JET_Affinity

At least so far as CS6 ('cause I won't rent business-critical software), Illustrator still can't actually crop a raster image; it can only mask it.
For the benefit of those who may not have "been there," Illustrator trailed versions (i.e., years) behind its historic nemesis, FreeHand, in all these areas:
Editing in Preview Mode. (In other words, simply being able to edit paths with their stroke and fill attributes showing.)
Compound Paths. (Making a path with a hole in it.)
Clipping Paths. (Ex: filling outlined text with a raster image.)
Performing alignment and distribution on Anchor Points (nodes). (Still sub-par compared to FreeHand because Illustrator's insistence on two separate primary selection tools effectively prevents it from "knowing the difference" between a path being selected at the object level, as opposed to merely having all its nodes selected.)
Page 2. Egads! What a concept! The garment-rending, sackcloth-and-ashes outrage from Illustrator devotees (who had practically zero experience with FreeHand or any other drawing program) anytime the need for multiple pages was even mentioned, was just laughable. Why, it was going to be the end of the world; the coming of the apocalypse; illustrators everywhere would be committing hara-kiri!
Predictably, many of those same users now no doubt couldn't live without it. Many probably think Adobe invented the idea. And Illustrator's treatment of it is still cumbersome compared to FreeHand's more straightforward interface.
Basic Math Operators in Value Fields. Another one still inferior to FreeHand's. Illustrator can still only manage a single type of operator in an expression (i.e., multiplication/division or addition/subtraction, but not both).
User-Defined Arrowheads. And man, what a hack job of an interface!
Converging Perspective Grids. Adobe had to acquire FreeHand to copy this one.
That's just off the top of my memory. I could go on.
I've said it here before, and I'll say it as long as it takes: Simple market share no more correlates to functional superiority in drawing software than it does in, say,  motorcycles (my other passion). Illustrator is not the program to emulate. I've seen evidence sufficient to convince me the spunky Affinity Team probably gets that. Some feature requesters...I'm not so sure.
And yeah, time is of the essence in the current window of opportunity opened by Adobe's Customer-alienating licensing change. But compared to the pace of Illustrator's development over decades, development progress of the Affinity line is lightspeed.
JET
 
 
 

Just for accuracy: Node handles do constrain to horizontal, vertical, and other grid angles (not lengths) when pressing Shift in 1.6.1.
No, the full functionality alluded to in the foregoing rants is not there, but it's not as dire as some make out.
I'm a technical illustrator, too, and for my own purposes, I went to the trouble of hacking out a couple of simple Javascripts to allow me to replicate handle lengths and angles between AnchorPoints in Illustrator. So I get the request. But it's not like the full desired functionality is present in even that ostensibly "leading" program.
The Affinity Team is doing great. I, for one, very much appreciate its members making time for direct involvement with Customers in this forum. And while I can be as passionate as anyone about facilitating technical drawing in mainstream general-purpose illustration programs, and though I will also push for "everything I can get" in that regard, I do hope my passion is not thought of as disgruntlement.
Right now, the Affinity line is the most promising platform poised to bring mainstream vector drawing out of its decades of lethargy. And the fact that there will soon be a dedicated Publisher to complete the static graphics triad is huge. Show me another single-source hopeful with all three key elements new from the ground up.
I love asking Illustrator, Canvas, Draw, and even Technical Designer users if they can key a trig function into their pet programs' value fields.

JET

Yes, but merely constraining angles to a desired set of axonometric axes can be accomplished in any program that provides a user-defined constraint angles feature (ex: the Constrain Guides sub-feature of Illustrator Smart Guides, and similar features in CorelDraw and ACD Canvas) without need for a page-spanning grid.
What's always missing in those features, though, is any provision for assigning correctly-proportional ruler scales along the constrain angles. That's why I said in my previous post "If grids are to serve as the rulers...":

I'm glad you're looking into that, because there has to be some provision for specifying properly foreshortened and accurate measures parallel to the three axis directions from any snapable point in the drawing.
One possible treatment might be a radio button set in the Transform palette labeled "Axis Scale" versus "Page Scale." That would at least provide a substitute for what other mainstream drawing programs are missing in their user-defined constrain angles. Moreover, it would enable entering measures in terms of true-measure values.
But even that does not emulate the direct intuitiveness or elegance of even a pre-computer drafting machine. Axonometric drawing, by definition, is all about making correctly-proportioned direct measures along three coordinate system axes (i.e.; each axis must have its own scale factor, and those scale factors must be correctly proportioned to each other), and zeroing those measures from elements of the drawing, not from increments of a grid. And performing such measurements should not require looking away from the drawing to a palette.
That's why I said that if grids are the only provision to serve as those on-page rulers, then the intersection of the grids needs to be able to be instantly and fluidly zeroed to any point wherever a mousedown occurs, just as the scale head of a physical track drafter effectively "moves" to the point of interest in the drawing and allows the illustrator to perform a measure from there without having to look away from the drawing.
The closest emulation I've yet seen of the kind of fluidity I envision is Lazy Nezumi Pro (so close, yet so far). In its isometric rulers preset, three rulers appear at and follow the cursor. Unfortunately, when set to other axonometric angles, those three rulers do not currently display proportional scales; all three still show the same scale. (I anticipate this changing, since LNP's converging perspective rulers do display proportional scales.) Plus, being an application-independent "overlay" seems to limit its functionality for vector drawing because (among other things) it is unaware of the program's zoom. And though tick marks were just recently added to the elliptical rulers (thereby allowing them to serve as elliptical protractors--something essential to serious axonometric drawing), the increments are not yet snapable.
But the interface concept is quite sound and elegant (and not unlike similar cursor-following interfaces of high-end 3D modeling applications). A similar treatment actually built into a 2D drawing program would not have those limitations.
For example, imaging drawing with the Pen Tool in its Straight Line mode:
The Axes feature is turned on. Three light-colored axonometric ruler guides appear, with their origin under the cursor. There are correctly-proportional tick marks along each of the three rulers. This set of guides always follows the cursor during mouseup, while the cursor responds to all the normal snapping candidates.
Upon mousedown, the rulers stay put.  The user drags along one of the axis guides. If he holds a modifier key, the cursor snaps to the tick marks of that axis. If he releases the modifier key, the angle constraint is still active, but the tick mark snapping is not. Either way, though, a distance readout (accurate to 4 decimals, please) continually appears next to the cursor.
That allows the illustrator to draw quickly with reasonable precision without having to look away from his drawing and toward a transform palette. But the transform palette (assuming its Axis Scale checkbox is on) can still be used to manually enter exact length.
Given such an interface, a page-spanning grid would not even be necessary. Sure, it would be useful when one wants to automatically "project" side views drawn "in the flat" onto the axonometric planes, and that's fine. But most of my drawing would be done with the grid display turned off. The whole purpose of axonometric methods is to allow the illustrator to intuitively draw directly into a mechanically-correct 3D orthographic perspective without having to draft side views first.
Oh, I'm all for some fresh innovation rather than just conventional wisdom. I look forward to seeing what you have in mind for the new feature. I just hope it's not too "locked in" to be open to some user feedback in terms of the implementation.
JET
 

