PDA

View Full Version : Help with a simple deformation



digitalimagery
10-07-2015, 02:01 PM
I have a logo bug I want to model. It resembles a sail. I've created the triangle shape and subdivided the poly for deforming (images attached).

What is the technique used to grab a point, allow me to pull it on the z axis and gradually affect the surrounding points less and less so as not to affect the points on two of the three edges - thereby creating a rounded forward edge.

Any help on the best way to model this would help. I'm not real familiar with spline modeling.

Thanks.
DI

ernpchan
10-07-2015, 02:54 PM
What is the technique used to grab a point, allow me to pull it on the z axis and gradually affect the surrounding points less and less so as not to affect the points on two of the three edges - thereby creating a rounded forward edge.



The move tool with a falloff? Which is also the magnet tool.

hunter
10-07-2015, 02:54 PM
I would stick to spline modeling for this. Obviously I haven't taken time to line everything up precisely but it will save you some headaches. Still, looking forward to how others might do it with cloth effects or Bullet...
130201

digitalimagery
10-07-2015, 03:00 PM
I would stick to spline modeling for this. Obviously I haven't taken time to line everything up precisely but it will save you some headaches. Still, looking forward to how others might do it with cloth effects or Bullet...
130201

Thanks for your effort. Although this does look like the way to go if "I" decided how it looks, this is a logo and it's a BIG no no to make any changes to the design. The rounded corners won't fly.

Again, thank you for the time you put into this. I really need to get spline modeling under my belt.

DI

- - - Updated - - -

Thanks Ernest, will give this a shot.

DI

hunter
10-07-2015, 03:10 PM
Thanks for your effort. Although this does look like the way to go if "I" decided how it looks, this is a logo and it's a BIG no no to make any changes to the design. The rounded corners won't fly.

Again, thank you for the time you put into this. I really need to get spline modeling under my belt.

DI

- - - Updated - - -

Thanks Ernest, will give this a shot.

DI

Of course I get your point, but I did it in about 5 minutes. More finessing and some subpatch weighting would take of the rounded corners...

m.d.
10-07-2015, 03:10 PM
could just pin the outside edges, make it a bullet softbody and change the gravity direction (or face the model down) and let it hang naturally to that shape, then save transformed

digitalimagery
10-07-2015, 03:24 PM
The move tool with a falloff? Which is also the magnet tool.

Thanks Ernchan. That did the trick. Now that I've gotten the curve nicely modeled, what's the best method to pull up the forward (lower) edge to match the shape. as if it's pulled up a bit? Position the center up at the top and use the same method (magnet)? My efforst are making the edges that need to stay stationary pull up as well.

Thanks.

DI

digitalimagery
10-07-2015, 03:29 PM
Hm, I just this week got the newest LW. Was on 10.1 - no bullet. But I've been looking at the tuts and it looks amazing. I'll play with this option when I get a chance.

Thanks!
DI

digitalimagery
10-07-2015, 03:47 PM
I see your point. If I knew more about how to do that I would have made an attempt. I really do need to do more experimentation with spline modeling - if only I didn't need to sleep. :o)

DI



Of course I get your point, but I did it in about 5 minutes. More finessing and some subpatch weighting would take of the rounded corners...

ernpchan
10-07-2015, 03:58 PM
Thanks Ernchan. That did the trick. Now that I've gotten the curve nicely modeled, what's the best method to pull up the forward (lower) edge to match the shape. as if it's pulled up a bit? Position the center up at the top and use the same method (magnet)? My efforst are making the edges that need to stay stationary pull up as well.


You could try that. I would model this with subds. For me there's too many polygons with that sliced up piece you have. It'll make doing a deformation rather tricky since there's so many polys.

Would I would do is model the shape flat in Z with that curve already there. Then pull your 'sails' outward. This way you're just dealing with moving things in Z instead of Z and now out to make that curve shape.

Doing logos are always a challenge cuz the design is usually 2d-centric. Not so much 3d.

digitalimagery
10-07-2015, 04:10 PM
Great idea modeling the curve in advance. I'll go that route since my modeling skills are pretty non existent. I've addressed this 2D/3D issue with the 2D designers in the past for this reason (they don't look correct in true 3D). They've promised to keep that in mind but - they still do it.

Thanks for the tip.

