Nice: 2 points of what I've read so far.'

“Upon creating a primitive object, no tool should be switched to automatically. Current active tool should be preserved, otherwise the user finds again having to switch back to whatever he or she had selected - boxing match.”

This appears to the option to Create at Object level or in Context option. Setting to ‘Create at Object ’wont switch the tool.

“Here’s another problem I currently do not know if or how it can be solved:

img_33.jpg can you tell what’s the difference between these?

From left to right: all polys (primitives) selected, object mode, all edges selected
Needles to say, this is far from optimal.
”

Try turning on ‘Fill Selections’ in the Display Option/Effects tab. Primitves will fill in with a colour

Thanks MartybNz for these.
First point does not address the problem, however it's a nice thing to know.

The 2nd one (Fill Selections) is clearly what I was looking for. I looked after it in that panel too but it didn't catch my attention. Will update accordingly.

Great additions! Ladder is great but I agree, the extra option set default incremental values for transform gizmos would be ideal.

Looks like you're tackling the View Tool next so…

I'd like to add (a lengthy) note, not modelling directly but pretty key for many tasks, I just find it most apparent when trying to model in Houdini. Viewport Choice:

Currently two methods of working various perspective and orthographic views.

Viewport Switch and Quad View. Both have basic problems.

Viewport switch - You switch to Top, Front, Right by pressing Space + 2, 3, 4. Space 5 gives you UV view. Space+1 brings you back to Perspective View, that's all functioning fine. The problem is it resets the view -every-single-time- you hit Space + Number. It needs to remember where the view is/was. Space-F should have to be hit to reset the view.

So if you're modelling or setting up a scene you're constantly re-hitting Space-G to focus on objects or re-navigating to where you just were. This is maddening and has made me adopt the Quad view despite prefering the Viewport Switch.



Quad View - You maximize a quad-view by hovering your mouse over a quadrant then hitting Space + B. Hitting the same combination to return to Quad View. This time, where you leave your views is rememberd.

The basic problem here is that a lot of times the event to find which viewport to make active never fires or gets stuck somewhere. So you hit Space + B to maximize, say Top view with the mouse clearly over it, but it constantly maximizes Right view, the last view you had maximized.

Another problem I find is that you can sometimes hold down Space, move your mouse to a quadrant, then hit B and it will maximize the quadrant under your mouse. Hit B again while still holding Space to return to Quad, move your mouse to another quadrant and maximize that view. All the while holding down Space Bar.

Other times I have to constantly let go of space, doing the entire combination. I myself would much prefer the former. Both times cans still invoke the first problem. But please. Consistency!




Now for myself, I would much like to see an option to have the current mouse position event to be more active. When typing in a field/channel, still after months/years of slowly learning Houdini I find myself messing up expressions and values by immediately hitting Space to navigate around the viewport. All the while the subtle non-blinking type indicator is still active.

What I'd like to see is this option; If mouse cursor IS over a viewport (strictly) - End typing automatically and invoke navigation. If mouse cursor IS NOT in a viewport, add space type character or whatever. Probably not a great idea, but damn do habits die hard, haha.

Create a box and set its primitive type to polygon mesh - the object enters in poly (aka primitive in Houdini) mode having all of them selected and the ‘Y’ tool is activated.

This is because the ‘Create in Context’ instead of ‘Create at Object’ is enabled.

Here, I have selected one point with ‘T’ tool active and all the other points have disappeared.

However as soon as I de-select the point by dragging a rectangle on empty space the points reappear

Try changing Display Option/Guides/Follow selection mask, and doing the same thing.

Also there seems to be a very dark shadow which might be video card drivers - to be confirmed.

This is a smooth shading mode and the primitive's normals are converging and there is an automatic headlamp in the scene. Try to cusp the normals with a EdgeCusp node and try different lighting, and shading modes.

Gyroscope you pretty much nailed it on the head. I was intending to address these issues as well and I will.

MartybNz

Create a box and set its primitive type to polygon mesh - the object enters in poly (aka primitive in Houdini) mode having all of them selected and the ‘Y’ tool is activated.

This is because the ‘Create in Context’ instead of ‘Create at Object’ is enabled.

No, this applies for Create in Object, not for Create at Context Please re-read carefully the steps I've described on the page and you'll be able to reproduce it.

Here, I have selected one point with ‘T’ tool active and all the other points have disappeared.

However as soon as I de-select the point by dragging a rectangle on empty space the points reappear

MartybNz
Try changing Display Option/Guides/Follow selection mask and doing the same thing.

I see no change with Follow selection mask on or off. You sure you understood the problem I addressed?

Also there seems to be a very dark shadow which might be video card drivers - to be confirmed.

