originally digitized to vector. Someone looked (electronically) at the paper chart and clicked on some (clearly not all) of the place names, to fix their location in the raster data. But where is that place? The cartographer knew, and found a good spot next to the island, cove, port, harbor, bay or town where the name would fit, and where it was clear to the user what it was marking.

But when the person doing the digitizing clicked on the name, they just fixed it wherever it had appeared on the chart, not on the actual place. Zoom in or out, and the place name looks like it's next to something else. With a different background, the spot where the name now falls may not a good spot any more. I see this all the time. The name of one island appears three islands over.

But many place names never show up at any zoom level, even though they're on the raster chart. Oddly enough, the ones that do make it through the process are often irrelevant, like some upland pond or village which has no connection to the navigable waterway.

Not to mention all the shoreside structure that's not on the raster charts. buildings, hills, towers and steeples, built-up areas, highways, marshlands, etc. Ideally, everything visible from the water should be on the chart, and that's what the cartographers used to shoot for. Apparently, not any more!

The whole process smacks of outsourced, low-bid, contract labor. They filled their quota, and didn't really care about the end user. Now that the digitizing project is over, the locations that have been lost or erroneously placed will never be recovered or fixed.

Well, that is a fundamental weakness of vector charts. But you don't care so much so long as you're not using them for navigation.



For many years, I used to do all my planning and navigation on paper, create a route, transfer to the nav system. This worked fine, except that it is a major PITA to keep that much paper on board, and basically impossible on a short handed pleasure vessel to keep it updated. Now I do it in OpenCPN on raster charts based on up-to-date official charts. With a large monitor, it is almost like having 1000 paper charts in your chart table. And the routes can be created electronically and can be transferred in a few seconds.



Once you've done this, then you actually appreciate the vector chart view on the small screen in the cockpit. I've tried to use raster charts on a 10" tablet; they are unusable on a screen of that size. I don't think you can say one is better than they other -- they are different, and each has its own purpose.



And BTW, not all vector charts are as badly made as the ones you describe. The Navionics ones are mostly pretty decent (although the quality varies according to the area), and the Platinum ones solve the problem you mention about land features, by overlaying Google Earth images. Albeit at a price!

[conachair]

Originally Posted by Tillsbury

I’m thinking that google earth may well be a primary source. Has anyone collated good sets of kap or similar files for some areas, or do we just get on with it ourselves?

Quite easy to make your own these days. One absolutely gold dust program IMHO no boat going anywhere remotely off the very beaten track should have is Sasplanet. Available here with the links set up to various satellite sources plus 2 nautical chart sources. Then set up the keyboard shortcuts and you can flick between the different sources to check if they agree and if the datum looks ok>

[Dockhead]

Originally Posted by conachair

Quite easy to make your own these days. One absolutely gold dust program IMHO no boat going anywhere remotely off the very beaten track should have is Sasplanet. Available here with the links set up to various satellite sources plus 2 nautical chart sources. Then set up the keyboard shortcuts and you can flick between the different sources to check if they agree and if the datum looks ok>

We did this for one area of Greenland, but unfortunately ended up going to a different area of Greenland, and kicked ourselves for not having GE2Kap files with us, as the charts were off by miles. It would have really helped a lot.


It's quite laborious, however. It would be really nice if there could be a system which would do a whole area rather than stitching together screenshot after screenshot.

[conachair]

Originally Posted by Dockhead

More than you ever wanted to know about it explained here:


https://chersoft.com/Resources/CherS...sterVector.pdf


"The key difference between raster and vector charts is in the way in which they are rendered, where rendering is process of translating the chart data into an image which can then be displayed on the screen. The screen is always raster display, that is, it is made up of dots. So:


"A raster chart is early rendered. The source data is analysed by a hydrographic office and the chart is drawn by a cartographer. The final image is decided by a cartographer and this image is captured for display on a computer screen.


"A vector chart by contrast is late rendered. In this case the hydrographic office collates a set of (vector) data which can be used to generate a chart image. However, the image is drawn by the computer and this is not actually done until just before it is displayed on the screen.

. . . .



"Early rendering means basically letting the cartographer do his job. This work is not only highly skilled but also time consuming. It is not unknown for a cartographer to literally spend all day looking at a new edition of a chart just to make sure that the layout and labelling and detail and myriad other factors are all just so; just right to make the chart clear, concise, uncluttered and useful for the navigator."


