Subscribe to this thread
Home - Cutting Edge / All posts - Manifold System 9.0.173.1
adamw


10,447 post(s)
#31-Dec-20 10:28

9.0.173.1

Here is a new build.

manifold-9.0.173.1-x64.zip

SHA256: a2822508479f4a933a213c2e8b19b83770654d657619f252539003157f8df82c

manifold-viewer-9.0.173.1-x64.zip

SHA256: 53d87b2de48745a4f5e89ceb1013306ace6eafe67ebb562b5b1c425d5f302de9

This build contains our first take on registration.

Happy New Year!

adamw


10,447 post(s)
#31-Dec-20 10:29

Registration

There is a new Register pane. Keyboard shortcuts for activating panes are adjusted to: Ctrl-1 = Project, Ctrl-2 = Info, Ctrl-3 = Layers, Ctrl-4 = Register, Ctrl-5 = Select, Ctrl-6 = Style, Ctrl-7 = Transform.

The registration process involves two windows: one for the component being registered (source) and another for the desired location (target). Each window has its own set of control points. Initially all windows works as registration sources. The Register pane allows making the active window a registration target for a different window which will work as a registration source. After that, the set of control points for the target window is synchronized with that for the source window. A target window cannot be a source window for some other window. If a source window is closed, all target windows for which it has been the source window return to being source windows.

The Register pane shows the list of control points for the active window. Each control point has a unique name and an optional location, shown using an icon on the right. For a source window, each control point always has a location. For a target window, a specific control point may or may not have a location.

The map window supports a new cursor mode for editing control points: Edit Control Points. The mode uses a distinctive cursor. Switching to the Edit Control Points mode shows the control points and shows the Register pane if it was hidden before.

For a source window, clicking in the map window places a new control point. The new control point is assigned a unique generated name.

For a target window, clicking in the map window assigns the location for the control point focused in the Register pane. If the control point already had a location assigned, it is changed to the new one. After the click, if there are control points with no assigned locations, the focus moves to the first such point.

Painting control points in the map window automatically moves the names around points to try to stay on screen.

The Register pane allows editing names of control points for a source window. If the newly entered name is not unique, it is forced to be unique by adding a numeric postfix. The names of control points in a target window are automatically adjusted to be the same as those in the source window.

The Register pane allows selecting control points. Selected control points can be deleted using the Delete key. For a source window, deleting control points removes them from the set. For a target window, deleting control points clears their target locations.

After both the source and target control point sets are placed, the Register pane for a target window allows registering any layer from the source window to the target locations. The list of layers omits labels and includes only drawings and images. The Register button performs the registration, creating a new drawing or image component and selecting it in the Project pane. The Preview button previews the results of the registration. The Edit Query button shows the registration query. Running the registration query will have the same effect as pressing the Register button would.

Registering a component allows specifying the registration method. Available methods are:

  • order 1 (affine)
  • order 1.5 (affine + cross-product)
  • order 2 (square)
  • order 3 (cubic)
  • triangulation

The default registration method is 'order 1 (affine)'. Avoid using 'order 2 (square)' and 'order 3 (cubic)' with small numbers of control points, use at least 5-6. Using 'triangulation' will only register data covered by the convex hull of the control points, you can use a set of control points that does not cover the whole component for a drawing, but not for an image.

Registering a drawing sets the coordinate system of the new drawing to that of the target window with local scales set to 1 and local offsets set to 0.

Registering an image sets the coordinate system of the new image to that of the target window with local scales adjusted to produce about the same number of pixels as in the original image and with local offsets set to the left-bottom corner of the image. The pixels in the new image are forced to be square.

Registering a component in the Register pane shows the time it took to perform the operation at the bottom of the pane. This helps confirm that pressing the Register button did have an effect when the operation ended up being too fast and the progress dialog never appeared. Registering a component also logs operation time in the log window.

New query function: CoordConverterMakeWarpNumeric - takes a line geom containing warp coordinates and a numeric order for the warp, returns a coordinate converter object that performs the warp. Each branch of the geom is interpreted contains the source location as the first coordinate and the target location as the last coordinate. The numeric order is rounded to one of the following values: 1, 1.5, 2, 3.