Congratz on this great achievement! May this be the beginning of AD and AP gaining a significant share of the graphic design market and ultimately top all overpriced competition.

Regarding the axonometric grids and ruler origin reset:
Hopefully before these two related sneak peek features reach a customer beta stage (in which feature schema and behavior is mostly already committed and focus is mostly just on bug testing), I want to throw this out, so I can sleep at night:
Having been doing isometric drawing since the days of drawing "on the board" before desktop computers, I dare say you won't find anyone more enthusiastic about adding some geometric intelligence (other than just snapping) to the plane grids (more akin to DrawPlus). Such grids are a great way to introduce commercial illustrators without prior experience to axonometric drawing.
So don't think it contradictory when I say this: In all those decades, frankly, I have never met a fellow serious axonometric illustrator who is highly dependent upon grids; neither before the advent of graphics software nor since. Here's why:
There is a fundamental concept which the trivial "isometric grid" features in mainstream drawing software typically gets completely "backward":
As usually implemented, grids make your drawing conform to the grids, when the grids should be adapting to the drawing.
Grids tend to force your drawing to conform to the increments of the fixed grid. That's fine for "fantasy" drawing like, for example, bird's eye view game artwork wherein the actual dimensions and spacing of whatever "boxy" shaped things you are drawing are entirely up to you. But in real-world technical drawing, it's not about just drawing conveniently "boxy" things, and it's not about making your drawing measures conform to a fixed grid; it's about having a set of freely moveable and correctly proportioned angled rulers which enable you to make correctly-scaled measures from any point in your drawing.
In pre-computer days, the only time you saw a tech illustrator using a grid was when he was away from his drawing board (or when his drawing board was not equipped with a track drafter). Newbie illustrators would sometimes use a printed axonometric grid under a sheet of tracing paper. And guess what: He would be constantly moving the grid around under his drawing sheet.
A technical illustrator is not the least bit concerned with measures incremented from any page origin. He's constantly gliding his properly-angled rulers to make measures from pre-existing points in his drawing.
If grids are to serve as the rulers for axonometric drawing, they need to be able to act like rulers and freely follow the cursor, not be stuck to any page origin. The origin of the grids (the intersection point of the three planes) needs to be able to snap to any snapping candidates in the artwork, completely free from interference from a page layout grid.
This is essentially why no grid-based approach has ever really matched the quick, easy, intuitive fluidity of a physical drawing table equipped with a mechanical track drafter. The closest software emulations of the fluidity of the physical tools metaphor are not grids, but three proportional rulers (axes) which follow the cursor, as in some 3D modelers.
But axonometric drawing is, by definition, a 2D construction method historically performed on a 2D sheet of paper. So there's no reason a similar interface could not be provided in a general-purpose 2D illustration software, based on 2D geometry.
JET

Yes. I've coded up alignment for the Node tool. So, you can use the Alignment panel to align the selected nodes in the Node tool, as you would align layers/objects in the Move tool.
 
I've also prototyped some widgets for doing snapping alignment, as an embellishment to the selection box.
 
NodeAlignment.mov
 
NodeAlignmentWidget.mov
 

Regarding the axonometric grids and ruler origin reset:
Hopefully before these two related sneak peek features reach a customer beta stage (in which feature schema and behavior is mostly already committed and focus is mostly just on bug testing), I want to throw this out, so I can sleep at night:
Having been doing isometric drawing since the days of drawing "on the board" before desktop computers, I dare say you won't find anyone more enthusiastic about adding some geometric intelligence (other than just snapping) to the plane grids (more akin to DrawPlus). Such grids are a great way to introduce commercial illustrators without prior experience to axonometric drawing.
So don't think it contradictory when I say this: In all those decades, frankly, I have never met a fellow serious axonometric illustrator who is highly dependent upon grids; neither before the advent of graphics software nor since. Here's why:
There is a fundamental concept which the trivial "isometric grid" features in mainstream drawing software typically gets completely "backward":
As usually implemented, grids make your drawing conform to the grids, when the grids should be adapting to the drawing.
Grids tend to force your drawing to conform to the increments of the fixed grid. That's fine for "fantasy" drawing like, for example, bird's eye view game artwork wherein the actual dimensions and spacing of whatever "boxy" shaped things you are drawing are entirely up to you. But in real-world technical drawing, it's not about just drawing conveniently "boxy" things, and it's not about making your drawing measures conform to a fixed grid; it's about having a set of freely moveable and correctly proportioned angled rulers which enable you to make correctly-scaled measures from any point in your drawing.
In pre-computer days, the only time you saw a tech illustrator using a grid was when he was away from his drawing board (or when his drawing board was not equipped with a track drafter). Newbie illustrators would sometimes use a printed axonometric grid under a sheet of tracing paper. And guess what: He would be constantly moving the grid around under his drawing sheet.
A technical illustrator is not the least bit concerned with measures incremented from any page origin. He's constantly gliding his properly-angled rulers to make measures from pre-existing points in his drawing.
If grids are to serve as the rulers for axonometric drawing, they need to be able to act like rulers and freely follow the cursor, not be stuck to any page origin. The origin of the grids (the intersection point of the three planes) needs to be able to snap to any snapping candidates in the artwork, completely free from interference from a page layout grid.
This is essentially why no grid-based approach has ever really matched the quick, easy, intuitive fluidity of a physical drawing table equipped with a mechanical track drafter. The closest software emulations of the fluidity of the physical tools metaphor are not grids, but three proportional rulers (axes) which follow the cursor, as in some 3D modelers.
But axonometric drawing is, by definition, a 2D construction method historically performed on a 2D sheet of paper. So there's no reason a similar interface could not be provided in a general-purpose 2D illustration software, based on 2D geometry.
JET

