5 point patch (troubleshooting)

[LenseOnLife]

Evening All from windswept, dark and wet West Cork (Ireland),

 

Maybe it is the weather, but the computer beckoned today and I hauled myself into learning AM again. Nothing difficult but I am again banging the sore head against a harder wall than before.

 

Been trying to cut a hole in a model and had varying levels of success - so much so I decided to get back to basics and see if I could fathom what I was doing wrong. Did a very simple outline, used an 8-sector lathe and formed a simple 3-d model.

 

Cut a hole in either side and then started filling in 5 point patches. To me (uninitiated) both holes were created exactly the same way.

 

The First Hole: All 5 point patches worked fine but when trying to fill in the last patch - it was a real pain. Selection1, 2 and 3 show the points being selected but the 5-point patch icon is not selectable.

 

 

It appeared to be dependant on only selecting points when they were clearly visible and not near the outline graphic of the selection as in 'Selection4'.

 

Selection5 shows the final patch created.

 

 

So, in this instance, I didn't change the CPs in any way, but it seems that AM is very sensitive to where / how the CPs are selected?

 

The Second Hole: The first 3 went OK (had a bit of bother with patch 2) but I am now on patch 4 and cannot get the 5-patch icon to become live. This is shown in Selection5 - each of the points that have been selected are clearly visible but still no 5-point selection tool.

 

 

 

Before posting I decided to have one last attempt.

 

I made a group called '5 points 1' of the group that wouldn't work, I then selected an area around where I wanted to work and hid the rest. Tried selecting the 5CPs again and this time the icon showed - so I named this group '5 points 2'. These are in the attachments '5pt Selection Group1' and 5pt Selection Group2'. Both groups look identical - why does one work and the other not?

 

Selecting the group '5 points 1' group shows that the 5-pt-patch icon is not enabled.

Selecting the group '5 points 2' group looks identical and now the 5-pt-patch icon is enabled.

 

Have I missed the plot somewhere and making life unnecessarily difficult for myself - or is it just the cold snap and rain affecting my brain and if I go to be will everything be right in the world when I wake up?

 

Sorry for being so long-winded

 

Oliver

 

 

[John Bigboote]

Hi Lens!

It is not you, this is a very common problem, I believe Rodney was just talking about this in another recent thread regarding 5 pointers. Somethng about using the period key to select them then hitting 'shift' to close the patch.

 

What I have always done is to hide everything BUT the 5 points that will not close, then drag a lassoo selection over all of them and the button becomes available. If not turn the view and try from another angle...

 

Hope you get this and continue your A:M journey!

[Fuchur]

There are a couple of things you need to be doing when selecting to get the 5-pointer-button to show up...

1.) If you are selecting the CPs with the shift or the crtl-key pressed, make sure to press the crtl / shift-key before you select the first CP. I am not sure why it is that way, but this seems to help.

2.) If you selected it accidently without pressing shift or ctrl before the first CP as been selected, hide everything (using the hide button) and use a drag-selection around the CPs. This should help.

3.) If you have a problem with differently selected CPs (one drag-selected, the others by shift-click for instance) you can use the comma-key to invert the selection and press it again to get the initial CPs back.

4.) If you have an unconnected CP or a double-cp somewhere, it is very likely not possible to create a 5-pointer. Make sure there aren't any.

5.) 5-points in direct connection to hooks are very likely not working. I would try to put in another spline-ring. If you absolutely have to use it (I do not recommend this) make the five-pointer first and attach the hooks afterwards.

 

Hope this helps .

See you
*Fuchur*

[Rodney]

Rodney was just talking about this in another recent thread regarding 5 pointers. Somethng about using the period key to select them then hitting 'shift' to close the patch.

 

 

Close but nope.

 

HIt the period key twice. Period. Period.

Once to select the complement (inverse) of your selection and then once again to have A;M reselect your original selection.

 

This is an easy solution that is a little hard to explain but I'll give it a try.

