AP - Curve thickness question

[Old Bruce]

Just now, Pšenda said:

It would be enough to make a screenshot with the Node Tool selected. I have nothing to try because I don't understand your picture - I don't know what points A - D are, I don't know what the left / right texts are. Is it one curve or several independent segments?

See Dark Clown's post from about an hour back regarding Left and Right.

A, B, C, and D are the rough locations of the four nodes.

[Pšenda]

2 minutes ago, Old Bruce said:

See Dark Clown's post from about an hour back regarding Left and Right.

A, B, C, and D are the rough locations of the four nodes.

Thank you for viewing when Node Tool is selected. Without marking the beginning and end of your curves, the image is completely vague, making it irrelevant to talk about the left and right sides of the curve. If you have a problem with this (display with the Node tool selected), we can end the discussion.

[Old Bruce]

A is the start. In both cases A is the start.

Try it yourself. I just wanted to point out that Left Right as substitutes for Inside Outside would have problems.

[R C-R]

2 hours ago, Old Bruce said:

A is the start. In both cases A is the start.

Try it yourself. I just wanted to point out that Left Right as substitutes for Inside Outside would have problems.

A is the start node & E is the end node in this Right is in or out?.afdesign file:

If I labeled everything correctly then right is both inside & outside & so is left, at least if I go by how the inside of that curve would fill with color.

I don't know about anybody else, but I would consider this a problem for that substitution.

[DarkClown]

13 hours ago, R C-R said:

If I labeled everything correctly then right is both inside & outside & so is left

Exactly ... And the align setting has the same probem with filled curves. That's why I'd prefere left and right instead of inside and outside for open curves. But when dragging nodes resulting in overlapping surfaces outside/inside behaviour is not the same as right/left! While positioning of inside/outside closed surfaces needs to change it remains the same when using right/left in an open curve

 

 

[Pšenda]

15 hours ago, R C-R said:

If I labeled everything correctly then right is both inside & outside & so is left

If the curve is open, there is no inside or outside, but only on the right and left.
The fact that you fill in the resulting curve, and thus even partially/virtually close it, has nothing to do with the problem of placing the stroke to the right or left.

Just in case someone doesn't understand, the explanation of the benefit of being able to place a stroke on the right or left (and not just centered as currently is possible) is that the width of the stroke could be changed without affecting its display.
If you are currently drawing an odd-width line (1, 3, 5, ... px) and you want to draw it exactly one / three / five pixels wide, you must place it exactly at the x.5 px position. If you require an accurate pixel rendering of an even-width line (2, 4, 6, ... px), then you must place it exactly at the x.0 px position. Which in itself is no problem - it just requires precise drawing work (and take advantage of snapping to a pixel grid). However, the problem occurs if, after creating a drawing, you decide - for example, after checking the output to the printer that the 1/3/5 px line needs to be extended to 2/4/6 px to be clearly visible. In this case, the whole design will fall apart because the lines will no longer be drawn in whole pixels, and you will have to adjust their positions (you will need to move them 0.5 pixels).
However, if open curves could be drawn with the stroke placed to the right or left (as well as inside/outside for closed shapes), then they would always be placed on whole pixels, and the stroke width 1/2/3/4/5/6 px would always be displayed on whole pixels. When changing the line thickness, there would be no need to change the position of the lines.

[Pšenda]

18 hours ago, Old Bruce said:

A is the start. In both cases A is the start.

In both cases, if point A is the start, then the right and left sides of the curve are described correctly in your images - which corresponds exactly to the DarkClown statement you respond to, and then I don't understand what should fall apart.

[DarkClown]

Just two little sidenotes:

1) The align functionality is not working on a closed curve once there ia a start- or endsymbol selected and the symbol is placed within the line? (it works, once the symbol is placed at the end of the line) ... really strange. From my perspective this should not depend on the alignment setting.

2) As well the "join" parameter is useless (I guess due to the implementation algorithm) once the closed curve is aligned "inside". Looks like the radius/bevel for the join parameter is based on the distance to the mathematical curve (which in this case is equal to 0) instead of the center of the inside aligned curve.

[R C-R]

3 hours ago, Pšenda said:

If the curve is open, there is no inside or outside, but only on the right and left.

Like I have said earlier, if you want the stroke alignment to be based on to the left, to the right, & along the centerline of the curve instead of inside, outside, & center alignments, this needs to be posted to the feature request section of the forum. Personally, I am fine with it as it is now -- I would not want the alignment to change if I reversed the direction of the curve, which I think this left/right option would do for both open & closed curves.

We also need to consider carefully how this would or should work with the 'quick shapes' (cogs, stars, arrows, & so on) since they do not have nodes & thus no defined start or end nodes. For them, inside & outside is well defined & intuitively obvious, but left & right is not since it depends on the start to end node direction.

