ARCHIMATIX PRO Node-based Parametric Modeling for Unity [Unity Awards Finalist]

Artist logic vs programmer logic, Probably no 8yo artist touched turtle, they noped right away to the color crayon.

Also typing text coordinate is not conducing to fast gestural iteration like artist like.

But at the same time, they can just prototype on paper the gesture, then report back the final coordinate, that's how artist did it in TRON when there was no fancy 3d software back in the day.

 

While you were busy whinging, I made you a shovel in Turtle so you can dig a deeper hole.

 

But substituting text coordinates for parameter variables makes for extremely fast artistic iteration, doesn't it.



Seriously, SHOW US how this could be faster, or easier.
Click a button, type the variable name, and your script is parameterized.

Enough empty devil's advocate crap. I bring receipts. If you know a faster system SHOW IT.

 

It's call macromedia flash, you draw, it vectorized for you then you manipulate handle, and power user select anchor and type coordinate, some does it through action script.

 

I think this looks a lot faster and easier, and much more like an artists work flow then editing turtle script.

No one is saying turtle script is hard. Just tedious, and is kinda the opposite of the idea of using a visual graph to create your model when you have to resort to code like commands.

 

We're gonna have to disagree on your definition of "a lot faster". Sorry.

But it's not a fair comparison, parameters vs clicking dots.
I'll stick with Turtle.

 

Hows this. Ill make several things, and record the process. Then you record the process of writing the turtle script to duplicate what i draw with my clicking dots. We will time them both.

Well actually the question wasn't parametric properties, it was the ease and speed at which something could be created and altered in the artistic workflow. You said show you something quicker. I did.

But you wanted parametric's right? Well for starters it generates Turtle Script, so using one of your own examples you can easily go back, create the parameters and then place them where they need to be in the Turtle Script. But no need to muck with turtle script during the design process.

But wait, there is more! I mean, the previous example is only relevant if you make a two rather large assumptions. That it doesn't do parametrics. And that it cant export those parametrics to AX. Which both seem like silly assumptions given my level of knowledge of AX and dizzing array of nodes and tools i produce for it.

I just didnt think that was the question at hand, so no need to include them in the video.

 

Hi wetcircuit, had to chop a bit off your post due to complaints, and no the dude above you did not file the complaint it was from some random person mashing the report button. Try to be better behaved than me at least.

I'll be monitoring this thread going forward (for all parties), and will just do thread bans if people keep on against each other.

Thanks all.

 

@syscrusher - 1.1.8 just hit the Asset Store. It has the fix for the orphaned GameObject when clicking Combine Meshes. Also, if any orphans do show up in the Hierarchy, they no longer select the AXModel object when clicked.

 

@dnoparker - the wait is over... AX 1.1.8, just released, now has Bezier curves with broken handles!

 

That sounds like a lot of fun, in fact surprised I never checked out this popular asset.

 

I yesterday I had fun trying out Shader Graph for the first time. I am happy to report that a Material using a Shader Graph in the Lightweight Render Pipeline works fine!

 

<snip>

If you would like to have an actual conversations about the merits of a visual shape designer vs turtle script I would be more then happy to, however I'm not sure what there is to really debate.

Note: Apparently @wetcircuit post to which I am responding has been partially moderated while i was responding. So what you are seeing here, is what is no longer there. - This is not the case, my post was modded as well.

@hippocoder While I appreciate you doing your job. I almost regret the post in question has been moderated. I prefer to let peoples words and actions show their colors. Since the post in question was modded. If mine has to be modded as well since its a response to that, do your thing i guess, but i feel i have acceptably responded.

 

Stay on topic, I will hate having to use thread ban feature. Use PM or such to continue off topic. On-topic is discussing the asset.

Sorry and thanks for understanding.

 

THANK YOU <3 <3

Haha it's a really handy feature! thank you for taking my input ^ ^

 

It's definitely worth a look! I'm currently using Archimatix on a rather large environment for a cinematic project, and once I got past the initial learning curve the productivity has been outstanding. I'm proficient in Blender, but AX allows me to iterate geometry right in Unity's scene editor, and also to place Unity prefabs in arrays or linear repeaters (or combinations of both) without actually making the prefab geometry part of the AX procedural meshes -- which helps a lot with occlusion culling in Unity, especially on such a large scene as this.