 

A:M's basic rule for 5 point patches is similar to the rule for any valid patch and A:M's eternal question (it asks itself as we are modeling) is, "Is this a valid patch?" If the answer is yes it renders the group as a patch. If no, then it doesn't.

 

I suspect in cases like this if you were to follow the splines all the way around to the other side of the model you might discover that the spline is one very long continuous spline (or at least a selection that A:M has determined is not a valid (5 point) patch.

 

But here's the thing... we CAN make a 5 point patch out of a single spline! We just have to tell A:M that is something we want to do.

So, we can select our set of 5 Control Points, whack that period key to get everything except those 5 points and then... whack the period key again to (properly) select the 5 points.

*I* suspect that A:M is reordering these points in some internal matrix and after the reorder it 'sees' the selection as valid for closing as a 5 point patch.

 

Hitting the period key twice after selection (and having the 5 point patch indicator stay greyed out) is what I'd call the 99 percent solution. It works (for me) *all of the time*.

The only exception (when it doesn't work) is when the selection isn't valid for forming a 5 point patch (i.e. I've selected at least one CP that isn't congruent and would not properly form a patch. Sometimes this is because I'm trying to make a 6 point patch and don't realize I've selected that 6th CP. Sometimes it's because I've only select 4 CPs. More often than not, I've selected 5 CPs that share the same spline or that or not congruent (i.e. do not form a valid patch) and A:M (initially) think we will not want to make that selection a 5 point patch. In other words, A:M sees our selection as invalid for forming a 5 point patch.

 

So, WHACK... do NOT gently select... thy period key... TWICE!

[robcat2075]

3.) If you have a problem with differently selected CPs (one drag-selected, the others by shift-click for instance) you can use the comma-key to invert the selection and press it again to get the initial CPs back.

 

I'll note that this should be the period key that is pressed twice to invert and un-vert the selection.

 

 

That works for most cases. Almost all.

 

Exceptions are when there is an additional dead end spline attached to one of the selected corners or the hook situation noted above.

[Rodney]

use the comma-key to invert the selection and press it again to get the initial CPs back.

 

Fuchur means period key here. The period key inverts any given selection.

If we use the comma key A:M will sound a loud blast of Pong! Pong! Pong! for every time we use the invalid key command after selecting Control Point (CP).

(Note that hitting the comma key after selecting a spline I believe will just have A:M sit there and stare at us... doing nothing... as if to say... yes, I've selected the whole spline as you requested...um, what else is it that you want me to do with that comma? My understanding is that A:M only knows how to select the remainder of a spline via the comma key if a spline, not a CP or set of CPs.

 

 

Edit: What Robert lectured.

 

 

Added: I'm tempted to capture a video of me making 5 point patches rapidly with all manner of selections to demonstrate just how useful deselect/reselect is when 'troubleshooting' 5 point patches. If we aren't using the period.period. method we aren't properly engaging our wannabe 5 point patches. It is important to note however that just because we can make a patch does not mean we should make that patch as A:M will allow us to create less-than optimal patches.

[LenseOnLife]

Hi,

 

Thanks for all that feedback

 

DOT DOT or PERIOD PERIOD rules the world.

 

I've tried it on many things that I had put on hold while trying to get to grips with this problem and they ALL were 'fixed' using the DOT DOT / PERIOD PERIOD method or maybe FULL STOP FULL STOP method

 

whatever

 

THANKS ALL

 

I'd better just cancel the trip to the funny farm as it APPEARS that I still have a couple of marbles left

[Rodney]

Here's a short video of me going through some randomly splined shapes and turning them into 5 point patches..

There are a few surprises (for me) along the way and in one instance... I almost... almost... think I find a 5 pointer that won't activate.

In the end that one gives in and forms a valid 5 point patch too.

 

The one example I should have included is the standard one that you demonstrate via your post.

Spline layouts like that are often created for places where arms or legs or necks will extend out of an area.

But those areas are just as easy to troubleshoot.