Regarding manipulation of nodes: I hope someone has their eyes on some of the innovative Bezier handling previewed of (and now present in) in the just-released FontLab Studio VI.
Others may remember that much of FreeHand's superior path drawing and selection interface stemmed from its progenitor, Altsys Fontographer. To my hopeful eye, this may signify at least potential for a long-needed renaissance in 2D Bezier-based vector drawing.
JET

Nice, Ben.
I hope everyone knows that the sensible, useable polygon behavior of this is absent in the Lasso Tool in Adobe Illustrator (even though its raster version has long been present in Photoshop).
However, does this not work if the path(s) are not already selected? And it's not limited to one path, correct?
JET
 

BrightBold,
No offense, but almost everyone says that about his/her most desired feature. As just one individual example, another "me, too" autotrace feature is not even on my list of desired features. I almost never use one, there are plenty of them out there (even free ones), and they all do pretty much the same thing (which I expect is unlikely to change unless and until an affordable one acquires some measure of shape recognition intelligence).
That doesn't make me "right" or you "wrong," but as has been explained, it's on the planned features list. But priorities are up to the developers. I would imagine that sometimes features are inter-dependent, and have to be developed in sequence.
JET

BrightBold,
No offense, but almost everyone says that about his/her most desired feature. As just one individual example, another "me, too" autotrace feature is not even on my list of desired features. I almost never use one, there are plenty of them out there (even free ones), and they all do pretty much the same thing (which I expect is unlikely to change unless and until an affordable one acquires some measure of shape recognition intelligence).
That doesn't make me "right" or you "wrong," but as has been explained, it's on the planned features list. But priorities are up to the developers. I would imagine that sometimes features are inter-dependent, and have to be developed in sequence.
JET

Halftoning effectively blurs the edges of line art.
A printer (imagesetter) has a fixed resolution. All it actually print is printer spots of the same size. Printer spots are the actual hardware resolution of the imagesetter (typically 3000 or more spots per inch).
Halftone dots are made up of printer spots..Each dot in a halftone is a collection of printer spots, trying to simulate a circle..The number of different circle sizes possible is therefore determined by the number of printer spots available to simulate them. Divide the number of available printer spots (SPI) by the halftone ruling (LPI), and that's the theoretical number of different-size halftone dots (levels of grey) the device can print. That's why you always get more banding from, say, a 600 SPI laser printer than you do from a 3000 SPI imagesetter.
Everything in a greyscale image gets halftoned. That means the raster is printed as halftone dots, at the line ruling of the halftone screen (typically 150 lines per inch). It also becomes effectively anti-aliased by the halftoning.process.That's why black text that is part of a raster image looks fuzzy compared to black vector text stacked in front of a raster image.
1 bit raster objects do not get halftoned at all. They are simply "filled in" with tiny printer spots. So it's common practice to, for example, create or scan line art (think of the inking of a comic book illustration) as 1-bit rasters at something like 1200 PPI, which overlay grayscale or full color raster images. The color artwork prints as 1/150th inch halftone dots. But the 1-bit raster actually prints as 1/1200th-inch squares, giving a crisp, sharp-edged, aliased (not anti-aliased) appearance.
You can sort of think of 1-bit color depth as the "vector" version of raster imaging in that exactly what you've "drawn" simply gets "filled in" with the tiniest printer spots of the given output device. Take a look at this PDF: Zoom into it as far as you can. Tell me if you think it is raster image or a vector line.
JET

Uses for knowing the area (or length) of a path would be very user-specific. Max may want to perform scaling based on area or path length, either of which, as Ben mentioned, is a simple matter of uniform scaling which can already be keyed as an expression into the dimension fields. I, on the other hand, may want to know the area or path length in order to convert the value to another unit of measure (e.g., acres or miles) in order to paste it into a dimension callout on a land plot drawing. Or any number of other things.
So there's no need for any elaborate re-work of the Transform palette for this. Object attributes like area and length just need to be visible in any sensibly unobtrusive location of the interface so the user can use them as values in whatever specific calculation needed.
Sure, the Transform palette is one logical place to display dimensional object attributes, but it's not the only place to do it. Consider: There are other kinds of object-level attributes also very useful to know (count of nodes, open or closed, etc.) which are not dimensional, and would not be used as transformations.
Such specific uses are not things every user would do every day in the same way. The user just needs to know the values in order to use them as fits the needs. They could appear in an attributes pane, or in a contextual cursor menu, for example. Let's trust the developers to know how to best integrate such things into the established interface schema so as to preserve the overall elegance.
JET

Re: Area:

Of course, visual appearance of "same size" is not just a simple matter of area (as any type designer knows), but it's at least a starting point for that, and useful for many other common things.

Personally, I've long maintained that every serious drawing program should provide path length and area as visible attributes (as, for example, ACD Canvas has for as long as I remember). If they were given, one could devise whatever calculation needed for use-specific purposes (especially given Affinity's more capable value fields).

The object model for Illustrator's JavaScript implementation includes area as a path property. So it's fairly trivial to write a JavaScript that will do whatever calculation you want that includes area as a factor. (I've used it in a few of my own AI Javascripts.) There are a couple of caveats; for example, the value shown is a simple sum of subpaths in a compound paths, regardless of winding order.