EDIT: By "initial learning curve" I mean no disrespect to the product or its documentation. There is a fine manual (RTFM!) by @roryo and also a series of excellent tutorials from @wetcircuit and possibly others. But any of these procedural geometry tools, from Archimatix to ProBuilder to CSG to Houdini, is a large toolbox. You can watch all the tutorials and read all the manuals you want, but eventually you really learn it by diving in and making something, committing the usual stupid new user tricks (I've done my share!), and gradually acquiring a feel for the tools in the toolbox.

 

Thank you!

 

Test result, in case anyone is curious: Archimatix works perfectly fine with Substance Designer materials, using Unity 2018.2 and the newest Substance in Unity plugin from Allegorithmic. I know this is something one would assume would work, but I thought it might be useful to have a specific test report to confirm. The UV control settings for scale and offset also work correctly in this situation.

 

Hi @flyingaudio - One constraint currently on rounding is that a positive value for Rounding only rounds convex corners while a negative value rounds only concave corners.



I should add a checkbox to allow rounding both. In the meantime, you can round both by passing the shape into an Offsetter node and give the output of the latter a negative Rounding:



You're observation that only closed shapes can be rounded is correct. However, you can use a ShapeMerger node, subtract out a Rectangle and choose/use the DifferenceRail as the open shape output:



Using the open rounded curve in a plan sweep:

 

May I suggest a more versatile alternative? You already have logic that is aware of the difference between convex and concave corners. Instead of a checkbox for rounding both, why not expose separate parameters for convex rounding and concave rounding, with the expectation of both parameters being positive numbers in the future?

If you make new fields for these parameters, but leave the old field present-but-deprecated, your code could look at the old value to automatically migrate. If it's positive, put it into the new convex parameter and zero the new concave parameter. If it's negative, populate the new concave field with -1.0 * the old value, and zero the new convex field. If it's zero. After either of these actions, zero the old value to mark it as "has been migrated". Wrap this whole thing in a conditional that does a no-op for migration if the old value is zero. That gracefully handles the case where the rounding was previously unused, and also when it was used but you've already done the migration once and now want to never again automatically alter the new fields.

 

YAY! Happy to be judged on words and actions. Also loving how active this thread has become!

Now back to the party with another AX video tutorial! The BIGGEST one yet!

ARCHIMATIX: EXPRESSIONS and MATH FUNCTIONS

Learn how to write:

• Relation Expressions – modify node-to-node parameter connections
• Parameter Expressions – modify a node with complex math functions and custom parameters
• Logic Expressions – embed functions within the Turtle script
• Functions discussed: Ratios, Greater, Lesser, Range, Sign, Sqrt, Atan2…
• Create a Math Function Node that calculates Hypotenuse and the angle Theta
• BONUS: an introduction to Turtle Script in Archimatix (spoiler: it's not hard)
• BONUS: add Conditional Logic and bool switches to enable and disable logic functions

Hit [CC] to turn on captions, or follow the transcript here:
http://community.archimatix.com/main-forum/video-tutorial-expressions-and-math-functions/

I use TTS voices because they speak twice as fast as I do (and the extra effort of getting them to pronounce "archimatix" lol), which allows me to pack a 30min tutorial into a 16min video. As always, I try to show EVERYTHING step-by-step and explain each concept for beginners and pros alike, but it goes fast and the voices aren't perfect….

This video is longer than the others because EXPRESSIONS are the "brains" of parametric and runtime models. As @syscrusher said, the real learning begins when you dive in and try it for yourself

 

:/

 

Thanks for another tutorial! I listened to the first few minutes (I'll finish it after working hours). If I may offer some small feedback on the voices -- and speaking only for myself -- at the accelerated pace I find the female voice much more comprehensible than the male. It is probably because the ratio between waveform frequency and speech pacing (syllables) is higher with the female voice, allowing the ear to more accurately discern the word breaks, but that's just my off-the-cuff theory. TL;DR, I have a bit of difficulty understanding the male voice but not the female. YMMV.

 

I agree. He has a pleasant "enthusiasm" voice, but gets gargly sometimes…. I will look for a "plainer" male voice.

