V17 Retopology Tool

[John Bigboote]

Hey Gang, I finally got a chance to try out the new V17 Retopology Tool, and I think it is pretty well thought feature. Kudos to Steffen! I know at this time V17 is in alpha and a lot of people don't have access, but will eventually and here is my 'illustrated' guide to getting the tool up and running.

 

I started with Fuchurs useful instructions from the V17 thread:

 

1.) Import a Prop (for instance an OBJ) into the chor-window.

2.) Create an empty model in the PWS and drag and drop it into the chor too.

3.) Good practice is to make everything but the empty model in the chor unpickable.

4.) Now select the model and go to the modeling-mode in the chor.

5.) Click on "Add Spline" and activate "Snap to Surface".

> Now you are ready to go.

 

Here is my journey down that road, with notes:

 

FIRST OF ALL, I needed a polygonal model so I went to TurboSquid, signed up, and browsed their free models. I wanted an OBJ but ended-up with a 3DS, didn't matter. Here is the link to the model I grabbed:

http://www.turbosquid.com/3d-models/polygo...head-3ds/265937

[John Bigboote]

Next, I booted-up my V17alpha2. Thanks to A:M for letting me be a tester on this, I will post bugs- but haven't found many yet.

 

Initially I made a mistake, and made a new model and tried to import the model into it. It bogged at 20% and I reread Fuchurs instructions and realized I needed to import the poly model as a prop DIRECTLY into the choreography, as shown.

[John Bigboote]

SO- Here is the polyonal model in the chor as a prop. I centered it on the XYZ grid and sized it up a little, and deleted the ground. I see the teeth are separate and off to the side, it would be nice delete them for this demo- but will save that for another app another day.

[John Bigboote]

Next, I made a new model- dragged it onto the choreography in the PWS, and locked the other items. I switched from Chor mode to Modeling Mode (F5) This will conclude the prep-work... I am ready to start modeling!

[John Bigboote]

Now it's time to start modeling! I make sure that the new 'SNAP TO SURFACE' button is ON, and click ADD CP and start laying splines.

[John Bigboote]

As I draw- I can see the resulting splines closely laid upon the polygonal surface! JUST WHAT WE ASKED FOR!

[Darkwing]

This looks incredible! Insanely intrigued as to how this turns out!

[John Bigboote]

Now as I look into the model, I see only my splines- I can play a little 'connect-the-dots' and get my A:M patches! THIS TOOL IS GREAT!

[John Bigboote]

AND, as I further play... I see I can delete my CP's and make a grid and apply it (via projection) to the surface as well! The denser the mesh, the better the results!

[largento]

Very nice! Thanks for demonstrating, Matt.

[John Bigboote]

But I do see there is some further work to be done to the tool. As I make the mesh denser and retry the 'Apply Group To Surface' feature, I notice it gets confused as to WHICH surface I mean, and about half the CPs place themselves at the BACKSIDE of the head. COULD BE the normals in the POLY model need to be checked to see if they are all facing outward... could be that A:M simply looks for the closest poly intersection in Z space, and finds a lot along the backside of the head. I see it would be useful to have a polygonal editor available when working with this tool, possibly.

[Darkwing]

V17 and I are gonna have a date with some star trek models I think when it's released!

[NancyGormezano]

terrific style of presentation Matt. Well done!

 

The denser the mesh, the better the results!

 

Hmmm...even tho there are still glitches...I thought the reason one might want to use retopo is to make it easier to remodel something done in another app but with hash spline topology. Usually way less dense than other systems? I do not travel back & forth between apps...so I may be missing the point.

 

Would not one want to lay out spline rings for "holes" and allow for 5 pointers? Or is the objective to not have 5 pointers?

[John Bigboote]

Would not one want to lay out spline rings for "holes" and allow for 5 pointers? Or is the objective to not have 5 pointers?

 

It's entirely up to the 'one'... that would be a main object of such tool, I am just showing other features available to it. The grid method might be more suitable to mechanical modellers, like if you were doing the fender of a car, hood, or entire vehicle. Here is the screen-cap of the problem I am having... perhaps I will submit this to A:MReports.

[Darkwing]

Would doing something like disabling back facing polys help or anything? This is a problem I have a lot when I'm doing legit modeling in AM. If I have two sides of something and so I select a CP, instead of it selecting the one in front in the view I'm in, it selects the back instead. Perhaps this is just an offshoot of whatever that is?