DI

JoePoe
10-07-2015, 05:35 PM
I went with "simple" subd. Started with half a cone.

Are these points tight enough ;).

The problem is.... which I didn't address,.... do these three parts have to be one piece. Then it gets harder.
I guess I could freeze what I have and weld them together. But adding the subDs together is going to be rough.
(maybe points line up better for joining with the spline method?)

Your kingdom for NURBS!!! Soon enough!

130215

pinkmouse
10-07-2015, 06:00 PM
Five minutes. Would have been perfect if I'd spent a bit more time on the weight map... :)

130218

130219

JoePoe
10-07-2015, 07:39 PM
Another potential gotcha....

Are you planning on staying with double sided surfaces, or, does there need to be thickness.

Double sided: fine.
If thickness: Once these three sails are joined they will form a Mobius strip. So running Thicken won't work. Thickness will have to be addressed during the initial modeling process.

jeric_synergy
10-07-2015, 08:45 PM
All the sails are identical, no? Wouldn't you make 1 and then duplicate it twice?

JoePoe
10-07-2015, 08:50 PM
All the sails are identical, no? Wouldn't you make 1 and then duplicate it twice?

Absolutely!

jeric_synergy
10-08-2015, 12:02 AM
Giving it a shot: oh, that reminds me: anybody else have TOGGLE BACKDROP non-working in 2015.3?

Pretty sure I submitted that as a bug months ago.
+++++++++++++++++++++
OK, here's my 5 minute try, constructed with spline modeling. Made one sail, copied, rotated, and moved. Not perfect obviously, but 5 minutes.

The sale used 4 splines: I used 10x10 divisions, but should have used 5x10 in the smaller part of the flag. (See layer 1)
130223

130224

JoePoe
10-08-2015, 10:50 AM
What a community :) :thumbsup:.

I connected the three sides and ran Thicken. Crashed the first time, but it took the second time. Most of the resulting weirdness is hidden (at the moment at least) on the undersides of the connection points. So depending on your animation maybe no problem.

130229

Started with an 8 sided cone this time. So each arch section has 4 polys defining the curve.
All attached and still subD.

jeric_synergy
10-08-2015, 05:54 PM
Started with an 8 sided cone this time. So each arch section has 4 polys defining the curve.
All attached and still subD.
Cone??? Me so confused.

(Anybody else look at a LWM screengrab and try to rotate the model? ....nobody? :foreheads: )

jboudreau
10-08-2015, 06:14 PM
What a community :) :thumbsup:.

Nice work. Not sure if you noticed or not but in the reference image it shows the white sail over the blue. You have the blue sale over the white.

Hope that helps

I'm going to give this a try myself

Thanks,
Jason

JoePoe
10-08-2015, 07:04 PM
Nice work. Not sure if you noticed or not but in the reference image it shows the white sail over the blue.

Thanks....
Are you sure :) ?
This is one of the problems going from 2D art to actual 3D. Looks to me like a 50% transparency on those colors. Bit of an optical illusion, no? Who knows what's supposed to be over what. It's a bit of a modeler's choice imho.

For example: here's my model (Blue over white) at 50% transparency. :D

130237

(this transparency thing, btw, is going to be an issue going forward with this object)

Jeric: I'll explain the cone thing in a bit. Can't put together a breakdown at the moment.

jboudreau
10-08-2015, 07:06 PM
Yeah you're right, definitely an optical illusion because now I'm seeing the white over the blue,

Nice work!!

Thanks,
Jason

jeric_synergy
10-08-2015, 08:10 PM
The white, being more luminous, 'naturally' jumps to the front. I agreed w/jboudreau, but as soon as the mere suggestion that it could be behind was raised, I could see it both ways.

Very interesting, and a reminder that color theory is deep and mysterious.

++++++++
BTW, this discussion has led me to find a serious bug in Curves, regarding the CONNECT command.

http://forums.newtek.com/showthread.php?148358-BUG-Warning-LWM-Curves-and-the-connect-command&p=1448642#post1448642

(Case LWB-1859) LWM: CONNECT can severely distort Curves

JoePoe
10-08-2015, 08:26 PM
White and gold dress or black and blue dress :D ?