New query function: CoordConverterMakeWarpTriangulation - takes a line geom containing warp coordinates, returns a coordinate converter object that performs the warp.

(Fix) The minimum height of the template list in the Select and Transform panes has been increased. (Could become too small if a pane was undocked and reduced in size.)

End of list.

(Planned additions: save / load control points, provide more control over registration parameters, report registration error for each control point.)

pslinder1
228 post(s)
#31-Dec-20 15:54

That's great. Looking forward to the video. Happy New Year.

Dimitri


7,413 post(s)
#01-Jan-21 06:57

Here's the new video. Happy New Year!

Mike Pelletier

2,122 post(s)
#31-Dec-20 18:12

Very glad to the new register pane. The preview option is fantastic as expected. Just experimented with a simple image and once you learn the steps it works very well.

It would be very handy to have a way to import turn on/off control points in the register pane that you want to apply in the georegistration. That way you can experiment with different control points without having to manually move them around.

Some thoughts on saving/exporting control points. I'm thinking it would be great to consider saving control points for each component automatically, so that it is easy to revisit at a future date. Then add a tool to export them to another component and rename them automatically as needed to avoid conflicts. Also, should have a way to import/export to a point drawing.

Happy New Year all

geozap
264 post(s)
#01-Jan-21 07:56

Adding of registration is a very nice step forward!

Some ideas:

a) Reports for errors. I like a lot that you will provide text reports for errors. The lack of such an option in M8 was one of the few things that seemed "unprofessional" in M8. The error surface that exists in M8 is a "nice to have" option for maybe later I think.

It is important to add not only report of error for each control point, but also the total registration error.

Also it would be really useful to see (in the pane) the error not only after the registration, but before too. So, one could place again problematic points, see if more points should be added, etc.

The after-registration report could be viewed in the description of the newly created image, so that it would "follow" the image for future reference.

b) Manually typing points coordinates and coordinates being visible in the pane. Apart from being sometimes a easier way to add control points, instead of having to place points entities and then snap to them, it also helps avoiding errors from no or erroneous snapping.

c)Snapping pointer. When snapping is enabled, the little blue snap square is not visible, like it is when creating objects. I think it should be visible to know if and where the user is snapping.

d) Control points visibility. Currently, control points are visible only while being in edit control point mode. There should be a way to keep them always on. For example, you may want to use the tracker to check an error using the preview. You cannot do this know because the points hide when you select the tracker tool.

e) Points activation/deactivation. The user should have the option to activate/deactivate each point in target/source, without having to delete it, to see how errors or the preview change.

f) Triangulation for images even when the points don't cover the total image. This could be done like in

M8, with the image outside the convex hull being deleted. Triangulation for images if the best and sometimes the only way to go when having to register scanned old paper maps with different distortions in X and Y, or being locally deformed.

If this way of triangulation is not possible now because of the current lack of by-pixel selection, at least manifold should add that option on the future, when by-pixel selection becomes available.

Nice to have in the future:

g) After having placed a couple of target points, a button (or right click menu) to pan on screen to the expected position of the next target point to place.

geozap
264 post(s)
#01-Jan-21 08:30

And a happy new year!

Dimitri


7,413 post(s)
#01-Jan-21 12:16

Great ideas! Some comments....

b) Manually typing points coordinates and coordinates being visible in the pane. Apart from being sometimes a easier way to add control points, instead of having to place points entities and then snap to them, it also helps avoiding errors from no or erroneous snapping.

You could do this by placing points in a drawing that is in the target map, and then having a control to use those points / importing them. One way or another you'll need a user interface to manually specify the coordinates of points, so may as well use the one already there that people have learned (adjusting the position of a point by manually specifying coordinates by entering them in the Info pane's Coordinates tab).

e) Points activation/deactivation. The user should have the option to activate/deactivate each point in target/source, without having to delete it, to see how errors or the preview change.