Basically, I just want consistency in how the stroke alignment options work for all objects that can be stroked, regardless of the object type, or its stroke direction (if it has one). The only way I can see for that is as it is now.

[R C-R]

1 hour ago, DarkClown said:

1) The align functionality is not working on a closed curve once there ia a start- or endsymbol selected and the symbol is placed within the line? (it works, once the symbol is placed at the end of the line) ... really strange. From my perspective this should not depend on the alignment setting.

I do not understand what you mean by the start- or endsymbol. What specifically are you placing "within the line" & what line do you mean (path, stroke, something else)?

1 hour ago, DarkClown said:

2) As well the "join" parameter is useless (I guess due to the implementation algorithm) once the closed curve is aligned "inside". Looks like the radius/bevel for the join parameter is based on the distance to the mathematical curve (which in this case is equal to 0) instead of the center of the inside aligned curve.

The "join" parameter determines what sharp nodes look like when the path changes direction. See for example this Inside joins.afdesign file to see the effects of the 3 different join settings for otherwise identical inside aligned closed curves:

[DarkClown]

9 hours ago, R C-R said:

do not understand what you mean by the start- or endsymbol.

[DarkClown]

9 hours ago, R C-R said:

The "join" parameter determines what sharp nodes look like when the path changes direction. See for example this Inside joins.afdesign file to see the effects of the 3 different join settings for otherwise identical inside aligned closed curves:

See attached video ... everything works as expected when align is set to center ... but switching to "inside" the angles (corners) remain the same)

 

[DarkClown]

10 hours ago, R C-R said:

Like I have said earlier, if you want the stroke alignment to be based on to the left, to the right, & along the centerline of the curve instead of inside, outside, & center alignments, this needs to be posted to the feature request section of the forum. Personally, I am fine with it as it is now -- I would not want the alignment to change if I reversed the direction of the curve, which I think this left/right option would do for both open & closed curves.

LOL ... Seriously .... I guess we are all aware that this is a purely academic (fun) discussion and not part of a "change request" dream (aside it would be useless anyhow)! Affinity products are lacking so many really substantially workflow improvements on a far less philosophical leval than this topic - but for that you don't find anyone who listens

Of course I can live with the existing setup with regards to curves ... it's got weak points but frankly spoken this did not stopp any of the creative minds from doing great things with affinity products. So, as I said - a philosophical discussion ...

I wish i could have some influence on the really essential worklflow topics 😞 (It's not that they aren't aware ... it's more an economically based decision not to listen )

[DarkClown]

10 hours ago, R C-R said:

We also need to consider carefully how this would or should work with the 'quick shapes' (cogs, stars, arrows, & so on) since they do not have nodes & thus no defined start or end nodes. For them, inside & outside is well defined & intuitively obvious, but left & right is not since it depends on the start to end node direction.

They have a special status anyhow ... convert them to curves and than treat them as curves ... s.a.

[R C-R]

1 hour ago, DarkClown said:

They have a special status anyhow ... convert them to curves and than treat them as curves ... s.a.

??? That doesn't mean you can just ignore that they have no left or right sides but always have an inside & outside if they are not converted to curves.

[NotMyFault]

17 hours ago, DarkClown said:

Just two little sidenotes:

1) The align functionality is not working on a closed curve once there ia a start- or endsymbol selected and the symbol is placed within the line? (it works, once the symbol is placed at the end of the line) ... really strange. From my perspective this should not depend on the alignment setting.

2) As well the "join" parameter is useless (I guess due to the implementation algorithm) once the closed curve is aligned "inside". Looks like the radius/bevel for the join parameter is based on the distance to the mathematical curve (which in this case is equal to 0) instead of the center of the inside aligned curve.

Why not give it a try and report as separate potential bugs?

[DarkClown]

7 hours ago, R C-R said:

??? That doesn't mean you can just ignore that they have no left or right sides but always have an inside & outside if they are not converted to curves.

I wouldn't say you should ignore them - but they are a special breed of object that doesn't follow other rules. They have no nodes and don't ever come with overlapping surfaces but sometimes end up in unresolvable states (see here: hmm.afdesign) so simply take ANY point on the curve look counterclockwise and declare the align-state as "right" (aka "outside) vs "left" (aka "inside").

[R C-R]

4 hours ago, DarkClown said:

They have no nodes and don't ever come with overlapping surfaces but sometimes end up in unresolvable states (see here: hmm.afdesign)

I have no idea how you created that weird Pie shape -- I can't duplicate it even if I use exactly the same parameters on a different one. It also won't change to any of the preset Pie shapes. Can you list the steps you took to create it?