So, as I said, raster charts are images of human-drawn charts, whereas vector charts are computer-drawn charts based on databases of data pulled out of raster charts.



The human-drawn charts represented in paper or raster charts are discrete views interpreted by real cartographers -- these views are absolutely invaluable for passage planning -- there is a lot of advice and discretion embodied in them -- the cartographer is showing what his professional knowledge tells him you need to know -- he's talking to you. Each view (which corresponds to a separate paper chart) has the right amount of detail for that view; you don't have the "zoom problem" of vector charts and you generally have far too much information to display useably on a small screen but just the right amount of information to visualize a whole passage and different significant parts of it.


Vector charts are created on the fly by a computer, and according to the exact zoom level you need, so are not limited to a limited set of views as are raster charts, but the views are mechanistically created and always lack important detail when zoomed out. This makes them better for pilotage and for use in chart plotters. But they are dreadful for passage planning, with regularly fatal results, which is why on my boat there is always a vector chart in the plotter at the helm and boat nav system, and a raster chart on a large 4k monitor at the nav table, rendered by OpenCPN, or else a paper chart.


Except in Danish waters, where I don't have any vector charts other than CM93, so I use an NV Charts raster chart chip, which doesn't work very well in the B&G Zeus nav system.



Read the linked white paper for more information.

Ta for the link, interesting reading. Seems I was pretty close, the data is exactly the same - just the presentation is different, though "Istr reading that raster charts are created from the exact same vector chart data tweaked to make it look nice" doesn't really come close to doing justice to the care, skill and time taken to create a raster chart .

Possibly the best answer to the first question is - if you need to ask then mvbe raster would suit you better. The link did seem quite dated, and possibly over egging the case a little for raster - the 2 images of Cowes looked like they had turned every layer on and had data density turned up to absolute max for the vector. The anti aliasing and redraw rendering haven't been a problem for years, even the humble raspberry pi can easily redraw and dozens of frames per second.


From using computers and CAD for decades I've no qualms whatsoever to using vector start to finish on a computer, completely comfortable with turning off or on layers which are of no interest to whatever is being done at the time, more so, I find raster really annoying as you can't filter out all the non relevant text etc or double click to see the notes etc. And the team vestas running aground just seem like "how could anyone at that level be so completely stupid???"
But again that's from years of CAD, for many or possibly most that experience and instinctive way of using vector data just isn't there. And team vestas DID run aground. So a strong case for raster. But up to date vector and up to date raster the data is the same AFAICS, just different presentation.

Overall I think you're being far too hard on vector, the data is there, you need to learn a different way to work as the display concept is different and for many that could well be a step too far. And at the end of the day all charts IMHO should be read with suspicion, data could be old, buoys could have moved last week....

All IMHO of course

[conachair]

Originally Posted by Dockhead

It's quite laborious, however. It would be really nice if there could be a system which would do a whole area rather than stitching together screenshot after screenshot.

It's not anymore, simple now. Selection manager offfers various choices - screen grab, polygon or line etc and download the mbtiles in sasplanet then opencpn can view them in their own chart groups, split screen if you want. Very

No need for any stitching.

[Dockhead]

Originally Posted by conachair

Ta for the link, interesting reading. Seems I was pretty close, the data is exactly the same - just the presentation is different, though "Istr reading that raster charts are created from the exact same vector chart data tweaked to make it look nice" doesn't really come close to doing justice to the care, skill and time taken to create a raster chart .

Possibly the best answer to the first question is - if you need to ask then mvbe raster would suit you better. The link did seem quite dated, and possibly over egging the case a little for raster - the 2 images of Cowes looked like they had turned every layer on and had data density turned up to absolute max for the vector. The anti aliasing and redraw rendering haven't been a problem for years, even the humble raspberry pi can easily redraw and dozens of frames per second.


From using computers and CAD for decades I've no qualms whatsoever to using vector start to finish on a computer, completely comfortable with turning off or on layers which are of no interest to whatever is being done at the time, more so, I find raster really annoying as you can't filter out all the non relevant text etc or double click to see the notes etc. And the team vestas running aground just seem like "how could anyone at that level be so completely stupid???"
But again that's from years of CAD, for many or possibly most that experience and instinctive way of using vector data just isn't there. And team vestas DID run aground. So a strong case for raster. But up to date vector and up to date raster the data is the same AFAICS, just different presentation.

