PDA

View Full Version : Please help to solve a brain block about transform



madno
04-12-2014, 12:41 AM
Hi,
I think a brain block hit me. I am trying to transform some bricks like shown in the image.

121403

I can't find out how to do this in Modeler. Tried all action centers with the standard transform tool and the new transform gizmo.
Can anybody help?

EDIT:
Sorry I meant stretch not transform.

Oedo 808
04-12-2014, 01:04 AM
If I understand you right, you want to select a point, edge or poly at the bottom > assign pivot and then use the pivot as the action centre.

I actually use a third party assign pivot because I often found LW's flaky, can't remember who it was by though.

According to my shortcut key assignment, it's PivotBB (https://www.lightwave3d.com/assets/plugins/entry/pivotbb/) from Carl Merritt.

You don't happen to have LWCAD do you? Scale Snap on 1D would make this very easy, unless I'm completely misunderstanding the intent.

RebelHill
04-12-2014, 04:25 AM
Use the transform gizmo... right click on the bottom edge of the bottom brick... drag on the scaler.

madno
04-12-2014, 10:27 AM
Oedo 808 luckily I have LWCad. I just thought I did it in the past with some simple native LW tools. But no, like most of the time it is LWCad that lets Modeler survive.

RebelHill, I tried that, but if I put the gozmo at the bottom I then cannot zoom to the top of the bricks to scale accurately. The gizmo gets off the screen and I can't grab the scaler anymore.

But anyway thanks for help.

RebelHill
04-12-2014, 01:08 PM
Turn on independent zoom for one of the viewports... perform the operation in one viewport... watch it in the other... or use numeric panel to operate your stretch tool.

hrgiger
04-12-2014, 02:44 PM
Yeah if you have LWCAD, its literally a snap with scale snap.

JoePoe
04-12-2014, 03:45 PM
You can use Transform....with a linear falloff (same as stretch), action center=selection or mouse...or anything really . This way you can zoom in as much as you like, wherever you like, in the actual action viewport. You can use a falloff with stretch tool too.

XswampyX
04-12-2014, 05:44 PM
I think you are all missing the point?

What madno is trying to say is if he scales from the origin @ 1% the block at the top moves below the line that he wants to scale to (ie too much).

If so...

Either, do your bad scale, undo and then use the numeric panel to scale by 99.99% repeatedly until it's the right size.

Or

Set a weight map on the whole set of blocks to 1% and use that in your scale falloff.

Possibly, maybe.... :D

dee
04-13-2014, 06:06 AM
Try Transform->More->Fit to Background or Set Scale.

Oedo 808
04-13-2014, 06:36 AM
Try Transform->More->Fit to Background or Set Scale.

I think there must be few tools I haven't used, but that would be two of them. Fit to Background does the job nicely for this, I guess I never used it because 99% of the time I reckon I would like it not to deform the individual pieces.

probiner
04-13-2014, 08:59 AM
PX_Resize https://www.lightwave3d.com/assets/plugins/entry/resize/
Absolute Size https://www.lightwave3d.com/assets/plugins/entry/as-absolute-size/
Are the stuff I use

madno
04-13-2014, 11:33 PM
Thanks again for the tips.
I solved it with LWCad.

But RebellHill's idea about two independent viewports is smart. Following it the standard stretch tool and new tranform gizmo became usefull.

But:

Both tools only stretch the geometry in discrete steps (I noticed this with other LW tools as well).
Watching the numeric panel it seems the smallest step size is 1%.

121440

To make it more accurate I need to use the numeric tool and enter guessed values and hit apply multple times (like XswampyX said).

Is this behaviour of fixed steps normal for the tools? Or is there a secret setting to make it more smooth?
At the current state it is a lot of guessing and clicking (not with LWCad of course, there it works smoothly out of the box).

spherical
04-13-2014, 11:52 PM
Both tools only stretch the geometry in discrete steps (I noticed this with other LW tools as well).
Watching the numeric panel it seems the smallest step size is 1%.

To make it more accurate I need to use the numeric tool and enter guessed values and hit apply multple times (like XswampyX said).

Exactly right.