AI used to only provide path length (among other object values) in its hidden programmer's dialog which was accessed by an "Easter egg" keyboard shortcut. After  a period of user demand, path length was eventually exposed in its Document Info palette (a frankly rather half-baked grab-bag feature), when the Objects and Selection Only palette options are on.
JET

Halftoning effectively blurs the edges of line art.
A printer (imagesetter) has a fixed resolution. All it actually print is printer spots of the same size. Printer spots are the actual hardware resolution of the imagesetter (typically 3000 or more spots per inch).
Halftone dots are made up of printer spots..Each dot in a halftone is a collection of printer spots, trying to simulate a circle..The number of different circle sizes possible is therefore determined by the number of printer spots available to simulate them. Divide the number of available printer spots (SPI) by the halftone ruling (LPI), and that's the theoretical number of different-size halftone dots (levels of grey) the device can print. That's why you always get more banding from, say, a 600 SPI laser printer than you do from a 3000 SPI imagesetter.
Everything in a greyscale image gets halftoned. That means the raster is printed as halftone dots, at the line ruling of the halftone screen (typically 150 lines per inch). It also becomes effectively anti-aliased by the halftoning.process.That's why black text that is part of a raster image looks fuzzy compared to black vector text stacked in front of a raster image.
1 bit raster objects do not get halftoned at all. They are simply "filled in" with tiny printer spots. So it's common practice to, for example, create or scan line art (think of the inking of a comic book illustration) as 1-bit rasters at something like 1200 PPI, which overlay grayscale or full color raster images. The color artwork prints as 1/150th inch halftone dots. But the 1-bit raster actually prints as 1/1200th-inch squares, giving a crisp, sharp-edged, aliased (not anti-aliased) appearance.
You can sort of think of 1-bit color depth as the "vector" version of raster imaging in that exactly what you've "drawn" simply gets "filled in" with the tiniest printer spots of the given output device. Take a look at this PDF: Zoom into it as far as you can. Tell me if you think it is raster image or a vector line.
JET

Since you mention that...
One of the most useful "shape tools" that would go hand-in-hand with the axo grids you're working on is a "threads" tools. For example, the one in Corel Technical Designer lets you simply drag to fill an ellipse with automatically trimmed half-ellipses to create a threaded hole. Huge time-saver.
Related: As I mentioned in one of the threads requesting a spiral too, ordinary "me, too" spiral tools are everywhere. But in both technical and general illustration, it's arguable that the need to draw a coil is actually more commonly needed than just yet another flat  "spiral." One of the oft-repeated threads in drawing software forums is "How do I draw a helix." And tech Illustrators need to draw springs just as often as they need to draw cogs.
Springs are tedious to draw manually. It's my most common use for the "path stretching" features ("Rubber Band Mode" in Corel Draw, or the "Reshape Tool" in Illustrator). But I see no reason why a live Spiral Shape tool shouldn't:
Provide control for both uniformly and progressively-spaced coils. Allow the spacing of uniform coils to go all the way to zero, so that the path coils effectively lay exactly on top of each other. Provide endpoint handles that can allow the spiral to be "stretched" so as to serve as the centerline of a spring. By way of example, here's a screenshot of my isometric springs "library" in Illustrator. Each of these was tediously derived by:
Drawing an ellipse. Cutting the ellipse at one of its nodes. Using the Reshape Tool to "stretch" the cut ellipse into a single "spiral" coil. Storing the coil in the "side tile" of a Pattern Brush. Drawing the end coils separately and storing them in the "end tiles." Repeating that process for every 5° increment about an isometric ellipse. You can imagine the tedium involved. But once done, each of the Brushes can be used to instantly create a spring of any diameter and any length. (And actually, of any bend, too; it's not just for tech illustration. I could use any one of the Brushes to draw a Slinky toy, for example.)

So in the above, I've used in non-obvious ways several features which many users consider the "high end" differentiators between ostensibly "professional" Illustrator and its competitors. But they're not really so "high end"; they're just needlessly cumbersome and not very well integrated.
I've done the same thing to create libraries of vector Brushes (or combinations of Brushes) to semi-automate drawing hex bolts, wires with terminals, wire rope, various kinds of chains, and more. Again, you can imagine the hours.

 
But I don't build such things in Illustrator anymore because I'm not going to continue to invest the effort in a program that I would have to rent (and which would thereby hold my own working files hostage). So I also do not share them, because I'm not really interested in promoting the use of Illustrator for the same reason.
Along similar lines, I've explored the related features in enough drawing programs to be convinced that such things and more could be accomplished more intuitively and more powerfully by a better-integrated set of more straightforward vector-based features, including:
Path Ends (not just arrowheads) Path Strokes (repeating or stretched) Symbols (as ends or repeating along strokes) Graphic Styles Blends As they say, "The devil is in the details." It's all about thoroughly and thoughtfully integrating the functionality between the features. For just one example, both Illustrator's Brushes and its Symbols are woefully debilitated by failure to abide by the option in the program's Transform palette to disallow scaling of stroke weights in the base artwork.
It's arguable that the piecemeal and standalone nature of such features in the very old programs is a consequence of their being added one at a time. I don't buy that, but Affinity's being new from the ground up should help avoid that "random grab bag" of functionality feeling of the long-in-the-tooth competitors.
Anyway, that's what I hope to see in Affinity.
JET

I've long maintained it would be less tedious and more accurate to simply provide a set of straightforward commands or buttons which allow the user to convert any ordinary path into a selection marquee or a cutting path (complete with appropriate contact-sensitive and select/subselect options). That way, any path drawing tool could be used, with all its accuracy advantages, instead of the conventional separate screen pixel-based marquee selection tool that:
Makes it far too tedious to weave around the desired selection in tight circumstances (a very frequent problem when working with the disjointed polylines exported from CAD programs). Prevents being able to zoom in or out while making a selection. Wreaks havoc when the "lasso" comes into proximity of the screen edges, causing you to have to start all over. I see no reason for separate functionally-limited "lasso" selection and "path cutting" tools, when the whole set of drawing tools could be used to perform both with more versatility.
Consider how ironic it is that conventional-wisdom vector drawing programs always resort to a clumsy screen-pixel tool for a so-called "lasso" tool, while Photoshop and similar raster imaging programs let you draw an accurate vector-based path and then convert it to a selection.
JET

Ben,

I apologize again for the length. I just want to be clear and thorough at the sacrifice of brevity, for the benefit of other interested forum participants.