Overall I think you're being far too hard on vector, the data is there, you need to learn a different way to work as the display concept is different and for many that could well be a step too far. And at the end of the day all charts IMHO should be read with suspicion, data could be old, buoys could have moved last week....

All IMHO of course

Well, I'm not being hard on vector, at all. I've written several times that I consider vector charts far more useful on chart plotters, and for real-time orientation during pilotage. You can zoom raster charts but they respond very badly to this -- there is one particular view for each chart, which is really useful, and further zooming around degrades the information. Vector charts on the other hand create a new view for each zoom level -- and that can be a huge advantage in pilotage.


I don't agree with you, however, about the data being exactly the same -- it's not. There is a lot of data added by the cartographer (including the notes!) which is not carried over into a vector chart.



But much more important than that -- are you missing this? -- raster charts are views produced applying the judgement of the cartographer. This is profoundly important. This is stripped out when data is extracted for use in the database of a vector chart, and substituted with computer judgement of which detail is needed at what zoom level. This is the real difference between vector and raster charts. This is intentionally exaggerated in the linked white paper, by the overcluttered view of Cowes, but the point is valid and very important.



Yes, you can twiddle the knobs on OpenCPN with a vector chart and improve the view, but the computer is still choosing what to show and what to suppress -- the knobs you can twiddle are only generalized rules -- with fatal results for some people using vector charts for navigation. And with a normal chart plotter, you don't even have these knobs to twiddle (however, to compensate, you commercial chart plotter does a really good job with the computer rendering, within the inherent limitations of this approach).



We had a thread on this a few years ago where I posted profoundly different views of the coast around Selsey Bill from raster and vector charts. Maybe you could dig that up -- the vector chart view was not exaggerated and you will see the massive difference. What was shown was that basically it is impossible with the vector chart to see the whole passage between the Solent and Brighton with significant hazards visible, but that this is presented in exquisite detail, plus drawn by a human in a highly comprehensive way, on the raster chart. It is pretty eye opening.

[transmitterdan]

I think very little data is “stripped out” of nationally produced vector charts. The data is there. But it takes some training and practice to use it. There is lots of text within vector charts but you have to access it. Clearly there is a lot more info in vector charts than in raster.

Now we speak about the wisdom of the raster cartographer in removing some data in the interest of clutter. I would suggest that the raster data is not often reprioritized thus the wisdom is quite old in many cases. Raster charts often contain useless information because they are updated but almost never redrawn from scratch.

Whether we like it or not raster is going to be phased out. The question is only when?

[ranger58sb]

Originally Posted by Dockhead

But much more important than that -- are you missing this? -- raster charts are views produced applying the judgement of the cartographer. This is profoundly important. This is stripped out when data is extracted for use in the database of a vector chart, and substituted with computer judgement of which detail is needed at what zoom level.

Yes, you can twiddle the knobs on OpenCPN with a vector chart and improve the view, but the computer is still choosing what to show and what to suppress -- the knobs you can twiddle are only generalized rules -- with fatal results for some people using vector charts for navigation. .

Ummm.... small point... There was a programmer somewhere behind that computer decision-making...

Of course some software developers are better than others, probably as some cartographers are better than others...

-Chris

[conachair]

Originally Posted by Dockhead

W
I don't agree with you, however, about the data being exactly the same -- it's not. There is a lot of data added by the cartographer (including the notes!) which is not carried over into a vector chart.

Don't think I've come across that. Tides diamonds might be missing but the other notes are on there, double click. Just checked a handful from visitmyharbour on cm93, double click the relevant icon at the place where the note relates to. Not as much fun as reading with the chart out on the saloon table though..

But much more important than that -- are you missing this? -- raster charts are views produced applying the judgement of the cartographer. This is profoundly important. This is stripped out when data is extracted for use in the database of a vector chart, and substituted with computer judgement of which detail is needed at what zoom level.

Not missing that at all and I find it one of the benefits, takes about 3 seconds to change the detail level on opencpn(just checked ) , but again - if you are used working with vector data and your head works that way.