My hope would be that folks approach 5 pointers with a confidence that they can and will form valid patches.

I get a sense that folks avoid them because they run into these cases where the 5 pointers don't immediately work.

The easy solution is Period.Period. If that doesn't work, inspect those connections a little closer and reattack!

 

Edit:

Glad to see it's working for you!

 

FULL STOP FULL STOP method

 

I like that.

 

 

Added: I should include for completeness sake that we can assign a shortcut key to selecting that 5 point patch icon. I recently assigned Control 5 as my shortcut keys (but should look at a better combo to go with hitting the period key twice... pehaps Shift Period? (I'll try that) In this way we can quickly hit... Period.Period.ShiftPeriod (or whatever final key combo) and fix those troubleshome 5 point patches. Edit: I think Ima gonna go with Shift Period that works well.

5pointers.mp4

[Rodney]

The Period.Period.ShiftPeriod method works so well that I might submit an A:M Report requesting 'Shift +.' be set as default/permanently for 'Make 5 Point Patch'.

In this way the standard answer to the question of troubleshooting 5 point patches would be, "Hit the Period Key twice (invert selection/return to original selection) and then Shift Period (to create the patch).

 

 

Hmmmm....

[LenseOnLife]

All I can say is that it has really changed my approach to model making in AM. I think it should be put up there in lights so that people who are new to the software (like myself) know about it - this means making sure that there is a video on the official website (that just about every newbie watches anyway) to get a toe-up [either under 'hash tutorials' or 'user tutorials']. Once the method is known, that is more than half the battle.

 

For me, if [period period] brings up the 5-star-patch icon them I'm a happy camper and I'd think that would be the same for the vast majority of people as well.

 

Thanks again for all your help,

 

Oliver

[robcat2075]

I'm supposed to do a quickstart modeling tutorial and I think I will mak a point to mention the period-period thing.

 

 

The mystery I'd like solved is, why does it make a difference?

[nemyax]

*I* suspect that A:M is reordering these points in some internal matrix and after the reorder it 'sees' the selection as valid for closing as a 5 point patch.

When you invert-invert, you don't necessarily end up with your original selection. Since CPs are fused at spline intersections, you may have selected only one of the fused CPs at a given intersection (if you click-selected rather than box-selected), but A:M might need the other one for a valid 5-pointer. When you invert for the first time, you deselect whatever you had selected at the intersection. When you invert again, you select everything at the intersection. Therefore, the selection becomes valid.

[Rodney]

I'll add the following to the mix for info purposes.

This is only slightly modified from one of Martin's papers:

Hash splines use a connectivity methodology such that any single control point knows its direct neighbors, and consequently, the neighbors of its neighbors, ad nauseam.

 

Example:

control point

{ control point *next, *prev, *junction }

So, I take it that A:M can read through a spline's CPs to determine if they can form a valid patch.

Or put another way:

 

A simple recursive algorithm walks the spline mesh’s control points, checking for enclosed areas that can become a valid patch definition. References to the control points are put into the patch array, the splines direction flags are determined at that time. When the model saves and loads, unique control point identifiers are used to reconstruct the references so that the patch finding algorithm need only occur when the spline mesh changes connectivity.

 

FindPatch( cp1, cp2, cp3, cp4, startcp )

For all junctioncp

if startcp then

if startcp is the junctioncp AddPatch( cp1, cp2, cp3, cp4 )

else if junctioncp->prev then

FindPatch( junctioncp->prev, cp1, cp2, cp3, cp4 )

if junctioncp->next then

FindPatch( junctioncp->next, cp1, cp2, cp3, cp4 )

I shall now guess that inverting a selection loads the new data into memory which by itself might jumpstart the recognition of an area that is valid as a 5 point patch from what previously was not known to be valid because it hadn't been saved/loaded but merely selected. I'm sure there is a way to test this to determine whether inversions allow for this to be true on not...

 

 

 

 

 

 

...