[John Bigboote]

On closer look, it appears my 1st suspicion is right. This may just be a limitation of the feature. So, denser/higher polycount models will work better... OR if you have a polygon editor app you could open the poly model and delete the backside of the head and this should then work just fine.

[Vertexspline]

Thanks for posting this-----despite some issues this seems like a really nice addition to our workflows.

[John Bigboote]

No doubt. This feature should prove to be the biggest modeling innovation to A:M since the distortion mode (or add your own fav) If properly promoted, it could add to both sales and userbase- as A:M is now officially bridged to the polygonal software packages. This will become the answer to many peoples problems.

[Vertexspline]

I think you are right John, The Poly world folks have kind of snubbed or ignored the spline world of AM despite its awesome animation capabilities ----now this may be a way for maybe both sides of the fence to enjoy more.

[Fuchur]

No doubt. This feature should prove to be the biggest modeling innovation to A:M since the distortion mode (or add your own fav) If properly promoted, it could add to both sales and userbase- as A:M is now officially bridged to the polygonal software packages. This will become the answer to many peoples problems.

 

Jep, it is... exactly what I thought of when asking for it. Steffen asked if he should include a more advanced algorithm for the normal-detection, so since it was much more work I didn't think it was necessary for now... maybe he could add it anyway, if you are having trouble...

For now: Keep in mind, that you can grab one point and move it a little to get the problem solved, if you are looking from the right side of the model. The point will slide on the surface and will very likely be placed where you want it afterwards.

 

See you

*Fuchur*

[jason1025]

This feature seams like it has so much potential. From an outside point of view it begs the question can we refine or evolve the feature so that you could lathe a dens cylinder around the prop and have the tool auto snap the cylinder to the prop. The top and bottom of the cylinder could be manually snapped to seal the hole at the top and bottum.

 

Steffen has a plugin that does something similar already with spline models. I forget what its called but its kind of like the old push plugin.

[John Bigboote]

This feature seams like it has so much potential. From an outside point of view it begs the question can we refine or evolve the feature so that you could lathe a dens cylinder around the prop and have the tool auto snap the cylinder to the prop. The top and bottom of the cylinder could be manually snapped to seal the hole at the top and bottum.

 

Steffen has a plugin that does something similar already with spline models. I forget what its called but its kind of like the old push plugin.

 

That sounds like a great idea, but UNfortunately- I think it defies how this feature works. From MY limited exposure, I have noticed that the 'Snap to Surface' tool- which is the underlying new mode behind the 'retopology' feature, only works in a PROJECTION manner... meaning- whatever surface lies behind the point(or group of points) to be applied as retopologized data, from your immediate POINT OF VIEW- will be where the points will be reassigned to. With this in mind, it is hard to see it working in a cylindrical manner. HOWEVER, I have not tested your idea yet. COULD WORK! but I doubt it.

[jason1025]

I think Steffen is creative enough and smart enough to figure that out eventually. I know its very easy to say coming from my point of view. The guy who does not know how to do it. But in my imagination I would think there has got to be a way to detect the mesh of the polys in 3d space and then constrain the cylinder around that prop.

 

I tired putting a deflector on a prop and then laying some cloth over it but although the prop accepted the deflector it had no effect against the cloth. I tried this on the mac version of AM. I will try again on pc.

 

If this were possible it might be an example of how some sort of detection could be applied and allow a cylinder to snap to the surface of the prop like cloth with heavy friction.

[Fuchur]

Have a look at the resurface plugin...

 

I think the better approach is to do it manually so... having to use a very dense mesh etc is just not good practice.

maybe for cloth or something....

 

See you

*Fuchur*

[Rodney]

While this will be something that should bring a lot of polygon users to the table the real beauty in this (from my perspective) is that splining itself has just gotten a whole lot easier. By using templates underneath as a helper to form our meshes we can spline surfaces faster and smoother than ever.

 

This is a lot like back when stitching was added to A:M but about 10x more extensible and versatile.

The old way worked nicely but the new way... way better.

 

v17 is looking SAWEEET!

[jason1025]

Have a look at the resurface plugin...

 

I think the better approach is to do it manually so... having to use a very dense mesh etc is just not good practice.

maybe for cloth or something....