Maybe in the Register pane there could be a column with checkboxes added, like in the Layer pane.

If this way of triangulation is not possible now because of the current lack of by-pixel selection, at least manifold should add that option on the future, when by-pixel selection becomes available

It's not about per-pixel selection, just about what conveniences you want to add for triangulation, such as automatic clipping of an image to just that portion that you want to triangulate, and what happens to the image outside the region where you've placed control points. Right now, in 173.1, the transformations that are done are sub-pixel transformations just like projection, which is better accuracy than you get with 8 or most other classic GIS.

Also, the video has been updated to add a small section near the end showing how to do live previews of using different georegistration algorithms (affine, triangulation, etc). Links in the forum posts and on the Videos page have been adjusted to point to the new video.

geozap
264 post(s)
#02-Jan-21 06:31

It's not about per-pixel selection, just about what conveniences you want to add for triangulation, such as automatic clipping of an image to just that portion that you want to triangulate, and what happens to the image outside the region where you've placed control points.

As a minimum, the image outside the convex hull could just be deleted, like in M8. I hope manifold at least adds triangulation for images using this method of dealing with pixels outside the convex hull. Triangulation is realy important because it is the best way to go in my opinion when dealing with old maps localy distorted from age (I think mostly because of humidity or constantily rolling and unrolling of the sheets) or when one wants to correct scanner distortions.

Regarding the cases when one wants to keep the outside of the convex hull too, in M8 I usually do 2 registrations. One triangulation for the part I can cover with control points and a second, affine transformation, for the other part. I prefer to use affine for the outside because usually I am dealing with scanned maps of property lots, an it is important to keep straight lines straight. (Polynomial transformations distort lines to curves.). Then I merge the 2 images and usually they match decently well.

It would be nice to have both methods: delete the area outiside the point covered area or somehow register and keep the outer part too.

adamw


10,447 post(s)
#16-Jan-21 09:33

Limiting registration for triangulation to the part covered by that triangulation is fine, we'll do this.

Extending it outside of the triangulation will necessarily involve guesswork, though. It is unclear where the locations outside of the triangulation should go. One might say "just do something, I don't care much what it is", but if that's the case, why not just put four extra points outside the triangulation to cover the whole image? We could try doing essentially the same thing automatically, of course, is this what is desired?

geozap
264 post(s)
#16-Jan-21 14:36

The cases I have needed "triangulation inside, affine outside" is when I deal with scanned old and worn paper maps, whose perimetric grid is torn apart. I am a land surveyor and I have dealt with a lot of these, but most GIS people, and possibly land surveyors elsewhere in the world, don't have such problems. Also, since you can now save control points in a drawing, and when selection by pixels comes, one could easily perform the 2 transformations, mask one of the 2 results, and merge the result images.

So what I was talking about is maybe a little needed luxury.

Mike Pelletier

2,122 post(s)
#16-Jan-21 15:20

You are not the only one :-) Working with historical maps can be very fascinating detective work. This tool would be very handy. Technically no idea what to do but here is the intent.

I like to first get the map in the ball park without warping, just scale/shift/rotate. Sometimes it ends there as it is more important to preserve the original map than the alignment. Other times it is nice to adjust what you can to align with modern day maps but keep the rest for context despite it being unaligned.

I've actually been waiting for this 9's registration tools for a long time before I tackle this job for all the historic maps in our area. :-)

Never done this but it might even be possible to merge into the original map some error registration raster to show the distortion. Might be worth trying in some cases.

geozap
264 post(s)
#03-Jan-21 07:46

An addition:

Simpler than 1st order affine transformations, that is move, move-rotate, move-rotate-scale, are also useful sometimes and I think should be added, as they exist in M8 too.

And a question, what kind of transformation is the the "1.5 affine+cross-product". Is it the one called "Projective"?