My convention for XYZ labels of the iso axes is not entirely arbitrary. I simply label the vertical axis Y and the rightward axis X just to be generally consistent with the page rulers in 2D drawing and page assembly programs. This is also in keeping with the Cartesian orientation of the two scales attached to a drafting machine head. So that simply leaves the Z label for the leftward axis.
But various programs (both 2D and 3D) do it differently, and it's not really onerous to adapt. Regardless of the specific orientation, I much prefer a straightforward X, Y, Z labeling as opposed to ambiguous terms sometimes used ostensibly in the name of "intuition" (Left, Right, Top; Front, Side, Top, etc.), even for general illustration.

Yes. A complete solution needs to provide for two general rotational tasks. More simply put: establishing the coordinate system itself and "off-axis" rotations of object edges which are not parallel to the coordinate system.
If you provide for that much, you have at least simulated the process of drawing isometrically "on the board" in pre-computer days.
Off-axis rotation, however, includes two commonly needed situations, which I differentiate as "simple" versus "compound."
What I favor (and what you have already demonstrated in your last movie file) is a third capability which can actually make drawing "isometrically" in software radically more powerful than drawing on the board: The ability to re-orient the coordinate system on-the-fly. The point of this is that the coordinate system should not have to be, as you say, "a constant for your whole design."
So the operations needed for a complete solution are  fourfold:
Initial Setup of the Axonometric Coordinate System Piecewise Simple Off-Axis Rotation (within the current coordinate system) Piecewise Compound Off-Axis Rotation (within the current coordinate system) Re-Orientation of the Coordinate System for major sections of the drawing Those four operations can be thought of as corresponding to "stages" or "levels" of supportive functionality as far as current 2D drawing software goes. (They can actually be performed in any mainstream 2D Bezier-based drawing system if the user is well-versed in the principles, but they are seldom and weakly overtly or expressly supported (if at all) in most programs:
Stage 1. Providing Axo Grids: As already in Designer 1.6. This is significantly less functional than DrawPlus. But because the feature can automatically generate correct orthographic proportion between three perpendicular planes, even this is at least far better than:
Programs that merely provide rectangular page grids (e.g., Adobe Illustrator) Programs that merely provide a "faux isometric" 2:1 aspect-ratio grid (e.g., Xara Designer Pro) Programs that merely provide a basic isometric grid (e.g., Inkscape) Giving this a cube interface as in your last screen recording makes it more intuitive and "friendly" for any other general illustration purposes. But if the main point of my previous post (rotating about each axis) were implemented, just a few thoughtfully selected pre-defined defaults which can be selected from a menu (or even a set of icons) would be plenty for a starting point. A selection something like:
Iso Bird's Eye Iso Worm's Eye Dimetric Top Favoring Dimetric Bottom Favoring Dimetric Sides Favoring Trimetric Top Favoring Trimetric Bottom Favoring Trimetric Left Side Favoring Trimetric Right Side Favoring That is would be not at all limiting because even at the beginning of a drawing, nothing would prevent the illustrator from "fine tuning" any of those orientations by incrementally rotating the proxy cube about any of its the three axes.
Stage 2. Providing Interactive Off-Axis Simple Rotation: This was demonstrated by your rotating star video clip. This moves toward the functionality of DrawPlus, which otherwise in the realm of mainstream 2D Bezier-based drawing programs (at least to my knowledge) is only provided in much more expensive Corel Technical Designer. (I assume TechDesigner's Projected Axes feature could be fairly easily added to Draw, and should have been long ago.)
At first blush, (as evidenced early-on in this thread), this "live connection" between a piece of the artwork and one of the grid-defined "planes" appears to just be all about merely putting flat artwork onto the three perpendicular planes of a box (which, unfortunately, is all that some think isometric drawing is).
While that's certainly a great and useful feature for illustration in general, its greater implication for axonometric drawing is more subtle: To an isometric illustrator, it's not just about rotation on one of the planes. It also constitutes the most basic use of an elliptical protractor; finding correctly angled and scaled rotation about the axis which is not even on that plane.
Now, in truth, just the Pie option of Affinity's Ellipse Tool alone effectively constitutes a serviceable elliptical protractor if one knows how to use it as such. And by the same token, it's also fairly trivial to effectively "project" (distort) artwork drawn "in the flat" onto any axonometric plane by means of a "manual" rotation and scale transformation. But the "live" (interactive) aspect, of course, makes such things quicker and more intuitive, and adjustable, as software should. (Suppose, for example, you need to construct a set of wheel spokes or a radiator fan in correct geometry relative to your axis setup.)
Stage 3. Providing Interactive Off-Axis Compound Rotation: This is not provided by your rotating star video clip. When drawing "on the board," that unassuming little piece of plastic called an isometric protractor was not just used for finding drawn lengths and angles rotated about a single axis. It also serves (albeit in two steps) to construct compound rotations—rotations about two axes—which, as you noted, can define any rotation in 3D space.
An isometric protractor template goes even beyond that. A good one has printed along its major diameter an odd-looking, rather esoteric scale which provides something else very commonly needed in conjunction with off-axis compound rotations:  the correct "angle" (aspect ratio) of an ellipse which is "pierced" by the off-axis line.
This (so far) is the "stage" of functionality missing in the peeks at the developing interface. Stage 2 and Stage 3 combined correspond to what you referred to as "the rotation of a component in the design relative to your logical coordinate system" and what I refer to as "compound off-axis rotation." It corresponds to the purpose of the "Unit Sphere" in the mockup I sent you and of my Flash-based tool; an interactive "spherical protractor":