MartybNz
This is a smooth shading mode and the primitive's normals are converging and there is an automatic headlamp in the scene. Try to cusp the normals with a EdgeCusp node and try different lighting, and shading modes.

This is a work-around, because there is a shading problem, right?

Thanks all. Page will get updated these days.

Also there seems to be a very dark shadow which might be video card drivers - to be confirmed.

The shading is because of normals, the default head light and Smooth (Wire) Shading, it's really sticks out on a typical box. This set up and a box are usually the very first things people see/do when playing with Houdini and it's really not a great look.

Adding the Cusp polygons is a node fix but gloablly you can just switch to Flat (Wire) Shaded. Or increase the Default Emission in Effects in the Display options. Basically adding an ambient light to the scene. Here you can also turn off the headlight specular.

In your viewport example on web pages (please add numbers/letters to be clear what issue is addressed):

The first issue on your page (entering inside object node) depends how do you set creation context before you create object (as mentioned earlier).

Tool option: set it to Create at Object Level
Create box in the viewport (shelf tool or tab menu in viewport).
Result: Box object is created, Network View stays at Obj context (level)
Box object is highlighted and default tool gizmo is on, you can use it to setup transforms of the object at object level.

Tool option: set it to Create in Context
Create box in the viewport (shelf tool or tab menu in viewport).
Result: Box object is created, Network View dives into newly created object (obj/box_object1/) so you are in the mode where you are able to edit geometry of the Box object.
Box SOP is highligted and default tool gizmo is on, you can use it to setup Size and Center parameters of Box SOP.



It doesn't enter in (poly) primitive mode. It doesn't select any polygon.
It enters inside SOP mode and select and display default SOP handle.

Then if you are inside Box_object1 and you want edit poly. Switch to selection tools (S key), change selection to poly (2 key) and in viewport hit Tab key and add Transform SOP. Now you can move selected poly.
Or you can directly move the poly without adding Transform SOP, that way it created automaticaly Edit SOP (but it didn't remember history).



Second issue:
Hidden points in transform tool - You need to turn on Display points to keep all points visible during any other tool then Selection tool. (It's of by default).

Point Selection tool temporary turns on Display points option so it's easy to select points in viewport but as soon as you switch to other tool (e.g. Transform tool it returns to previous state that depends on your display settings).

To verify it: Create cube, turn on Select tool (S key), switch to point component (2 key). You can select points without any disappearing. Then select some points and switch to Transform tool. Now if you have Display points off then only affected points are visible. If you have Display Points on then all points are visible.



Third issue: I don't see why it should be problem that default selection includes all components. You can specify groups usually in most of the appended sop or directly in viewport before creating that SOP (you don't have to deselect all, you just start picking new desired components).

With selection memory it would be no problem at all. At first is selection set up to all components, because usually (with procedular workflow) we want that appended SOP affects all components (partition, connectivity, group filtered by expression, fracture,…) and specify groups by type and expressions after (and not by selection that could change with different input object). Then when you change the selection to your wish it remembers it and when you return to this sop and want to edit it more it restores the remembered selection.

When you specify selection of components by select in viewport and then add another SOP in viewport (e.g. Transfer SOP) it remembers what was selected in its group parameter. And when you return to this sop later and switch to default handle (y key) it highlights this selection.

1.0 - Brush picking (let's keep it Brush please, but enhance it)

Brush picking in Houdini is more closer to Softimage Paint Selection (F11 key) How to improve it?

+ Please add MMB drag (middle mouse button) drag to adjust radius (or Shift+(any mouse button) how it works on Paint SOP now)

+ Add visualisation of radius before and during the stroke is made (like Paint SOP)

+ Add possibility to turn on/off dynamic change of brush radius based on pressure (tablet support)
(more pressure = bigger radius, small pressure = smaller radius. And in preferences add toggle Radius by pressure on/off, possibility to define minimum radius, maximum radius, and maybe even pressure to radius curve)


That way if radius is set to very small value (or light stroke) it behaves as Raycast Freeform Selection in XSI, if radius is larger it behaves as Paint Selection tool.
Two in one.


So I can't see any good reason why it should be crippled to behave only like Raycast Freeform selection from XSI.

In your viewport example on web pages (please add numbers/letters to be clear what issue is addressed):

The first issue on your page (entering inside object node) depends how do you set creation context before you create object (as mentioned earlier).

Tool option: set it to Create at Object Level
Create box in the viewport (shelf tool or tab menu in viewport).
Result: Box object is created, Network View stays at Obj context (level)
Box object is highlighted and default tool gizmo is on, you can use it to setup transforms of the object at object level.