 

The speed lines are a nice touch. That does look FAST.

 

Thanks!! -- I'm glad you like them!






Some things to mention about this new visual design suggestion for the Freecurve:

Point List Context Menu:
So things like variables and draggable handles should be able to populate the "Point List" menu. This "Point List" menu could either be a right-click menu (appearing in the scene view) or displayed in the inspector (which would be a /little/ slower, but still useful enough).

Parametric Shape Entities inside "Normal" Shapes:
Also, "Groups" of "Normal Points" could be made into a single "Parametric Point" shape entity.
This would allow offsets and repeats of entire "2D Shapes" inside of a shape. I forgot to add rotation options and point origin offset to the image showing the right-click "Point List" menu, which would rotate around the current "Parametric Point" offset.

Rotations:
Things like circles and flowers could be created with a "Parametric Point" offset as the circle's origin with a rotation angle and an additional rotation amount (per repetition) and the "Parametric Point" repetitions property would simply repeat these according to the origin, offset, and rotation amounts.

Positions and Variables:
Each subsequent parametric point (P1 and brethren) can be tweaked visually (by mouse drag) and hierarchically (as long as at least one of the points that make up the "Parametric Sub-shape" is currently selected), and therefore their individual origins/rotations/repetitions and subsequent "offsets" for each repetition can be adjusted visually too.
These points can be given a parametric variable (or expression) as a "label" that can relate to some other parameter(s) in the AX Graph, allowing them to be fully controllable with draggable numerical fields (as in @wetcircuit's example of the "shovel").

Parametric Groups of Parametric Shapes:
The picture displays only one node point being a "parametric sub-shape" (with an "implied" end-point), but in actuality, any group of points could behave this way, and all of them could contain parametric shapes within parametric shapes with this model.

Disconnected Points in Shapes:
These simply have a checkbox in the RMB "Point List" menu for each individual point to determine whether it connects (or not) to the next point. Under the hood, this could operate as the DRW command in TurtleScript does to some extent (i.e. absolute positioning), but with all subsequent shapes that are not marked as "not connected" (with the checkbox) simply following the normal (relative) 2D shape path that we all know and love.


Plus, it is as "fast" to visually edit as you might imagine it could be (while still being relatable.)

 

Robot Kyle is glad that FreeCurve is finally getting its first steps toward parameterization! He wonders if the sudden development surge had something to do with todays, ahem, discussion theme....



You can now add parameter names in the X or Y fields for a point. When you change the parameter value in the Geometry Controls, all the points that call it have their values modified.



Also, When you drag a point in the SceneView, the values for the parameters are modified and ripple through the graph. For example, in the GIF below, the Cylinder's Translation is being modified as the FreeCurve point is being dragged.



I have been meaning to get to this feature for over a year and in the end thinking it would be a challenge, but it only took a couple hours. More to come....

 

What are we arguing about next week, Guys?
3D Splines or C# API? I could definitely come to fisticuffs for Lofter….

 

Hi @steveR - Thanks for posting this. I just did the tutorials for Shape Sprite. Really interesting. Has me thinking of all sorts of ideas for 2D parametrics.

 

Thanks for these ideas, @awesomedata. They raise the issue of a 2D shape editor that allows one to edit smart shapes without going to the node graph to use ShapeRepeater nodes, ShapeCombiners nodes, etc. Going through the Shape Sprite tutorials has been interesting because, though they are not parametric, one edits almost entirely in 2D SceneView with immediate feedback and play-mode testing.

 

No problem, @roryo -- I'm really glad to see you interested in my suggestion on freecurve parametrics. Let me know if you need any clarification on parametric repetition of point-groups and/or shape-rotation and propagation stuff -- I can offer a better write-up on any areas that aren't clear. My goal is to be able to recreate the AX library (such as the parametric staircases, pillars, flowers with alternating petal sizes while rotating, and rooftops with parametric relations on angles/widths/heights/repetitions/rotations) using a visual tool with immediate feedback.