Very VERY interesting.
Color/luminosity had nothing to do with my height perception.
For me, it was all about the curvature of that edge. To me, a more intense curve meant a higher profile, and the softer part... more gradual/lower.

jboudreau
10-08-2015, 08:30 PM
The white, being more luminous, 'naturally' jumps to the front. I agreed w/jboudreau, but as soon as the mere suggestion that it could be behind was raised, I could see it both ways.

Very interesting, and a reminder that color theory is deep and mysterious.

++++++++
BTW, this discussion has led me to find a serious bug in Curves, regarding the CONNECT command.

what's the bug? I'll see if I can reproduce it

Thanks,
Jason

jeric_synergy
10-08-2015, 08:33 PM
Color/luminosity had nothing to do with my height perception.
Maybe consciously, but years of research say most humans perceive it that way.

JoePoe
10-08-2015, 08:39 PM
Maybe consciously, but years of research say most humans perceive it that way.

Hmmm, yeah, maybe.
Then how do you explain the red/blue overlap? the purple is less "luminous" but, in fact, the red is over the blue.

Maybe I should have worded it differently. I looked at the image, concluded (imho) that the color mix was 50/50 and therefore "untrustworthy" and let the other (also ambiguous) visual input to take the lead.

jeric_synergy
10-08-2015, 09:41 PM
Hmmm, yeah, maybe.
Then how do you explain the red/blue overlap? the purple is less "luminous" but, in fact, the red is over the blue.
In optical illusions, once one bit is 'decided', it informs the interpretation of the other bits.

130241 130243 130244

http://www.jstor.org/stable/1413561?seq=1#page_scan_tab_contents

JoePoe
10-08-2015, 10:05 PM
I'd like to think my interpretation was more driven by personal experience.

sail shape.... 130245

Also the shape of a wing.... same thing. :hey:

jeric_synergy
10-08-2015, 11:11 PM
You can try to deny your biological heritage, but it always wins in the end.

JoePoe
10-08-2015, 11:23 PM
I can lead a horse to water...... :D

MonroePoteet
10-09-2015, 04:26 AM
Not sure if it'll help, but here's a flat version of the logo. The "sails" aren't billowed like a true sail. They're just flat geometry with the spline curve simulating the sail curvature.

The modeling of the lower left sail was done with three Spline Draws (the curve and the two straight edges) in clockwise order, weld the end points, then select the bottom edge, left edge and curve in clockwise order (on behalf of Patch) and use Construct => Patch with perpendicular and parallel counts set to 20 to create the submesh. Then a Radial Array to create the three sails, nudging the original vertically to get the corners to line up.

Getting the saturation on the Red and Blue over a white background while simultaneously making them semi-transparent is pretty tough. With the white background, once the transparency is non-zero, the saturation of the colors goes down (i.e. gets whiter), and if the specific colors of the logo are trademarked (i.e. need to be close or exact), it wouldn't work. The overlapping areas would have to be set as separate surfaces and their colors explicitly set, or any animation moving the sails into place with transparency would dissolve to the true logo with the trademarked colors. Maybe there's a better way to do the overlapping areas than transparency.

Note that in terms of which is on top, I could only get the overlapping areas to look reasonably close to the original by having the White in front, then Red, then Blue. Otherwise, the overlapping areas weren't as close as they are. Maybe I just didn't play with it enough.


mTp

MonroePoteet
10-10-2015, 07:48 AM
FYI, the saturation issue I mentioned in the previous post can be overcome by duplicating the sails behind the front set of three, and making the transparency of the back three sails 0%. Thus, the overlapping portions are a result of the transparency of the front sails, but they show through to solid sails to provide the color saturation. What I wasn't able to fix simply is the red overlapping the blue producing purple rather than dark red, as in the original logo. I ended up adding a "Fake In Dark Red" panel on top of the red panel to force the right color.

RE: modeling (original post question), the added panel was modeled the same way: three Draw Splines, weld end points, Construct=>Patch with appropriate curve selection order and settings. The points on the inner curve were then dragged to match the original on the Red sail with Magnet ( : ) and right-drag to set the influence radius.

mTp

P.S. a careful comparison of the generated logo and the original, the overlapping blue needs to be a bit more green, so unless the WhiteSail and WhiteSailBack colors can be adjust to get that, another fake-in panel might be required to match the original logo colors. Or, as I mentioned in a previous post, any preliminary animation of the sails could dissolve to the precise original logo.