Mike Pelletier

2,122 post(s)
#05-Jan-21 16:46

Really like everyone's suggestions in this thread. Agree with just a number for the control point names to ease renaming. Geozap's suggestions would be especially helpful for my work. Hope those are getting to engineering :-)

Dimitri


7,413 post(s)
#06-Jan-21 08:31

Hope those are getting to engineering :-)

Suggestions are the tangible way of putting hopes into action. They are the direct pipeline to engineering and take but a moment to send in. Have you sent in yours? :-)

If not, it's risky to assume somebody else has sent in a suggestion that matches your preferences. Even if others want to be supportive, it's easy to miss a post, assume the poster has already sent in their own suggestion, or just get busy with their own priorities and forget about a specific comment.

I've sent in about a dozen suggestions based on my own use of the Register pane, but that's only one person's preferences and priorities, which might not be the same as yours.

Dimitri


7,413 post(s)
#06-Jan-21 08:38

Simpler than 1st order affine transformations, that is move, move-rotate, move-rotate-scale, are also useful sometimes and I think should be added, as they exist in M8 too.

Already there: that is what first order affine transformations do. They shift, scale and rotate.

what kind of transformation is the the "1.5 affine+cross-product".

First order affine transformations are shift, scale and rotate. Second order transformations do curvatures. 1.5 + cross products are midway between first and second order, in that they do shift, scale, and rotate with some curvature added.

ESRI has a nice discussion here.

tjhb
10,094 post(s)
#06-Jan-21 08:49

As far as I can see the ESRI discussion does not cover "1.5 + cross product", not in those terms but I think not at all.

What it sounds like intuitively might be shift, scale, rotate and shear? I don't know.

Dimitri


7,413 post(s)
#06-Jan-21 12:25

ESRI's discussion is useful for a general overview of first order, second order and third order polynomial georeferencing transformations. They cover "shear" in their discussion of first order transformations:

"The first-order polynomial transformation is commonly used to georeference an image. Use a first-order or affine transformation to shift, scale, and rotate a raster dataset. This generally results in straight lines on the raster dataset mapped as straight lines in the warped raster dataset. Thus, squares and rectangles on the raster dataset are commonly changed into parallelograms of arbitrary scaling and angle orientation."

The ESRI ArcGIS Pro names/transformations are not a one to one match to Manifold's currently available methods (more are planned). For example, their "zero order" transformation (shift only) for vectors is the Reshape : shift transform template in Manifold, done as part of the Transform pane and not as part of the Register pane. But for convenience, Manifold could add that to the first order affine method, so that if you used just one control point it would simply be a shift.

So far, both ESRI and (pending) Manifold documentation on the methods are qualitative, and not quantitative. For example, ESRI doesn't really say what "similarity polynomial", "adjust", or "projective" are, or provide details on "spline" (likely to be thin plate spline). I expect Manifold will provide more technical descriptions as the documentation fills out and more methods appear (thin plate spline is upcoming in 9, for example).

If you drill way, way into past documentation for ESRI, you discover that "similarity" is simply a constrained first order affine, that is, without any independent scaling of axes or any skew (ESRI's word for "shear" as used in this thread), thus maintaining aspect ratio. "Adjust" likewise appears to be a limited form of affine.

I expect Manifold will sooner or later provide all the ESRI methods just for convenience, as they are trivial variations on the basics.

Interestingly enough, ESRI uses a completely different user interface from raster georeferencing for their surprisingly limited ability to georegister vectors. It's the "spatial adjustment" tool, which provides only Affine, Projective, Similarity, Rubbersheet (an ESRI'zed version of triangultion) and, in ArcMap but apparently not in Pro, "Edge snap". It doesn't use control points but "displacement links".

I much prefer Manifold's way, where the same user interface works to georegister either raster images or vector drawings. That also allows you to simultaneously deal with a huge stack of mixed rasters and vectors in a source map that are to be georegistered to a known good target map constellation of layers.

geozap
264 post(s)
#06-Jan-21 08:53

Affine 1st order sheers too, not just move-scale-rotate.