Stage 4. Providing for On-The-Fly Reorientation of the Coordinate System.  This is what your last movie file demonstrates. Its interface treatment, however, is the element about which I have the greatest misgivings so far, because a more intuitve treatment of this would make available and immediately intuitive a concept widely misunderstood by begining and experienced axonometric illustrators alike: The fact that isometric, dimetric, and trimetric are not three arbitrarily "disconnected" conventions. When done right, they are in fact entirely geometrically compatible and can therefore be used to advantage even within the same drawing.
This (given the main edit I suggested in my previous post) is what would really take the solution far beyond competing attempts. But thinking of it as "a constant for your whole design" is a step backward. Completely re-orienting the whole coordinate system for every single piecewise off-axis roation (stages 2 and 3) would be tediously cumbersome. But doing it for significant contiguous portions of the whole drawing (as in my motorcycle example) would be hugely advantageous.
Provision for what I've described as Stage 3 and a modification of Stage 4 would put this feature set in a class of its own, far ahead of any of the half-baked piecemeal features provided in any of the mainstream 2D drawing progams. One way to do that might be:
To Provide Stage 3 Functionality: Provide a Unit Sphere widget or tool. This would basically consists of the functional equivalent to three "Pie" ellipses (and a little bit of on-screen "cues" for interactivity and intuitiveness) which, when invoked, appears at the cursor location (if a tool), or on the page (if a widget, perhaps stored as one of the default Smart Shapes objects). In either case, it conforms to the current grids orientation (if any) when it appears. Otherwise it appears in isometric orientation.
Referring roughly (at least in principle) to my screenshot above, a handle circle would appear at the intersection of, say, the X axis and the "equator" ellipse. Its movement is constrained along the ellipses it encounters, and is followed by a "ghost" ellipse guide (the grey ellipse in my image) indicating the current "latitude," which also serves as a drag guide. Cursor readouts (two values) appear during the drag, indicating rotation about the constraining horizontal ellipse (i.e., "longitudinal" angle) and about the constraining vertical ellipse (i.e., "latitudinal" angle). Throughout the process, a radius line between the sphere's center and the cursor is displayed, along with an ellipse which is "pierced" by the radius. On mouseup, that radius becomes a normal single-segment path, and its "thrust ellipse" becomes a normal Pie Ellipse. A prompt appears in which to enter the true-measure length of the radius (and major diameter of the thrust ellipse).
To Improve Stage 4 Functionality: This would be just as explained in my previous post: Modify the interface so as to provide interactive and numeric rotation of the proxy cube about each of the three axes, instead of about just one axis and a "pitch" thumbwheel" as in your movie file.
Something I did not make clear in my previous post is that I would not expect re-orienting the overall gridset (the cube) to have to be so elaborate as to maintain object-specific "links" to the planes of the previous original cube orientation. That is, once I reached the steering stem in the motorcycle example, and decided to re-orient the axonometric coordinate system to more directly assist in construction of the front assembly, I would consider it fine if previously drawn objects were to become "flattened" regarding their "live" connection to the previous coordinate system. Again, the concept I'm trying to convey is that changing the overall coordinate system would be powerfully useful. But it should be used to facilitate drawing significant areas of the drawing, not to perform each and every piecewise rotation (which are typically many in real-world use). I'm not expecting this to act like 3D modeling, but like the long-established discipline of 2D axonometric drawing which is all about referencing the axes, for not just linear measures but rotations also; and not about freely twirling a "virtual trackball."
Something functionally equivalent to this approach would be vastly more intuitive in real-world use than having to mentally "step back" and imagine what combination of rotation about Y and overall elevation (values from two different manipulation metaphors) are needed to, for example, "Rotate the toy tank's muzzle 15 degrees about X and then rotate it and the turret 65 degrees about Y." Those are the terms in which the illustrator is thinking when drawing axonometrically. He is not thinking in terms of tilting and rotating the whole coordinate system for every piecewise off-axis rotation.
I'd like to suggest a detail modifications to the cube-based interface:
Unless it represents some other not yet explained functionality, I don't really see the value of the smaller inner cube. I would much rather see an ellipse displayed on each face of the large cube. One of the most common errors in axonometric illustration is disproportion between ellipses and axial measures. Without going into more detail here, It has to do with the issue that was commonly (and awkwardly) described as "isometric drawing" as opposed to "isometric projection." The error is just as easy to commit in software as it is on the board (if not more so).
Again, thanks for your work and attention to detail (and patience), and sorry for the length.
JET