Regarding implementation details:

The major thing you'll want to keep in mind is that you're only keeping track of groups (and hierarchical subgroups) of points and how they act or relate to the next point in line.
Some of these points in a group can be "implied" points (e.g. when a parametric point-group is selected by the user, labeled, and repeated from any other point (including any "free" or "floating" implied point.)
These parametric shapes should be able to be rotated/offset/etc any number of times on a per-point (or implied point) basis.
Changing the location/rotation/orientation or even the bezier's curvature of any repeated (implied) point changes the location/rotation/orientation of all other points in the repetition prior to it (but not after it -- since the repetitions are done hierarchically -- unless the user specifies it.)
This subsequent repetition would, effectively, be a "variant" prefab -- if you were to equate it to the new "NestedPrefabs" system Unity has implemented in 2018.3.
Please let me know if you need some further implementation details for any unusual technical scenarios you can think of or run into -- I've probably already thought of ways to subvert them and just forgot to mention them here.

Also, consider adding a node for "repeating along a point group" in a shape by adding a variable label name to a parametric "point group" you've selected, and then assigning that parametric "point group" label to a point in a shape. This would allow one to "mask" areas of a 2D shape and only apply repetition to certain sides/angles/portions of a 2D shape.

I believe Ferr2D is what the SpriteShape tool was modeled after actually, though Ferr2D itself was modeled after the UbiArt Engine for Raymand Legends and Rayman Origins (and other Ubisoft games.)
In the UbiArt engine interviews (and some GDC talks about it, they mention that they used a LOT of parametrics in Rayman Origins/Legends, so the technology is for 2D parametrics in this sprite context is not new at all -- However, applying those techniques to splines generating 3D models in the way AX does (while offering a better way to author parametric splines) in a high-level "easy-to-use" / "easy-to-understand" way -- THAT actually IS new -- at least in 3D modeling -- and is why I suggested it for usage in AX.)

 

Oh -- and don't forget about parametrically distributing "greebles" along certain (masked) portions of a 2d shape. ;D



In all seriousness though -- as an aside to you as a fellow designer, @roryo:


I really like the concept of AX. -- but I want to make clear that I've studied lots of art-creation tools over the years. I've found that the more they solve the pain-points of the overall iteration workflow (in terms of speed) for users to quickly accomplish their larger goal (in usually a novel but fast/effective way) and prioritize less the "novel feature" / "selling-point" aspects of the tool, the more well-received these tools become, even just through word-of-mouth alone (the best advertising money can't buy). Most of the time, a separate tool/editor is required that has to be written from scratch to accomplish such a task, but this tool extension is almost always worth the effort in the end due to the praise your users will give you for the value (in speed and freedom!) you've added to their tool (and workflows!) and the fact that your tool is invaluable for helping them accomplish their larger goal faster than anything else -- and they'll tell their friends/colleagues about it too.

I know how much of a pain it can be to be both the programmer *and* designer for something as abstract (in use-cases) as AX can tend to be, but the power of "designing tools for speed" should never be overlooked.
Zbrush is a very good example of a great tool with crazy use-cases -- it has a terrible UI (but that can still be used quickly) and a high learning curve (not unlike AX).

Zbrush looks (and acts!) nothing like other 3d modelers, BUT it does offer amazing speed boosts for users' workflows that make it worth the high learning curve (i.e. automatic controllable retopology, 3d vector displacement maps [with overlap for detailing sculpts quickly with things like scales/feathers/etc.], the ability to edit model geometry directly in Zbrush despite Zbrush "not being a traditional 3d modeling package", etc., etc.)
The developers keep adding unique workflow enhancements like the ones mentioned above to eliminate overall pain points for the larger problem of 3d modeling ( -- something AX already does well.) It doesn't let "scope" define its boundaries though -- If there is a pain point, then they design a way around it (which is not unlike Unity these days!) -- They can sit back and let the cash roll in -- but they don't -- and users respect that. Thanks to this approach, even very demanding users (like me!) can tolerate quirks in the interface or other weird issues (and even the occasional bug or three), just as long as it isn't too annoying or -- more importantly -- tedious or severe enough that is slows them down on the overall computer-design iteration process (e.g. losing work, for example, is a big no-no in most apps since that "design slowdown" is revealed in a seriously eggregious way).


Not trying to preach -- and many here may not like my opinions -- but I do feel it's worth pointing out these (very specific) things for any other programmers looking to (sincerely) undertake the burden of designing proper tools for artists...

