Hi all,
I don't remember if anyone already asked for it but it could be handy to be able to define a by default material for the glass panels.
Right now, there are 2 materials (one for single-sided and one other for 2-sided).
I usually define my frame and glass material on the first FlexWindow I generate in the model and then duplicate it then tweak the sizes to need. The clones share the same material definition like that.

you're welcome;
it's the least I can do for my favourite dev

Here's an example of terrain mesh I made with VT and QFT.

Here's the default align to world axis gizmo behaviour.

Here's how I would like the gizmo to behave.

And for the last question, yes, sometimes in the move mode, some vertices end selected after moving.
It happens when I select a vertice (after the white circle to be visible), move it and then move it again but this time without the white circle to be visible. the vertice remains red (selected) and I need to go to the select mode to deselect all to get rid of the selection. The simplest way to resolve the issue could be to be able to deselect all vertices by clicking in the empty space just like all tools in Sketchup.

@thomthom said:

So you are picking vertices one by one? Not by selection drags?

it depends in my needs of course.
when I draw a turning road, I usually start with 2 vertices matching the road’s width, ctrl-move them with vertextools (or extrude them along a vector), and then rotate/scale the copied vertices with vertextool’s gizmo, ctrl-move them again, etc. until I end up sketching the whole curb, and them come back to some segments to adjust them more precisely. I usually work in top view, no perspective (cad way) to make the vertices match the dwg blueprint underneath.

so, yes, sometimes I select vertices with a marquee, sometimes I use QFT’s loop and parallel edges selections, it depends on what I need. but it’s mainly when I need to make small adjustments that the gizmo orientation issue is tricky.

By the way, is it possible to move the vertice freely in a xy plane instead of just along x or y axis (of the gizmo)?
In 3dsmax UI, if you click on the small square formed by the gizmo xy axis, you can freely move the vertice in the xy plane. (ditto for the xz and yz planes btw). I guess this could be a solution to simplify the workflow with vertextools.
Of course I know that you can move the vertices freely with the move mode but it annoys me to always have to switch between select and move mode to be able to move the vertice freely. I think eventually move and rotate mode should disappear if the gizmo of the select mode was more practical, because I feel they’re redundant.

I like the move mode when no vertices are preselected (with the small white circle when you hover over a vertice). Yet I dunno why, but after several moves, I can’t get rid of a selection (after the move, the vertices remain selected), and need to switch back to select mode to deselect everything before switching to move mode again to continue fast select and move of vertices. is there a workaround to avoid this? perhaps a shortcut to deselect all vertices...?

thanks for you reply

that’why I said to use the last two selected vertices when selecting more than two.

@thomthom said:

Yea, it would be doable to create a mode where two vertices define the gizmo orientation. Could infer orientation from the surface normal as well.

But what would you do with only two vertices selected?

move the vertices along the axis, scale their relative position, etc.
I've been doing a lot of terrains recently, and I needed this feature OFTEN, mainly to adjust vertices after creating them.

@thomthom said:

I see, so if there was an easy way to align the gizmo to an edge that would work?
(Possibly other methods to align to other things, like face normals etc...)

Thanks for your reply.
I think that the easiest behaviour for me would be a mode in which the gizmo inherits at least the red axis from the last selected two vertex. (And probably inherit the green axis from a tangent plane from the 2 vertex. Anyway once the red axis is correctly in place, it's easy to rotate the gizmo to fit the green axis in the plane we want).
But yes, If you can find an easy way to align the gizmo to an edge without proceeding to the right button/gizmo/align to custom affair (even if I already have shortcuts for everything - my head is full of shortcuts for every soft I use so the less shortcuts the better ), I'd be very happy!

Unless I buy a 3d navigator and always use the top view, then rotate it with the navigator's yaw to be able to use the gizmo in align to view ...?

PS There still is TIG's rotate plan script but you have to type numeric angle values and each time it zooms out the model so it's not handy.

@thomthom said:

Any chance you can show me a gif or video of how you do this in max?

I was meaning that I misunderstand vertice/vertex for edge because I'm a french speaker, and that in french vertice is more interpreted as an edge than a point.

For what I was asking you initially, it annoys me that each time I select an edge (=2 adjacent vertices), I have to right-click, go to gizmo, aligh to custom, then draw the red and green axis and then I finally am able to stretch or move the edge in a predictable direction.
I could use gizmo->Align to View but I'd use this mode more often if there was a way to keep the gizmo in a horizontal plane (X/Y plane or blue axis always vertical).
Previously I was using Julia's Relative Top View script for the view to remain horizontal but it doensn't work with 2017-2018 anymore (so it seems).