[DarkClown]

9 minutes ago, R C-R said:

I can't duplicate it even if I use exactly the same parameters on a different one. It also won't change to any of the preset Pie shapes. Can you list the steps you took to create it?

 

It is actually not a pie (no clue, why AP renamed it - wasn't me) .... it is a donut squeesed (transformed) in one direction until it becomes a single line ... and after that AP is lost with the shape ...

[R C-R]

11 minutes ago, DarkClown said:

It is actually not a pie (no clue, why AP renamed it - wasn't me) .... it is a donut squeesed (transformed) in one direction until it becomes a single line ... and after that AP is lost with the shape ...

I tried to duplicate that with a new donut by setting its height or width to zero in the Transform panel but I could not do that -- the smallest height or width it would permit was 1 px. So I wonder how you managed to squeeze it so far down that it actually became a single line. I did notice that in the Transform panel its width was showing as "0 px" but it was greyed out & could not be changed.

As for the name, donuts seem to be treated as a special case of the pie shape, as evidenced by the "Close Pie" item on the Node or Move tool's context toolbar, & by the presets on the context toolbar showing pie presets when a donut shape is selected.

A side note: you wrote "AP" (the abbreviation for Affinity Photo) but the file you furnished was an afdesign one. I did test it in both AD & AP, but in neither one could I set its height to zero.

As long as we are on the subject of donut & pie shapes, & your suggestion that you could take ANY point on the curve & look counterclockwise to declare the stroke alignment as "right" (aka "outside) vs "left" (aka "inside"), consider either shape with a non-zero hole radius, like these:

non-zeroholes.afdesign

I hope it is obvious that no matter at what point on either curve you start with, as you move along the path the direction you move changes from CCW to CW, making this notion of how to define left vs. right ambiguous.

[DarkClown]

4 hours ago, R C-R said:

I tried to duplicate that with a new donut by setting its height or width to zero in the Transform panel but I could not do that -- the smallest height or width it would permit was 1 px. So I wonder how you managed to squeeze it so far down that it actually became a single line. I did notice that in the Transform panel its width was showing as "0 px" but it was greyed out & could not be changed.

Don't use the transform panel but do it with the mouse on a grid ... I can reproduce the effect easily.

4 hours ago, R C-R said:

As for the name, donuts seem to be treated as a special case of the pie shape, as evidenced by the "Close Pie" item on the Node or Move tool's context toolbar, & by the presets on the context toolbar showing pie presets when a donut shape is selected.

Hmm explains things. But interetingly somewhere in the process ov manipulating the donut the naming on the layers panel seems to change automatically .... funny

 

4 hours ago, R C-R said:

As long as we are on the subject of donut & pie shapes, & your suggestion that you could take ANY point on the curve & look counterclockwise to declare the stroke alignment as "right" (aka "outside) vs "left" (aka "inside"), consider either shape with a non-zero hole radius, like these:

Yes ... where's the problem? Everything I said remains correct ... anywhere on the curve - counterclockwise right side is outside.

[R C-R]

23 minutes ago, DarkClown said:

Yes ... where's the problem? Everything I said remains correct ... anywhere on the curve - counterclockwise right side is outside.

Start moving along the curve in a CCW direction, beginning at a point on the inner radius. Note that by the time you have moved to a point on the outer edge you are now moving in a CW direction. Start instead at any point on the outer edge, moving in a CCW direction & by the time you are on the inner radius you will be moving in a CW direction.

[R C-R]

29 minutes ago, DarkClown said:

Don't use the transform panel but do it with the mouse on a grid ... I can reproduce the effect easily.

OK, I can do that using either guides or the grid, but when I do the zero px width or height in the Transform panel greys out & the shape becomes unresponsive to changes in that zero px value. The same thing happens with all the other quick shapes like cogs or trapezoids, so I think this should be reported as a bug, if it has not already been.

[DarkClown]

2 hours ago, R C-R said:

Start moving along the curve in a CCW direction, beginning at a point on the inner radius. Note that by the time you have moved to a point on the outer edge you are now moving in a CW direction. Start instead at any point on the outer edge, moving in a CCW direction & by the time you are on the inner radius you will be moving in a CW direction.

Sorry, I don't get you. Once you start moving in a ccw direction you keep on doing that regardles where you are or where the curve runs ....  you can only surround the surface either ccw or cw and this direction does not change wherever you are on the curve.

[R C-R]

2 hours ago, DarkClown said:

... you can only surround the surface either ccw or cw and this direction does not change wherever you are on the curve.

You were talking about moving around the curve, starting at any point on it. So if you start at point A, moving CCW, by the time you get to point B you will be moving CW.