 

An AXer wrote emailed a question about creating a tunnel from a self-intersecting Shape or spline. Here's a new short tutorial based on it:

 

Hi,
I'm not sure where to ask a question or address an issue, so I'll do it here. Please, let me know if there is more proper place to do it!

You can seen on screenshot, right tower of default model "GridliaCastle" has proper repeating boxes, while boxes of the left one are not in a rounded pattern.

I've noticed it first after doing youtube tutorial called "05 Archimatix - Repeat after me!", and my model has the same problem. Whats weird, it's always one tower having this issue
screenshot

It looks ugly and wrong. Please help!

 

Hi @anton88pro - I believe this was fixed a few versions back. What version of Unity and Archimatix are you using?

 

Today I am working on improving the FreeCurve editing work flow.



When you are in 2D Mode now, the drawing plane automatically goes to XY. Also, in 2D mode, calling up items from the Shape library automatically puts them on this plane. Alt-clicking a point repeatedly cycle converts the point through the three modes. Also there are now menu buttons for these modes a la SpriteShape. When you convert a point to Bezier, handles are automatically extended. Now to add menu items to allow you to fill and stroke the shapes, create Merge and Combine sets, all from SceneView.

 

I've just tested it (by creating GridliaCastle) in Unity 2018.2.8f1 and 2018.2.2f1 with Archimatix version 1.1.8 (on screenshots it was 1.1.7) and there is no change. I've created new project and it's still the same thing
I even tried the beta (2018.3.0b3) and here models didn't even show up (there were some errors in console).

 

Would it be possible to have the option to toggle the default drawing plane between XY and XZ instead (for topdown floorplan shapes?)

 

The anticipation of the desired drawing plane is one of the key features in this version. When you go into "2D" mode in the SceneView, the default plane is XY. When you go to orthographic Y-axis in the SceneView (looking down), the default drawing plane is XZ. When you are in 3D mode, it finds the plane that you are most closely looking toward to set the default view. These are a game-changer in FreeCurve editing. At least, Robot Kyle thinks so!

 

Well, this is a head-scratcher! I can't seem to replicate this. @anton88pro - could you email me a copy of the scene (not the whole project, just the scene where Gridlia Castle is misbehaving). I'd love to get to the bottom of this. Thanks!

 

I've used the contact form from Archimatix website to send the scene.
Thank you for trying to solve the issue.

 

Thanks for the scene - and the error has now reared its ugly head on my machine, er, I mean, now I have the bug reproduced! Rolling up sleeves....

 

A bug that reproduces reliably is always way better than one of those@#$#@ intermittent ones. Happy hunting!

 

Would it be possible to add in a fast way to create anglular relationship parameters? -- For example, to simplify creating parametric shape relationships for triangular roof pitches, widths, and angles (mainly for things like bridges and rooftops.)



In practice, the user would just select the points he wants a relationship with and click a button. The order of selection and modifier keys held upon selection would determine relationship direction(s). The user would then click a button at the top of the scene view which would populate/replace the text in the input fields with the math (and variable names) for these relationships.

I tend to have to look up the particular formulas when I want simple mathematical relationships between my points (such as pitch/height/width/angles). This would let me populate the common relationship formulas visually.

 

@anton88pro - the bug has been fixed! It will be in the next version, due to hist the AssetStore early next week. Thanks for reporting this - it was a tricky one!

 

i love archimatix, i made the monastery in my new game entirely in AX and it works great

Just to chime in, softly, on the turtle vs visual debate - i think people work differently. For some people typing in turtle is faster. And for other people like me, a visual design tool is faster. One isnt objectively faster than the other, and its nice, when possible, if both methods are available. My firends an engineer and all he does is keyboard shortcuts. Whereas i almost never touch the keyboard unless i have to and use the mouse almost exclusively.

 

Nice lighting work on this! Care to share which render pipeline you're using, and whether realtime or baked?

 

Wow. This game looks good. When there's some action added to the game record a clip and show us. It will be a good way to show off what Archimatix can do in an actual game situation.

 

Thanks Its the default (built in) render pipeline, and its real-time lights.

 

I do have some shots of running through one of the hallways

And a screenshot of some of the architecture made in AX. My favorite is the lathe making a bunch of cool rounded shapes