I don't understand why both you and Marty don't read what I say exactly as I say it.
On the page I said: Create a box and set its primitive type to polygon mesh
I didn't say Houdini will select polys immediately after I create the box.
I said "and set its primitive", therefore after you create it, you enter in SOP and set the primitive type to polygon mesh.
Then it does select all the polys.

Second issue:
Hidden points in transform tool - You need to turn on Display points to keep all points visible during any other tool then Selection tool. (It's of by default).

Point Selection tool temporary turns on Display points option so it's easy to select points in viewport but as soon as you switch to other tool (e.g. Transform tool it returns to previous state that depends on your display settings).

To verify it: Create cube, turn on Select tool (S key), switch to point component (2 key). You can select points without any disappearing. Then select some points and switch to Transform tool. Now if you have Display points off then only affected points are visible. If you have Display Points on then all points are visible.

I was expecting that that setting would display points when in object mode only, not when in point mode when one intuitively expects to see all the points regardless of what tool (selection or transform) is selected. What purpose does it serve to hide (by deactivating display points setting) points when in both object as well as point mode?

1.0 - Brush picking (let's keep it Brush please, but enhance it)


Two in one.

So I can't see any good reason why it should be crippled to behave only like Raycast Freeform selection from XSI.

If the brush tool would not be able to select by simply LMB clicking but only by making a stroke (like it is now), then we need a raycast like selection tool. Otherwise I agree with the improvements you proposed to which I'd add the option to save a default after making a customization.


In many areas Houdini tries to reinvent the wheel for things already sorted out which work simple and intuitively.
I get that each program is unique and has ways of doing stuff necessarily made so because of its inner connections, but when you have people coming from all kinds of 3d apps and say that Houdini is on a different level of unintuitive and backwardsness (new word alert) regarding the way you interact with your objects in the scene, I say there must be some kernel of truth somewhere in there.
Thankfully SESI is aware of this and hopefully they'll do something about it sooner rather than later even if that involves ruffling a few seasoned Houdini users' feathers.

McNistor
No, this applies for Create in Object, not for Create at Context Please re-read carefully the steps I've described on the page and you'll be able to reproduce it.

The explanation for polygon/primitive selections is that you have a the primitive selection tool active, whereas all the points will be selected if you have the points selection tool active upon box creation.

For the question ‘Why?’ It has advantages to immediately creates groups, apply sops to entire group like mountainSop etc.

This, coupled with other things I’ll discuss below, creates a stuttered experience, it feels like you’re in a boxing match with the interface where you constantly have to revert, change, switch, etc. because the program has a mind of its own.
Nothing should the program select for the user, be it object or transform tool. Neither should it enter nor select points,edge,polys unless requested to do so.

There is no boxing match once you learn it and then it feels natural. i.e. an analogy is that a bicycle is terrible- it falls over and the direction has a mind of it's own, that is, until you learn to ride it, a car too, and, any skilled technique. The idea that the program selects any points/prims on it's own could be a preference as it feels right 100% of the time coming from a Houdini background, and maybe 0% of the time coming from other software.

That's the same reason why ‘Alt key’ was added as a viewport tool recently. Houdini artists have no need for it but anyone else seems to need it. bizarre!

I see no change with Follow selection mask on or off. You sure you understood the problem I addressed?

It may be as OsX is currently limited to GL2 viewport and there is difference to the GL3 viewport, but, I certainly see differences. With the ‘Follow selection mask’ off, when selecting the points selection tool no points are visible and I need to turn on point display to select points, whereas with ‘Follow selection mask’ on, as soon as I switch to points selection tool the points are visible and I can select them. This then explains your ‘However as soon as I de-select the point by dragging a rectangle on empty space the points reappear’

Also there seems to be a very dark shadow which might be video card drivers - to be confirmed.

MartybNz
This is a smooth shading mode and the primitive's normals are converging and there is an automatic headlamp in the scene. Try to cusp the normals with a EdgeCusp node and try different lighting, and shading modes.

This is a work-around, because there is a shading problem, right?

If this is a workaround then Mantra also exhibits the same issue The SmoothShaded viewport is most likely using Phong shading, when a primitive with smooth normals is viewed with a headlight this is the correct shading. I read somewhere that Maya, probably SI, puts down a cube with cusped normals automatically, which then explains why people are not be familiar with this.

You can toggle the ‘Consolidate Corner Points’ checkbox on Box obj, and, if you turn on normals display in the viewport you will see so-called ‘correct’ shading

This is a common complaint on the forums, that for someone testing Houdini says it looks terrible, so maybe it could be changed to default to ‘Consolidate Corner Points’, though the advantage is it does force you to learn more about normals, which is good in the long run.

The shortcomings of a non-realtime discussion.
You might want to review your post as you were probably compositing it while I posted mine.
I will reply to yours shortly.