Dimitri


7,413 post(s)
#06-Jan-21 11:20

I think that the classic "shift, scale and rotate" description of first order, affine, polynomial transformations in vector transformations is intended to cover the notion that shifts of individual coordinates might be first order different, thus producing a shear effect.

adamw


10,447 post(s)
#16-Jan-21 09:41

Order 1.5 (affine + cross-product) is the case where new X = aX + bY + cXY + d, same for new Y. Without the term for XY, this becomes order 1 (affine). With terms for XX and YY this becomes order 2 (square).

Regarding methods simpler than affine, we will likely add means to merely scale + shift - without altering any of the actual data, simply by adjusting the parameters in the coordinate system. We can add optional restrictions to the regular affine as well - what would be desired: 'shift + scale only'? something else?

geozap
264 post(s)
#16-Jan-21 14:45

For images, more useful than shift-scale is shift-scale-rotate, because an aerial photo or scanned paper map are not going to be exactly rotated to real world grids. (But in those cases I think most people would anyway prefer triangulation or affine 1st order with shearing.)

For vector drawings such as floor plans shift-rotate would be useful sometimes because you don't want to change scales or lose lines being perpendicular (no shearing wanted).

pslinder1
228 post(s)
#09-Jan-21 16:13

I think a), b), d) are particularly important to add.

I am very impressed with the new georegistering process. In a previous life I had the dubious pleasure of being responsible for the georegistration of building floorplans on an industrial scale and I can say that the new process is simpler and quicker than the process in arcgis, mapinfo or qgis. Bravo.

It would be cool though to add a completely graphical work flow that would allow a user to overlay the image (with a variable opacity setting) that needs to be registered on a map or drawing. The user then could rotate the image and scale it by manipulating on screen controls. Once the user had the image more or less lined up the way they want it, instead of adding control points they could could click on locations in their image and drag them to the exact spot on the underlying map to which they want to register it.

This type of workflow would not be ideal for a highly repetitive process like georeferencing 100 floorplans for each floor of a high rise building, but it could be ideally simple and intuitive for a lot of one-off type of georegistering projects.

adamw


10,447 post(s)
#16-Jan-21 09:47

This is a good idea, we might do this in the future. Ie, get an image roughly to where it should be first by dropping it into a map and using interactive scale / shift / rotate tools. Then perhaps register it more precisely using the control points. We could extend our definition of the coordinate system to allow for local rotation (similarly to local scale / local offset), this way the first interactive step would only be altering the coordinate system and would not be losing any data to rounding.

pslinder1
228 post(s)
#16-Jan-21 17:57

This is a good idea, we might do this in the future. Ie, get an image roughly to where it should be first by dropping it into a map and using interactive scale / shift / rotate tools. Then perhaps register it more precisely using the control points.

Exactly!

adamw


10,447 post(s)
#16-Jan-21 09:22

Revisiting this after 173.2.

We did (e). We are also going to do (a), (b), (c) and (f). Some of those are already done in the internal builds.

Regarding (d), for now we keep showing control points only in the Edit Control Points mode. If control points can be forced to be shown in any mode, we are afraid this might interfere with the editing - multiple overlays shown at the same moment in the same window can be overwhelming. We already sometimes have multiple overlays with preview + editing, maybe we shouldn't be allowing that either. In any case, we can now save control points to a drawing, so if you want to have them perma-visible, to be able to snap to them, and so on - you can.

For (g), maybe this could be a visual hint in the window which you could snap to. This could be practical for scenarios where all you need is a simple shift + scale + rotate and you are looking for control point entry speed because you are registering many components to different locations.

Thanks for a great post!

geozap
264 post(s)
#16-Jan-21 14:56

What I have in mind (actually used a similar tool years ago in MicroStation Descartes) is useful when you want to place many points. So after having placed a couple you can have a button to instantly move the screen to the expected target position of the next point you want. Not to snap to it, but to pan and zoom fast. Actualy, in most cases you wouldn't want to snap and you might not mind the indicated position being even approximate, because if the transformation was already ok you wouldn't be placing more control points. Such a tool is very useful when one wants to register many pictures with many points.

geozap
264 post(s)
#16-Jan-21 15:32

An addition: Since a tool like this would be useful to move fast rather than snapping, when having triangulation to do, a case without same transformation parameters for all source map, the approximate location of the target point could be calculated by using another method, such as affine 1st order.

adamw


10,447 post(s)
#18-Jan-21 15:01

Well, yes, we'd use affine. Higher-order numeric methods could produce results that are much less intuitive, and triangulation would only produce results in the convex hull of already entered control points. Using affine would be safest and the most predictable.