What was shown was that basically it is impossible with the vector chart to see the whole passage between the Solent and Brighton with significant hazards visible, but that this is presented in exquisite detail, plus drawn by a human in a highly comprehensive way, on the raster chart. It is pretty eye opening.

Well, just did a quick passage on opencpn with no issues whatsoever - very little difference between flipping between VMH & cm93. Other than cm93 being old of course Mouse zoom in and out to check along the way.

Anyway , each to their own, vector is fine for me but can see how it could lead to issues with the tech if not used to it.

[Dockhead]

Originally Posted by transmitterdan

I think very little data is “stripped out” of nationally produced vector charts. The data is there. But it takes some training and practice to use it. There is lots of text within vector charts but you have to access it. Clearly there is a lot more info in vector charts than in raster.

Now we speak about the wisdom of the raster cartographer in removing some data in the interest of clutter. I would suggest that the raster data is not often reprioritized thus the wisdom is quite old in many cases. Raster charts often contain useless information because they are updated but almost never redrawn from scratch.

Whether we like it or not raster is going to be phased out. The question is only when?

It is not the underlying data which is stripped out -- it is the PRESENTATION of the data in particular coherent views. Even if drawn 100 years ago, these are still totally relevant and valuable, if the details are updated, so the age of the original drawing has nothing to do with it. The cartographer's art is highly relevant still today.



I appreciate that computers get better and better at everything, and I would love to see vector charts which are capable of the very useful views we get today from say the official Admiralty paper or raster charts, but we have a very long ways to go.



I am not aware of any plans by the UKHO to phase out either paper or raster charts. Maybe some other cartographic offices, but I haven't heard about it. I suppose one day all this must happen, and one day human cartographers will be replaced by computers, like probably everything else human, but I personally doubt that this will happen during my lifetime.

[LoudMusic]

I can't speak to the value of navigation chart option in regards to vector versus raster, but from a technological standpoint, vector is a vastly superior way of storing and displaying mapping data. Whether or not anyone is making quality marine maps this way is not my area of expertise, but they absolutely should be.


Is anyone (Garmin?) collecting sounding information from their users and processing it into their databases and then redistributing it to their users? Seems like they have all the systems in place to "crowd source" charts.

[conachair]

Originally Posted by Dockhead

It's quite laborious, however. It would be really nice if there could be a system which would do a whole area rather than stitching together screenshot after screenshot.

Real quick tutorial....
Download sasplanet from here, it's got all the data sources etc set up
Index of /Downloads


then operations-selection manager - polyline selecton.

Make a line along a coast or river ..

click the green button.

In selection manager try these settings -

19 might be a bit much, I changed to 18 * 15 after taking this snapshot - huge slow downloads.
You can cancel this and go back into operations-selection manager - edit last selection to tweak the area.
Then when happy click start . Then wait for a very long time as all the tiles download.

Click quit then operations-selection manager - last selection.
Go to the export tab >

Have the zoom settings the same as the downloaded zoom settings

Then save and point opencpn the the same location and update chart database

[transmitterdan]

Originally Posted by LoudMusic

I can't speak to the value of navigation chart option in regards to vector versus raster, but from a technological standpoint, vector is a vastly superior way of storing and displaying mapping data. Whether or not anyone is making quality marine maps this way is not my area of expertise, but they absolutely should be.



Is anyone (Garmin?) collecting sounding information from their users and processing it into their databases and then redistributing it to their users? Seems like they have all the systems in place to "crowd source" charts.

Yes, marine charts are all available in vector format for many years. The entire world has been vectorized with varying levels of detail and accuracy. All US navigable waters have vector charts. Inland rivers are only available in vector format.

With modern compression algorithms the advantage of vector charts in terms of storage and fast display are getting less over time. I would say today raster charts are displayed as fast or faster than vector data. OpenCPN has to pre-build vector ENC charts else it would be much slower than for raster. The ENC file format isn’t so speed friendly.

Navionics collect crowd sourced depth data. However, the accuracy is not that great since calibration of the sensor is a bit unknown. A more popular use of crowd sourced data is locations of anchorages, local amenities and the like that help cruisers plan their trips and select destinations.

I am surprised no one is lamenting the uptick in using satellite imagery in place of official charts. These images are pretty much devoid of any human manipulation yet they are wildly popular especially for areas where official charts are known to be inaccurate or very old (or both). But i also hear of people using sat imagery for locations where official charts are readily available.