(a feature to this forum comes to mind - one that alerts the user that at least one reply was added before submitting the reply)

For the question ‘Why?’ It has advantages to immediately creates groups, apply sops to entire group like mountainSop etc.

OK then, as pezetko said, if selection memory is implemented, then this is a minor inconvenience.

There is no boxing match once you learn it and then it feels natural. i.e. an analogy is that a bicycle is terrible- it falls over and the direction has a mind of it's own, that is, until you learn to ride it, a car too, and, any skilled technique. The idea that the program selects any points/prims on it's own could be a preference as it feels right 100% of the time coming from a Houdini background, and maybe 0% of the time coming from other software.

Yes there's a boxing match, at least for a modeler that because most of he/she does is select and interact in the viewport.
The analogy here is between bicycles, not bicycles and cucumbers. Thus, if there's such a thing called "intuitively to ride bicycle“ then one could imagine also a ”hard as hell to ride bicycle".
You could argue that after you learn to ride the second type then it's not that hard which might be true, but that still doesn't invalidate the fact that by comparison it's still hard as hell to ride.

Initial selection, as I said would not be a big deal. I will remove it as I see this has drawn more attention than intended.

That's the same reason why ‘Alt key’ was added as a viewport tool recently. Houdini artists have no need for it but anyone else seems to need it. bizarre!

I agree the alt key for navigation is weird inheritance due to a historical idiosyncrasy and it doesn't make much sense in the light of ergonomic usage of the keyboard. It's much easier for the hand to keep the thumb on the space key and it also facilitates access to other keys.

It may be as OsX is currently limited to GL2 viewport and there is difference to the GL3 viewport, but, I certainly see differences. With the ‘Follow selection mask’ off, when selecting the points selection tool no points are visible and I need to turn on point display to select points, whereas with ‘Follow selection mask’ on, as soon as I switch to points selection tool the points are visible and I can select them. This then explains your ‘However as soon as I de-select the point by dragging a rectangle on empty space the points reappear’

I still have no idea, but I'll tackle this later.

This is a smooth shading mode and the primitive's normals are converging and there is an automatic headlamp in the scene. Try to cusp the normals with a EdgeCusp node and try different lighting, and shading modes.

If this is a workaround then Mantra also exhibits the same issue The SmoothShaded viewport is most likely using Phong shading, when a primitive with smooth normals is viewed with a headlight this is the correct shading. I read somewhere that Maya, probably SI, puts down a cube with cusped normals automatically, which then explains why people are not be familiar with this.

You can toggle the ‘Consolidate Corner Points’ checkbox on Box obj, and, if you turn on normals display in the viewport you will see so-called ‘correct’ shading

This is a common complaint on the forums, that for someone testing Houdini says it looks terrible, so maybe it could be changed to default to ‘Consolidate Corner Points’, though the advantage is it does force you to learn more about normals, which is good in the long run.

OK, thanks for the info.
The only suggestion to make here then is to have that setting off by default (for the viewport display, not at rendertime if that propagates to there) and the reason to this is to avoid future discussions started by new users.

get that each program is unique and has ways of doing stuff necessarily made so because of its inner connections, but when you have people coming from all kinds of 3d apps and say that Houdini is on a different level of unintuitive and backwardsness (new word alert) regarding the way you interact with your objects in the scene, I say there must be some kernel of truth somewhere in there.
Thankfully SESI is aware of this and hopefully they'll do something about it sooner rather than later even if that involves ruffling a few seasoned Houdini users' feathers.

That's cool. There's definitely some truth in that, if everyone simply argues the pros and cons of each new idea, as you have been doing very well, we won't call anybody crazy it's always easiest not to change things, so one needs to a great and logical argument.

McNistor
(a feature to this forum comes to mind - one that alerts the user that at least one reply was added before submitting the reply)

OdForce forums have this, one reason its a great area to discuss Houdini stuff too.

Indeed.
Before I update my page I want to ask you something about a thing I find extremely annoying:
does it bother anyone the way that upper toolbar changes height based on the tool you have active?
It makes the viewport window change dimensions and a very disruptive pop up and down occurs.
Know what I refer to?

I've logged that as a bug but only for HighDpi mode. Thought it didn't happen at large and smaller ui sizes.

When did you log it?

I just checked - it's something else - UVEdit and HighDPI mode. the UI bounces.

McNistor
Yes there's a boxing match, at least for a modeler that because most of he/she does is select and interact in the viewport.

When does this happen, apart from creating a primitive with ‘Create in Context’ mode?

Also check Twod last post:
http://www.sidefx.com/index.php?option=com_forum&Itemid=172&page=viewtopic&t=30844&start=25 [sidefx.com]

Snapping is rewritten already for the next release