 

See you

*Fuchur*

 

 

Yes but isn't the goal to be able to import a poly model and use it? The rotopoligy tool is designed to help expedite that right? So Why not try to go all the way if we can? Speed things up if its possible. I agree a dence mesh is not that great. Except for cloth.

[Rodney]

So Why not try to go all the way if we can?

 

I'm sure Steffen will when he can.

In the meantime some of us plan to use the heck out of this feature.

 

 

BTW - I'm pretty sure Matt meant v17Alpha2 up there. I don't think the Beta cycle has yet been released.

[Fuchur]

Yes but isn't the goal to be able to import a poly model and use it? The rotopoligy tool is designed to help expedite that right? So Why not try to go all the way if we can? Speed things up if its possible. I agree a dens mesh if not that great. Except for cloth.

 

I think that it would be great to have that but that it is not very likely to happen (in near future). We striked for something like that for years now and it could not be made possible... so I searched for something that will make it possible in another way (without automatic polygon-conversion, because this is more or less un-doable) and this is a very nice way to go about that.

 

Even other software like 3d coat has gone this route and they are not even using patches but polys and it works well... they have a auto-retopology-tool too, but for me it is not a good way to strike for unrealistical aims, because that will not happen in the short term and depress everyone who is dreaming about such things. Step-by-step is the best way in opinion and for now, this is the way to go. Who knows what the future brings...

 

And no matter what will happen: The best way is to do it manually, because it will create the best topology for animation and for the use in A:M (the computer just can't know what you plan to do with the model) and it just is now much easier to do it than before. I have the next step in my mind, but I will not talk about it till I know it is doable... just doesn't help to rise the expectations if you can't fullfill them.

 

See you

*Fuchur*

[Vertexspline]

Fuchur,

I think the current planned option makes it best for AM'ers- who would like some helpful modeling tools. I think it is gonna be so so helpful . A auto tool although a great time saver perhaps does not advance ones splinnage skills either as much. So given limited resources for developement ---I think what we are going to get is awesome.

[Fuchur]

Fuchur,

I think the current planned option makes it best for AM'ers- who would like some helpful modeling tools. I think it is gonna be so so helpful . A auto tool although a great time saver perhaps does not advance ones splinnage skills either as much. So given limited resources for developement ---I think what we are going to get is awesome.

 

Exactly what I thought .

And for all of you who want to see how it is done, I have created a small video-demonstration of the new feature...

Keep in mind: I am not a native speaker, but I do my best .

 

Link to Video-Tutorial: Retopology

 

Best wishes

*Fuchur*

[largento]

Noting that you can do this to existing patch models, it occurs to me that this would also be great for creating low-patch count substitutes for heavy models.

[John Bigboote]

BTW - I'm pretty sure Matt meant v17Alpha2 up there. I don't think the Beta cycle has yet been released.

 

 

Good catch. Fixed it, thanks Rodney!

[John Bigboote]

One thing that did occur to me, as later I was trying to do more of the face- is that you will still need good modeling skills. 'Tracing' the polygonal base is a great start, but it will only get you so far. You will still need a plan, know how and when to use 5pointers, hooks etc. This will not automatically make great modelers out of newbies... the challenges of patch modeling are still right there.

[NancyGormezano]

I have created a small video-demonstration of the new feature...

 

Wonderful new feature! Well done tutorial!

 

Will make it easier in creating paths for animating models, or special effects moving across rugged, irregular terrain or contours of other models.

 

Love it. Sure to open up many new, yet to be thought of uses.

 

EDIT:

 

Another possibility is to use a "rough 3D roto/proxy model", ie 3D roto built in A:M, composed of unconnected spheres & simple shapes, and then use that to create a more detailed connected spline model.

Edited December 11, 2011 by NancyGormezano

[Fuchur]

I have created a small video-demonstration of the new feature...

 

Wonderful new feature! Well done tutorial!

 

Will make it easier in creating paths for animating models, or special effects moving across rugged, irregular terrain or contours of other models.

 

Love it. Sure to open up many new, yet to be thought of uses.

 

EDIT:

 

Another possibility is to use a "rough 3D roto/proxy model", ie 3D roto built in A:M, composed of unconnected spheres & simple shapes, and then use that to create a more detailed connected spline model.

 

Jep, have thought of that too. Like a manual "meta-balls" or "z-sphere"-approach... another great example is, how easy it is to create cloth with such a feature and so on... many possibilities!