Is this behaviour of fixed steps normal for the tools? Or is there a secret setting to make it more smooth?

No. Dragging the mouse will only get you a default resolution. It's not all that difficult to enter the finer resolution increments into the Numeric panel and hit Apply a couple of times; narrowing down the value as you approach the intended goal. UnDo helps greatly in arriving at a good initial value, from which you can fine tune.

Oedo 808
04-14-2014, 01:24 AM
But RebellHill's idea about two independent viewports is smart. Following it the standard stretch tool and new tranform gizmo became usefull.

Just a tip you may or may not know if you think you might use this workflow a lot, you can make one of the viewports independent then assign it (single pane) to one of the numeric viewport presets (ctrl+numpad) set to record independence only, also assigning another key set to independence off. That way if you need to break zoom independence for a moment on one of the views you can hit numpad 6* and then remove the independence by hitting numpad 9* when you're done. This screws with my independent perspective viewport colour so I don't use it on that, and have that set to constant zoom independence anyway. Not sure if there are any other caveats to using this method.

*My keys for example.


It's not all that difficult to enter the finer resolution increments into the Numeric panel and hit Apply a couple of times

Are you kidding!? Every time you use the numeric panel instead of the viewport to make an adjustment, a fairy dies. :cry:

JoePoe
04-14-2014, 10:55 AM
You can have much much much finer control with the mouse only.
Use swampys other (weight) suggestion. Or (i think... havent actually vetted this and away from computer now) my suggestion with the linear falloff.... just overshot the influence by a lot (no need for independent zoom...but that's a cool thing anyway :)).

jeric_synergy
04-14-2014, 11:10 AM
Sorry for jumping in late/wrong, but:
looking at the OP's sketch, it looks like s/he wants the radii on the blocks to remain the same (??). Won't Transform/Scale squash those radii, not simply move all those parts together?

I vaguely remember someone doing something slick via cloning and then MOVE(W.MAP). The clone is "in place", with a gradated w.map applied to each clone, and then MOVE(W.MAP) moved each clone discretely as a whole, adjusting only the spacing, not distorting the mesh. (It was kUuL.)

erikals
04-14-2014, 02:47 PM
I think there must be few tools I haven't used, but that would be two of them. Fit to Background does the job nicely for this, I guess I never used it because 99% of the time I reckon I would like it not to deform the individual pieces.

btw... \ :]


https://www.youtube.com/watch?v=hcvuhyLMEz4

erikals
04-14-2014, 04:57 PM
I vaguely remember someone doing something slick via cloning and then MOVE(W.MAP). The clone is "in place", with a gradated w.map applied to each clone, and then MOVE(W.MAP) moved each clone discretely as a whole, adjusting only the spacing, not distorting the mesh. (It was kUuL.)

maybe like this >


http://youtu.be/dnuoc66DRjI


or like this >
https://www.youtube.com/watch?v=gK-9zXb5d-A

jeric_synergy
04-14-2014, 05:07 PM
I'm pretty sure it was Ken Nign's :bowdown: "Weight Clone":

https://www.lightwave3d.com/assets/plugins/entry/weight-clone/

Some brilliant tools there.

wishlist: a tool that would apply weights PARTS-WISE, so that one could make Parts (in the DPKit sense) have equal weights, for use with all the tools that respect w.maps. So the user could 'paint' weights in any direction like we can with the Weight Tool, but it would be applied in a Parts-wise fashion (i.e. "islands"). Maybe Sensei has already written that one.

JoePoe
04-14-2014, 07:49 PM
... Maybe Sensei has already written that one.

Well yes . Weight Polygon Groups (http://www2.trueart.pl/?URIType=Directory&URI=Products/Plug-Ins/WeightPolygonGroups)

jeric_synergy
04-14-2014, 09:47 PM
Well yes . Weight Polygon Groups (http://www2.trueart.pl/?URIType=Directory&URI=Products/Plug-Ins/WeightPolygonGroups)

Most excellent, sounds perfect (no time RN to watch the video) since it says "interactive" -- my concern is that a simple tool would be restricted to obvious axes.