We'll think about adding a button to pan to the expected location / show it on the screen (if we don't show it and just put it into the center of the screen, telling where it is exactly could be problematic).

mdsumner


4,260 post(s)
#01-Jan-21 20:38

Awesome, I'd add a request for indicator lines on each axis, with callout vertical/horizontal lines - these are handy to pick out axis ticks. (ah yes look at this old email)


https://github.com/mdsumner

adamw


10,447 post(s)
#16-Jan-21 09:56

This could be useful, agree. Perhaps something that could be turned on and off like snap.

Thanks for the request.

tonyw
736 post(s)
#01-Jan-21 00:23

I'm looking forward to getting proficient with the registration pane. I had a quick try already. One thing I notice with Manifold is I often get the opposite result than intended, no different here with the registration pane. I ended up altering my shapefile with known good points. Other places I end up with opposite results are in the Overlay transformations in M9 and clip or intersect in M8. It might have to do with semantics, in the case of the new registration pane in M9, what is a target (image) to be georeferenced and what is the source of known points? The use of the two terms in M9 for target and source, for me, are reversed and after re-reading the manual, I now understand that in M9 the target is the source of the known points and the source is objective (target) to be georegistered. I come at Manifold as a non-programmer and to programmers source and target may have specific meanings. I'm a native English speaker with a relatively good command of the language and its semantics. I wonder if others run into this mirror image of semantics in Manifold? It results in trial and error on my part as I use Manifold infrequently and not in a production environment.

I can understand boiling down a description to one word is a necessity (e.g. "source" and "target") rather than repeating "the image that you just imported" and "the component with the known good points". Using letters A and B just adds another layer to the mental juggling that A= "the image that you just imported" and B = "the component with the known good points".

I'm sure deciding on the terminology to use in Manifold is not a trivial matter. So just a comment.

A question for the surveyors out there, what terminology do you use when you are tieing in a survey to known coordinates?

Dimitri


7,413 post(s)
#01-Jan-21 06:57

It's not a programmer thing but a GIS inertia thing... using "source" for the item to be georegistered and "target" for the known-good destination is the same nomenclature used by ESRI.

mdsumner


4,260 post(s)
#01-Jan-21 20:35

I think source and target are perfect when you refer to a before transformation and after transformation task, it's not really a source in that sense here because we effectively ignore all information (apart from relative position). So that's interesting ... I wasn't aware that esri uses those terms that way, I use them routinely in other workflows re before/after.


https://github.com/mdsumner

tjhb
10,094 post(s)
#01-Jan-21 20:56

I don't think it matters much. Source and target are familiar. Perhaps only "target" makes perfect sense here, but it does. [Well, "source" makes sense too. We are mapping the source to the target, unknown points to known points.]

I loved the video, thanks Dimitri.

The only thing I would change--this drove me nuts in 8 as well--is the naming scheme for control points. The words "Control Point" add nothing of value. We know that, thank you! All they do is make it slightly harder to rename, that is, to give control points descriptive names, usually identifying features. (This only becomes really useful given save and load, export and import, but that is coming.)

I would suggest a better default naming scheme would be a sequential number alone.

The first name should be 1.

P.s. being able to preview the result of registration is worth millions. A fantastic feature, makes all the difference.

tonyw
736 post(s)
#02-Jan-21 00:38

I agree, once I understood the Register pane in M9 the terminology doesn't matter. I work in a back and forth manner when georegisterng. With the scanned image and the known good layer side-by-side I'd start hunting for common points. When I find and confirm one I would put down one control point on each of the source and target. Then look for the next common point. When I find the next common point, it doesn't matter which layer has the focus, the one currently with focus gets the control point, then the other layer gets the corresponding control point.