The most telling detail of FreeHand's elegance was its selection and path manipulation interface. Illustrators who never used it just don't realize that its single selection tool did more—and did it more intuitively and efficiently— than any program with the now pandemic separate node selection tool. The insistence on two separate main selection tools is so ingrained due to Illustrator's market dominance that I fear most users will never know how much better it could be.
I knew FreeHand's demise was immanent as soon as Macromedia added the completely useless white pointer, just to appease Illustrator users. (It was literally just that. It wasn't until the very last version that the white pointer actually gained any ability in FreeHand that couldn't be done with the black pointer; and even that was a token detail, unworthy of a separate tool.)
Nonetheless, up to that point, FreeHand underwent marvelous advancements while under Macromedia's control. A huge one was its complete interface rebuild to an Inspector-based one. Everything you needed to know about the current selection was visible and settable in the efficiently designed Inspector. No drawing program's attempt at so called control panels and object attributes ribbons has come close to the efficiency of FreeHand's Inspector palette. Once again, Illustrator's schizophrenic attempt—which can't seem to figure out if it's a tool options bar or an object attributes bar or a commands bar—wins worst-of-class.
(Ironically, FileMaker Pro—a relational database management program, of all things—comes close with the comprehensive Inspector palette of its Layout Mode interface.)
My nostalgia is not rose-colored glasses, though. Even when FreeHand was in its hey day, I was quite vocally bemoaning the fact that the interfaces of all the Bezier drawing programs (including FreeHand) were actually more analogous to a mere "line up table" or "paste up table" (a glorified light table with a T-square, used for flat design, stripping film, etc.) than to a proper drawing table equipped with a Mutoh track drafter (used for illustration). It was so refreshing when FreeHand's full-blown Perspective Grids feature appeared. I wonder how few know the one in Adobe Illustrator is a direct copy of it; just one of those things for which Adobe gets the credit by merely acquiring it from elsewhere, and very belatedly adding it to Illustrator. And though I sung its praises when it first appeared in FreeHand, I was deeply disappointed that they had chosen to address converging perspective before parallel perspective, which I think is arguably more amenable to 2D drawing programs by its nature.
Still today, with few exceptions, almost everything in a "drawing" program's interface is tyrannically oriented toward the horizontal and vertical. When drawing, an Illustrator couldn't care less about the page edges. An illustrator is thinking in terms of the spatial orientation and angular view of the subject being drawn. When you really think about it, we are usually struggling against the conventional-wisdom features when trying to draw real-world things. After over thirty years of ostensibly "rocket fast" computer and software development, little has really changed from that in the 2D drawing genre.
That's why I find it so refreshing whenever I see a feature explicitly designed to support illustration in a mainstream Bezier drawing program, like Affinity's axonometric grids. Axonometric is a particular passion with me, because it's such an elegant system. It's so neat to watch your constructions just come together and fit perfectly together throughout the drawing with geometric accuracy equal to that of mechanical drafting. And especially today its application is far broader for commercial illustration than commonly assumed. I blame the misconception on decades of neglect, both in software and in general art classes.
By the way, I never had a FreeHand file fail to RIP. In the days of slower processors and early PostScript, too many users indiscriminately built their files using all kinds of willy-nilly, sloppy, convoluted constructs without a thought beyond on the monitor appearance, thereby effectively begging for output problems. Such problems occurred with all graphics applications. CorelDraw gained an undeserved bad reputation in that regard, largely just because it was so feature-rich. Too many users still do the same today, but the output systems have become much more forgiving in terms of error handling.
For one example, back then stray points (single-point paths) could cause output problems, and Adobe Illustrator (precisely because of its awkward selection interface) is the program most prone to inadvertently creating stray points.
JET

I've long maintained it would be less tedious and more accurate to simply provide a set of straightforward commands or buttons which allow the user to convert any ordinary path into a selection marquee or a cutting path (complete with appropriate contact-sensitive and select/subselect options). That way, any path drawing tool could be used, with all its accuracy advantages, instead of the conventional separate screen pixel-based marquee selection tool that:
Makes it far too tedious to weave around the desired selection in tight circumstances (a very frequent problem when working with the disjointed polylines exported from CAD programs). Prevents being able to zoom in or out while making a selection. Wreaks havoc when the "lasso" comes into proximity of the screen edges, causing you to have to start all over. I see no reason for separate functionally-limited "lasso" selection and "path cutting" tools, when the whole set of drawing tools could be used to perform both with more versatility.
Consider how ironic it is that conventional-wisdom vector drawing programs always resort to a clumsy screen-pixel tool for a so-called "lasso" tool, while Photoshop and similar raster imaging programs let you draw an accurate vector-based path and then convert it to a selection.
JET

A man after my own heart, Alfred.
JET

And for those who had no experience with FreeHand: This ability is just one of many things which FreeHand had years before Adobe Illustrator. And even when it finally appeared in Illustrator, FreeHand's treatment was still more powerful.
It's one of my favorite examples of how ill-conceived fundamental elements of Illustrator's general interface (in this case, the simple principle of "selection") cascades upward through all of its features. Without going into detail, it boils down to the fact that Illustrator's interface doesn't "know" the difference between a path's being selected as an object, as opposed to having all its nodes selected. This stems, in part, from the infernal insistence on two separate selection tools.
In this case, the result is that FreeHand could perform all the same alignments and distributions on nodes as on whole objects. For example, in FreeHand, you can select all the nodes of a given path—or of multiple paths—and align them. Try that in Illustrator, and you'll find that you must deselect at least one node.
Thus, my automatic refrain: When building a better drawing program, Illustrator is not the program to emulate.
(By the way, FreeHand could also perform alignment or distribution of multiple pages (yet another major feature which Illustrator was decades late in providing) in the same straightforward fashion. The tedium of doing that in Illustrator, at least through version CS6, is laughable by comparison.)
JET

Since you mention that...
One of the most useful "shape tools" that would go hand-in-hand with the axo grids you're working on is a "threads" tools. For example, the one in Corel Technical Designer lets you simply drag to fill an ellipse with automatically trimmed half-ellipses to create a threaded hole. Huge time-saver.
Related: As I mentioned in one of the threads requesting a spiral too, ordinary "me, too" spiral tools are everywhere. But in both technical and general illustration, it's arguable that the need to draw a coil is actually more commonly needed than just yet another flat  "spiral." One of the oft-repeated threads in drawing software forums is "How do I draw a helix." And tech Illustrators need to draw springs just as often as they need to draw cogs.
Springs are tedious to draw manually. It's my most common use for the "path stretching" features ("Rubber Band Mode" in Corel Draw, or the "Reshape Tool" in Illustrator). But I see no reason why a live Spiral Shape tool shouldn't:
Provide control for both uniformly and progressively-spaced coils. Allow the spacing of uniform coils to go all the way to zero, so that the path coils effectively lay exactly on top of each other. Provide endpoint handles that can allow the spiral to be "stretched" so as to serve as the centerline of a spring. By way of example, here's a screenshot of my isometric springs "library" in Illustrator. Each of these was tediously derived by:
Drawing an ellipse. Cutting the ellipse at one of its nodes. Using the Reshape Tool to "stretch" the cut ellipse into a single "spiral" coil. Storing the coil in the "side tile" of a Pattern Brush. Drawing the end coils separately and storing them in the "end tiles." Repeating that process for every 5° increment about an isometric ellipse. You can imagine the tedium involved. But once done, each of the Brushes can be used to instantly create a spring of any diameter and any length. (And actually, of any bend, too; it's not just for tech illustration. I could use any one of the Brushes to draw a Slinky toy, for example.)