 

See you

*Fuchur*

[Fuchur]

One thing that did occur to me, as later I was trying to do more of the face- is that you will still need good modeling skills. 'Tracing' the polygonal base is a great start, but it will only get you so far. You will still need a plan, know how and when to use 5pointers, hooks etc. This will not automatically make great modelers out of newbies... the challenges of patch modeling are still right there.

 

...yes, that is part of the cool thing and for inexperienced users a bad thing about it. You need to know how to model and how to layout your splines... like that you will create very nice topologies, which are easy to animate... and that is the part that a computer can't do (easily) and why it is necessary to involve the user...

 

See you

*Fuchur*

[jason1025]

thanks for the tutorial

[Malo]

Thank you to show us the new retopology tool.

[Robert]

Wow, it is a great new feature.

 

Thanks to Steffen for the developement and to you for the tutorial.

[Fuchur]

Wow, it is a great new feature.

 

Thanks to Steffen for the developement and to you for the tutorial.

 

No biggy guys.

I had fun creating the tut.

 

See you

*Fuchur*

[pixelplucker]

That looks pretty cool, what kind of poly counts can it handle? I do a lot of work in 3dCoat and a direct export of the voxel would be sweet but unlikely AM can handle a few mil poly stl.

[Fuchur]

That looks pretty cool, what kind of poly counts can it handle? I do a lot of work in 3dCoat and a direct export of the voxel would be sweet but unlikely AM can handle a few mil poly stl.

 

The limitation is the prop-importer. Using Millions of polygones is quite heavy, so I doubt that would be possible, but if you could provide me with a few test-models (for example one with around 10k of polygones, one with 100k of polygones, one with 1000k of polygones), I could try it out.

 

See you

*Fuchur*

[pixelplucker]

Can you import stl files or just want a plain ole obj?

[John Bigboote]

Can you import stl files or just want a plain ole obj?

 

 

The V17 choreography prop import list is:

 

PLY

OBJ

LWO

DXF

3DStudio (autodeskLTD)

[Fuchur]

Exactly. Everything you can import as a prop can be used for Retopology-Tool.

STL is not importable into A:M in any way, so you would need to convert it first using another programm.

Anyway, since STLs are often used for 3d-scanning: If you try it with a 3d scan you will very likely run into problems there.

 

See you

*Fuchur*

[pixelplucker]

Here is an eagle model done in 3d coat. I exported approx poly counts, 100k, 50k and 2k. Anything below 50k really started to degrade in quality but I don't think anyone would really want to spline that much anyways.

eaglemodel.zip

[Fuchur]

Here is an eagle model done in 3d coat. I exported approx poly counts, 100k, 50k and 2k. Anything below 50k really started to degrade in quality but I don't think anyone would really want to spline that much anyways.

 

Hi!

 

None of these does even slow A:M down noticeable when importated as a prop and brought to the choreography... although the algorithm of resurface doesnt have any problems with the amount of polygones in the model... everything works fine and with no noticeable performance drop. Very cool!

 

See you

*Fuchur*

[Gerry]

Great tut, Gerald, thanks! this is a great new tool, can't wait to try it.

[pixelplucker]

Nice!

[Heiner]

Hi there,

 

the new Retopo is very cool, and it really bridges the gap to the polygon world!

What i miss is a "snap to vertex" enhancement for the retopo.

 

Most polygon models have already a topological structure, which is by itself already a good guide

for a retopo in A:M. Having a "snap to vertex" would help very much with models which

are not ultra polygon heavy. For example, the head John Bigboote is using in his

tutorial here

 

http://www.hash.com/forums/index.php?showtopic=40992

 

would be remodeled in no time with a "snap to Vertex".

I am no programmer, but i guess the "snap to surface" was much more complicated to make

than a "snap to Vertex". Most likeley most of the logic and math needed is already in the

"snap to surface".

 

It would serve well in models with a average poly count, and the realy polycount heavy

models like they come from ZB or 3D Coat can be treated best with with the "Snap to surface".

 

Just for explanation why i got that "Snap to Vertex" into my mind:

I just had to turn a polygon model in a NURBS model using Rhino.

Snap to vertex was the most important fuction. And boy, while working on that,

how often did i whish to use A:Ms Splines modelling with a StV function!

 

Best regards

Heiner