Well done, Sensei!

http://www2.trueart.pl/?URIType=Directory&URI=Products/Plug-Ins/WeightPolygonGroups

JoePoe
04-15-2014, 03:57 PM
Quick update:
My suggestion in post #15 which should have read "...overshoot the influence..." does work :hey: :).

_______

@ Erikals .... love that second vid :thumbsup:.

erikals
04-15-2014, 04:09 PM
@ Erikals .... love that second vid :thumbsup:.

thank you... http://erikalstad.com/backup/misc.php_files/smile.gif

this one too i hope you like... :yoda:


https://www.youtube.com/watch?v=soJCqM-ARvE

JoePoe
04-15-2014, 04:46 PM
Yes, I saw that in the other thread :thumbsup: And I have used that technique in the past (except for the absolute value part.... clever).

Right now, though, I am trying to think of a way to automate the weight assigning process. It's easy enough if you have just a handful of objects to assign intervals of weight values, but more than 10.... :bangwall:

The weight clone plug is great if you need a lot of ....... clones :D. But like in your vid, often times you need this technique for differing objects.
And Sensei's plug is great for that. Heck I'm the one who gave his plug a plug! But it is $300, so sorry Sensei I am digging around in the dirt trying to uncover a workaround :angel:.

erikals
04-15-2014, 06:06 PM
Right now, though, I am trying to think of a way to automate the weight assigning process. It's easy enough if you have just a handful of objects to assign intervals of weight values, but more than 10.... :bangwall:

like this... ? http://erikalstad.com/backup/misc.php_files/smile.gif


https://www.youtube.com/watch?v=cxx4nmXWilo

jeric_synergy
04-15-2014, 08:11 PM
Is there a way to accomplish the same thing (movement by Weight Map) but:




along a spline
interactively
in MODELER?



Must meet all 3 conditions.

jeric_synergy
04-15-2014, 08:20 PM
like this... ? http://erikalstad.com/backup/misc.php_files/smile.gif

https://www.youtube.com/watch?v=cxx4nmXWilo
Huhhhh? WHY does that work????

Or does it just approximately work, but close enough?

Specifically, why does it quantize to parts?

AND: an lscript that would execute "W.Map Blur" a given # of times would be easy, no?

JoePoe
04-15-2014, 10:27 PM
like this... ? http://erikalstad.com/backup/misc.php_files/smile.gif

Very nice!!!

Same Question(s) as Jeric.... why?

Zero is still zero and 100% is still 100% right? I would think blurring would just (try) to make the transition in between smoother.
Acting more as stepped islands is counterintuitive to me. And as Jeric is asking as well, again same question....I assume it is indeed "merely" (a very clever merely) an approximation. Clearly(?) this does not assign a uniform value to each individual object.

I also wonder (I will try myself, don't worry) on a test with a collection of shapes of differing sizes and shapes. An object within the group that takes up more vertical height might stubbornly hold on to some stretching behavior.

jeric_synergy
04-15-2014, 11:28 PM
JoePoe, I continue to be baffled as to why this works, BUT, here's an example:


made an array
applied a 100% w.map
using WEIGHTS:LINEAR FALLOFF made a gradient across the geometry (btw, I h8 that tool)
using MOVE:WEIGHT MAP moved the array and got the following:



121466

Eyeball exam by moving the furthest cube to the lowest cube reveals no distortion, sooooooooo....

All I can think is that WEIGHT BLUR is actually not blurring, literally (big surprise, misleading nomenclature in a 3d app), but doing some kind of averaging between points. Very possibly more complex geometry might not work so neatly, but with simplistic cubes, we're gold baby.

Hell of a find by erikalst-- I would NEVER have expected WEIGHT BLUR to have this effect. :chicken: :stumped: :bangwall:

Oedo 808
04-15-2014, 11:35 PM
btw... \ :]

Ha, yeah it is pretty cool, it would be nice if created meshes retained their local space so that you could match local rotation to the background object also.

Regarding the topic overall, it would be good to see the option to manipulate based on parts as a whole without deforming the individual parts.

djwaterman
04-16-2014, 12:02 AM
Does any of this work for animation in Layout, using the weight maps?

jeric_synergy
04-16-2014, 12:37 AM
IIRC, Dana (dwburman) has a video showing EXACTLY how to use w.maps (or something) to move parts independently. The video was illustrated with an electric motor "assembling itself".