So in the above, I've used in non-obvious ways several features which many users consider the "high end" differentiators between ostensibly "professional" Illustrator and its competitors. But they're not really so "high end"; they're just needlessly cumbersome and not very well integrated.
I've done the same thing to create libraries of vector Brushes (or combinations of Brushes) to semi-automate drawing hex bolts, wires with terminals, wire rope, various kinds of chains, and more. Again, you can imagine the hours.

 
But I don't build such things in Illustrator anymore because I'm not going to continue to invest the effort in a program that I would have to rent (and which would thereby hold my own working files hostage). So I also do not share them, because I'm not really interested in promoting the use of Illustrator for the same reason.
Along similar lines, I've explored the related features in enough drawing programs to be convinced that such things and more could be accomplished more intuitively and more powerfully by a better-integrated set of more straightforward vector-based features, including:
Path Ends (not just arrowheads) Path Strokes (repeating or stretched) Symbols (as ends or repeating along strokes) Graphic Styles Blends As they say, "The devil is in the details." It's all about thoroughly and thoughtfully integrating the functionality between the features. For just one example, both Illustrator's Brushes and its Symbols are woefully debilitated by failure to abide by the option in the program's Transform palette to disallow scaling of stroke weights in the base artwork.
It's arguable that the piecemeal and standalone nature of such features in the very old programs is a consequence of their being added one at a time. I don't buy that, but Affinity's being new from the ground up should help avoid that "random grab bag" of functionality feeling of the long-in-the-tooth competitors.
Anyway, that's what I hope to see in Affinity.
JET

Since you mention that...
One of the most useful "shape tools" that would go hand-in-hand with the axo grids you're working on is a "threads" tools. For example, the one in Corel Technical Designer lets you simply drag to fill an ellipse with automatically trimmed half-ellipses to create a threaded hole. Huge time-saver.
Related: As I mentioned in one of the threads requesting a spiral too, ordinary "me, too" spiral tools are everywhere. But in both technical and general illustration, it's arguable that the need to draw a coil is actually more commonly needed than just yet another flat  "spiral." One of the oft-repeated threads in drawing software forums is "How do I draw a helix." And tech Illustrators need to draw springs just as often as they need to draw cogs.
Springs are tedious to draw manually. It's my most common use for the "path stretching" features ("Rubber Band Mode" in Corel Draw, or the "Reshape Tool" in Illustrator). But I see no reason why a live Spiral Shape tool shouldn't:
Provide control for both uniformly and progressively-spaced coils. Allow the spacing of uniform coils to go all the way to zero, so that the path coils effectively lay exactly on top of each other. Provide endpoint handles that can allow the spiral to be "stretched" so as to serve as the centerline of a spring. By way of example, here's a screenshot of my isometric springs "library" in Illustrator. Each of these was tediously derived by:
Drawing an ellipse. Cutting the ellipse at one of its nodes. Using the Reshape Tool to "stretch" the cut ellipse into a single "spiral" coil. Storing the coil in the "side tile" of a Pattern Brush. Drawing the end coils separately and storing them in the "end tiles." Repeating that process for every 5° increment about an isometric ellipse. You can imagine the tedium involved. But once done, each of the Brushes can be used to instantly create a spring of any diameter and any length. (And actually, of any bend, too; it's not just for tech illustration. I could use any one of the Brushes to draw a Slinky toy, for example.)

So in the above, I've used in non-obvious ways several features which many users consider the "high end" differentiators between ostensibly "professional" Illustrator and its competitors. But they're not really so "high end"; they're just needlessly cumbersome and not very well integrated.
I've done the same thing to create libraries of vector Brushes (or combinations of Brushes) to semi-automate drawing hex bolts, wires with terminals, wire rope, various kinds of chains, and more. Again, you can imagine the hours.

 
But I don't build such things in Illustrator anymore because I'm not going to continue to invest the effort in a program that I would have to rent (and which would thereby hold my own working files hostage). So I also do not share them, because I'm not really interested in promoting the use of Illustrator for the same reason.
Along similar lines, I've explored the related features in enough drawing programs to be convinced that such things and more could be accomplished more intuitively and more powerfully by a better-integrated set of more straightforward vector-based features, including:
Path Ends (not just arrowheads) Path Strokes (repeating or stretched) Symbols (as ends or repeating along strokes) Graphic Styles Blends As they say, "The devil is in the details." It's all about thoroughly and thoughtfully integrating the functionality between the features. For just one example, both Illustrator's Brushes and its Symbols are woefully debilitated by failure to abide by the option in the program's Transform palette to disallow scaling of stroke weights in the base artwork.
It's arguable that the piecemeal and standalone nature of such features in the very old programs is a consequence of their being added one at a time. I don't buy that, but Affinity's being new from the ground up should help avoid that "random grab bag" of functionality feeling of the long-in-the-tooth competitors.
Anyway, that's what I hope to see in Affinity.
JET

But lest anyone under-appreciate the significance of this feature, here's a simple example of what I'm talking about in the context of your billboard question:



So you see, such a feature set does indeed facilitate "rotation on multiple planes".
Although this is an entirely 2D construction method, and although there is no "live" rotation connection between the default isometric and the custom dimetric grid presets, it's still quite practical in terms of expedience and its result is just as geometrically correct as if it had been generated by a 3D CAE program. This is what I meant in saying that isometric and dimetric (and trimetric) methods can actually be used together in the same drawing. They are not separate, unrelated arbitrary conventions. The grids feature will help you do that.
And much more. The above is fairly trivial. Axonometric drawing is not just about drawing "boxy" objects, or using "clever tricks" to "project" flat designs onto a plane, like the ubiquitous mockup of a cereal box which is all many illustrators with little exposure to isometric drawing mistakenly think it's about. The axonometric grids will be just as useful for constructing mechanically-correct parallel perspective drawings of objects of any shape, and the opportunity it represents to commercial illustrators is significant.
JET