Thanks for reading.

@thomthom said:

How would you expect it to orient to a vertex? A single point doesn't have any direction by itself.

sorry, I meant edge...
It's because of my 3dsmax habits... to move edges, I select vertex sub-objects

Hi Thom

Is it possible to get the gizmo to auto-orient according to the last selected vertice? a kind of smart gizmo... that way, the scale function would always work as you expect...

I know that the gizmo is already "smart" and can follow the orientation of the last operation done (move, rotate), but it would be very handy if the gizmo was able to automatically adapt to the last selected vertice.

Ok, thanks.
Good to know!
EDIT: you're a genius

Hello Yoni,

I realized that when scaling manually, the stretching works correctly (FW keeping its dynamic compo attributes), but when I scale numerically (typing a value in the VCB like 0,25 or 0,5), the compo loses its dynamic compo attribute and becomes a regular compo (and the "smart" stretching fails).

Ex.: I create a FW, stretches it with anchor points and smart inference snaps in my model. ok.
Then I want to reduce its size, and during a non-uniform scaling operation (like, just the width), I type 0,25 to reduce its width to 25%, then the stretching occurs alla regular sketchup way, that's to say with the vertical frames becoming superthin, instead of keeping their width, Flex way.

Is it a normal limitation or a bug?

Cheers

Renderer: GeForce GTX TITAN X/PCIe/SSE2
GL version: 4.5.0 NVIDIA 382.05
Pixel Format: DBL RGBA:8-8-8-8 Depth: 24 Stencil:8 MSAA:16 FF:1
Res: 1920x1200

@rv1974 said:

overlapping geometry?
what happens when you bring back default blue? Update\downgrage your grivers.

Nope. it's not overlapping geometry unless Thom says otherwise.
It's clearly related to the OpenGL support of SketchUp, because WrapR suffers the same symptoms.

@rv1974 said:

Be more specific. what do you mean by 'trying visualize edges'? What's your aim?
Are you trying to make all edges black, same width?
Have you used some dedicated plugins Wrapr, color edges by direction etc?
check edges colorization in your style, if its 'by material'- switch to 'all the same'.
And it's not so bad idea to post your .skp file (a part?)

Don't you see that some edges are a mix of red and black?
The 2 examples show the exact same selection, but differ from an anti-aliasing point.

Above, on the AA4x, you can clearly see what is red and what is black (selected vs unselected), as it is supposed to be. Yet, according to the geometry, sometimes the red gets hidden under the geometry, dunno why. If you use a line only mode (F2), you can understand if an edge is red or not, but it's not always confortable to work in line mode.

Beneath, on the AA16x, some (supposed to be) selected edges are red AND black, and on some parts the red fades to black, some edges look even black even though they are selected.

I can't even imagine the struggle on 4k screens (I use FullHD only).
So for my understanding, there is an OpenGL fight between the red and black line, which is enhanced by the anti-aliasing. Then a solution could be to make the selected red line thicker?

Really cool timelapse!
I see you have the same issues as me with autofolds and mirrors (delete, remirror, redelete, reremirror, etc.). And to flatten faces you use scale?

Here are some printscreens as asked.

btw I never use profile lines in SU (except for good-looking final outputs, but never during modeling)

Hi Thomas,

I have sometimes problems trying to visualize selected edges, as with anti-aliasing SU tends to merge black lines and red selected lines. If I choose no anti-aliasing, or low anti-aliasing (2x, 4x), the selection is visible (red lines are fatter than black lines) but if I select 16x anti-aliasing, the red selection are barely visible.

So my question is:
Is it possible to change the width of the red (selected) edges? Btw I have the same issue with selected seam visibility with WrapR so I guess it's a SU thing...

Thanks in advance,

Thanks!
Another example... Here I only WrapR'ed the tires and painted them with Substance Painter, then exported back the texture to Sketchup, rendered with Thea...
Previously I had to add the markings with PS or stamp the text as vector shapes in raw sketchup geometry.

Often subdivision creates redundant loops which are useless (basically in flat areas), and some geometry refinement where it matters (in round, convex/concave areas).
So I'm asking if I can get rid of these "useless" loops, and build upon previous subdivisions, explode and keep subdividing while simplifying the control mesh by removing redundant loops each time.
SubD is limited to 4 subdivisions I think, so this could allow for more detailed meshes.

I haven't tried this "method" though, but was wondering if there was any problem with this approach of modelling.