Check on Liberty3d.com (http://Liberty3d.com) for Dana's video tutorial packages.

+++++
The "Weight Blur" parts quantizing effect does NOT work on more complex meshes (at around 500 iterations), so we don't get a free "polygon island" tool. Also, "Weight Blur" slows markedly as the point count increases.

JoePoe
04-16-2014, 09:04 AM
Ha! I've been waiting for the "do it in Layout" post. Yeah, I guess... gradients w maps or even Part Move perhaps (?).
But I still like the Modeler puzzle.


....+++++
The "Weight Blur" parts quantizing effect does NOT work on more complex meshes (at around 500 iterations), so we don't get a free "polygon island" tool.

Thanks for checking with more complex objects. I suspected as much.
As for the free poly island tool... Ken has a tool that is very close (closer than weight clone) Random Weight per Connected (https://www.lightwave3d.com/assets/plugins/entry/random-weight-per-connected/). Gives a consistent value to each island, "we" (:hey:) just need to swap out the random part in favor of falloff options.

jeric_synergy
04-16-2014, 09:38 AM
JoePoe, my tests implied that if one did ENOUGH Weight Blur iterations (~2K?) the faux-quantization might be acceptable for use.

If someone did an LScript to automate that (with a progress bar!!!!!) it would become a viable tactic.

JoePoe
04-16-2014, 10:38 AM
Lol. I guess my focus was drawn to the bigger bolder underscored CAPITALIZED "not" :D.

As I sit here waiting for a weight blur to complete (waaaaaay too slow on even a modest bump up in poly count, at least on my machine)....
I go back to the OP.....

If the OPer does not care about the radii, then a simple falloff will do (fine adjustment by mouse or numeric control included!). As will the other techniques/tools mentioned.

If he does care and it's more about shrinking the spaces in between the objects (at least in a linear situation) it would be so much easier to simply place the last object where you want it and run Align and Distribute (https://www.lightwave3d.com/assets/plugins/entry/align-and-distribute/). A bit less interactive but much less labor intensive. Expanding out, of course, would eliminate the risk of overlap that shrinking down has.

erikals
04-16-2014, 01:45 PM
All I can think is that WEIGHT BLUR is actually not blurring, literally (big surprise, misleading nomenclature in a 3d app), but doing some kind of averaging between points. Very possibly more complex geometry might not work so neatly, but with simplistic cubes, we're gold baby.

Hell of a find by erikalst-- I would NEVER have expected WEIGHT BLUR to have this effect. :chicken: :stumped: :bangwall:

Thanks... http://erikalstad.com/backup/misc.php_files/smile.gif

i do think it does blur, but on a box it will only blur between two points and create an average, afaik.

so on a regular box (not skewed) it will work fine, but on more complex models you will have to blur it many times.

though after blurring enough times, it stops being approximate, and becomes only one weightmap value per object, keeping the geometry volume 100% the same.

yep, should be fairly easy to make a Lscript that runs WeightBlur many times over, but can't recall just how right now.

jeric_synergy
04-16-2014, 07:17 PM
It's a question of which would be faster to run as a script: a repeated W.BLUR operation, or "detect PARTS and assign w.map value to entire part", which still doesn't give the kind of interactivity Sensei's product probably affords, if you want to achieve a directional gradient (for instance).

The former would be a snap to program, but you might have to go have a cup of coffee.....

Sensei
04-17-2014, 07:11 AM
And Sensei's plug is great for that. Heck I'm the one who gave his plug a plug! But it is $300, so sorry Sensei I am digging around in the dirt trying to uncover a workaround :angel:.

In that price you have whole TrueArt's Modeling Pack
http://modelingpack.trueart.eu
Weight Polygon Groups is just one of hundred tools.

BTW, scripts can't be interactive. Only true compiled plugins.
Not to mention that finding group of polys is very time consuming task.
Running script with something bigger than a few hundred polys and you will be waiting minutes or hours.