The preview and the non-destructive registering in M9 are great. Preview is a great new feature, it makes georegistration interactive which is what I was looking for.

Some suggestions to spur discussion

1. In M8 if I recall, when placing a new control point I could hold the mouse down (I use a stylus, same effect), move and position the control point and the control point would be deposited only when I let the mouse button up (=lift the stylus). In this way I could place the control point with better accuracy. In M9 it seems the control point is placed even when the left mouse button is still depressed. Someone using a mouse, what is the behaviour for you? I know in M9 I can delete a point and try again, but being able to move the control point for fine tuning then letting the mouse button/lift the stylus worked great in M8.

2. Agree with the comment that the word "Control " in the control point label isn't needed. I couldn't find a way to turn off control point labels as the label covered a location where I was going to place the next control point. Being able to turn off the labels, as in M8, would work for me. (I realize now I could have zoomed in until the next control point location was uncovered).

adamw


10,447 post(s)
#16-Jan-21 10:15

Regarding (1), we'll allow repositioning an already placed control point. This is already possible for control points in a target window, we'll allow doing this for control points in a source window as well. We'll also allow editing the coordinates directly.

For what it is worth, on the use of source vs target. The notion is that the registration process transforms (warps) data from its original source location to a desired target location. So, 'source' is where the component currently is (unregistered, or registered to a wrong place), and 'target' is where it should go.

Dimitri


7,413 post(s)
#13-Jan-21 12:38

The words "Control Point" add nothingof value. We know that, thank you!

Agreed, but that's easy to change. Should have thought of this earlier... You can change that using localization.

If the naming scheme using "Control Point" is too long, just change this line in default.ui.txt:

ValueRegisterCoord=Control Point

to something like

ValueRegisterCoord=CP

Save the file to a ui name, like ui.cp.txt and place it in the same folder as the manifold.exe file you're launching. In Tools - Options you can choose the new cp option for a language, restart, and then the control points will be named by default using that string.

Example (georegistering a historic map):

Attachments:
cp.png

tjhb
10,094 post(s)
#13-Jan-21 17:53

Nice Dimitri! Thanks very much.

adamw


10,447 post(s)
#16-Jan-21 10:07

We agree about the names being too long. We shortened them in 173.2 to 'P XXX' (and started with 'P 1').

Just in case, the 'P' prefix can be edited in the localization file. (I wouldn't recommend setting it to an empty string though. Empty strings are fairly frequently interpreted as 'use the default value', so even if an empty string will indeed reduce produced names to just the numeric part at the moment - I didn't check, but it might - this might stop working in the future.)

lionel

995 post(s)
#06-Jan-21 19:22

Shear transform not supported


Book about Science , cosmological model , Interstellar travels

Boyle surface fr ,en

adamw


10,447 post(s)
#16-Jan-21 10:20

Yes, you can perform shear using registration with order 1 (affine). Need 3 pairs of control points.

pjrs33 post(s)
#12-Jan-21 22:09

What a good start to the New Year. Just when I need it to develop a historical map, Georegistration arrives and so does a good explanatory video. The Norwich 1500 (the UK city) map is now in place and I can get on with placing features. Thanks to all. I'll leave you pros to debate the finer points - I'm more than pleased.

enri10 post(s)
#24-Mar-21 15:33

Can I Register vector data which is made from the Satelite Image after the image datasource is registered?

I mean I've traced the satelite image and got a vector dataset, and then I found the satelite image need to be registered more precisely . But if I register the image, the vector data would not be compatible with the image. So can I register both of them at the same time?

Thanks!

Dimitri


7,413 post(s)
#25-Mar-21 04:54

Post this in a new thread... it's too hard to find your question buried in an old thread where there's no subject to tell people what your question is about.

And yes, you can do both at the same time, assuming you had the forethought to save the control points you used to better register the satellite image. There's a similar video, about georeferencing many CAD layers at the same time.

Manifold User Community Use Agreement Copyright (C) 2007-2021 Manifold Software Limited. All rights reserved.