Jump to content

Michael Goldshteyn

Regular Member
  • Content Count

  • Joined

  • Last visited

Community Reputation

0 Noble Beginner

About Michael Goldshteyn

  • Rank
    Cafe Ronin

Profile Information

  • First Name
  • Last Name
  • Location

Cinema 4D Information

  • C4D Version
    R20.026 Studio
  1. I just saw a technique for converting the triangles of a cylindrical caps to quads that I hadn't seen before. Specifically, the modeler used the technique pointed to by the red arrow in the image below to convert the central triangles into quads. He basically created a zig-zag starburst pattern, which cut the number of radial edges in half and created quads as a result. Usually, I see people creating quads by just removing every other edge from the radial edges that go to the central pole. The image above could be converted to this form by taking those points around the central point that are closest to it and pushing them away from the center until they are the same distance as the next farthest set of radial points, producing something along the lines of the image shown below. What are the pros and cons of these two modeling approaches with respect to each other - a starburst like pattern as in the top image vs a more conventional circular arrangement as shown in the bottom one.
  2. Let me prefix by saying that these question are specific to and centered around Cinema 4D (R21/R22): It seems that coping with initial topology complexity as well as changes to existing topology, especially when adding "Yet Another Element of Detail To a Curved Surface," takes a very significant amount of time and effort in the context of subdivision-surface modeling. In addition, I realize more and more that keeping things as procedural and non-destructive as possible for as long as possible is truly the ultimate goal. Having learned as many subd patterns as one can find on the web and books, I am left to wonder if there is a certain ordering of steps, techiniques, or even free/paid plugins, that experienced Cinema 4D modelers would advocate to help ease the pain (e.g., move the surface you are working on to (and align it with) the origin while working on it, bevel last, add detail only where necessary, etc.), but what tools and or techniques do you guys use to automate and make procedural as much of the subd workflow as possible, so as not to have to manually wrangle with never-ending edge(-loop) geometry in an effort to maintain quads and avoid shading artifacts due to n-gons/triangles, in light of model complexity and change. I realize that this question is somewhat general and open ended. I am certainly not expecting a silver bullet reply, just the sharing of tips, ideas, and workflows that, with experience and learning, you have found help expedite your SubD surface workflow. I am certain that any experienced SubD modeler has discovered techniques, aids, and perhaps plugins that they find indispensable. By sharing these, perhaps we can learn from each others' findings and enhance our own abilities. I will start by mentioning some of my tips (and do bear in mind that edge cases where one or more tips are invalid, do exist, [i.e., don't take any tip or set of tips as gospel]) : Use snaps, don't "wing it." In almost all subd educational videos, the so called experts eyeball movement and alignment of points and edges, instead of using snaps if and whenever possible. This often leads to non-planar faces, shading artifacts, stray points/edges, and other anomalies down the road. Also, be mindful of world coordinates as you model and adjust your topology. Boxing out your object (use a picture if and whenever possible as the background - one of the best pieces of advice right there) and go from featureless boxes (and box holes) on top or in other boxes, as a starting point. Then add detail. Combine this with the next piece of advice. Bevel last and keep flat surfaces free of detail (yes as NGons) for as long as possible. I will add to this that you should limit the addition of control edges to bordering geometry for as long as possible as well. You can always subdivide, cut, and use the Bevel (in solid mode) to connect and control edges later, adding edges where and when you actually need them, once other surrounding geometry finds its final place. Breaking up large flat NGons before you need to, adding control edges early on, only to have to move, adjust, and perhaps even remove those edges later, can really complicate details that always (perhaps inadvertently and often unexpectedly) come into contact with said edges. In other words, model and isolate (details on the inside) first, connect last, (also, from the inside out). The inner details should lead the outer topology and not vice-versa. Way to often you see people going outside-in, only to have to go right back to inside out when they realize they made a mistake on their outer topology while trying to model the inner details. Try to contain inner details via loops and not let them propagate out (along a flat surface which hopefully you are keeping as an NGon for as long as possible). Tie as many parameters that should be shared among objects (and especially parts of objects) as possible, preferably to a master object that you create and use to affect the parts. For example if you have a plane with many holes, break it up into many identical (little sub-plane with single hole) objects - one master, and the rest instances. It is best to keep the master at the origin, if possible and work on it there, while the clones, perhaps in another view if its far from the origins, get distributed and aligned to where the copies need to be. Work on the topology of the master object, while the clone objects stay in sync, instead of making the same changes repetitively to each and every one of them. If you want to see results as you go (e.g., via subd effects, etc...), place the instances into a Connect and the Connect into a Subdivision Surface. When you are done, preserve a (backup) copy of the master object (and possibly the clones as well, to retain their locations and individual properties distinct from the master) and flatten the Connect into a single editable object. Use symmetry whenever possible! This is one that even the most experienced pro modelers tend often overlook. Always keep an eye out for symmetry, mirror/radial/or even ad-hoc, and use a corresponding (minimal) Generator to limit the duplication of the components of a symmetric sequence of parts. From least overhead and (modeling time) effort to most, use {Symmetry, Array (for radial symmetry), Duplicate w/instances, Cloner, Sweep/Extrude [along/of a spline that (procedurally) defines the path of geometry}, and manually positioned instances that either sit in a Null with the master or are children of the master}. Learn the terms under the "Modeling Axis" tab of the Move/Rotate/Scale tools and when to use which for the Axis, Orientation, and Along Normals checkbox. Learn to use guides and nulls in combination with Snapping to help you position and align. Avoid snapping in the perspective view and 3D snapping in general. Try to use one or more of the 2D Orthogonal views to do all your snapping. Also, don't forget about the measure tool, when you are unsure of or want to ensure certain angles and/or distances. Name early, name often. As soon as you add anything to the scene, give it an appropriate name. This should be used in combination with the next bullet point, best summarized as "Group early and group often." Make your hierarchy deeper rather than longer. Try to group things whenever possible in an effort to conserve vertical screen real estate in the Object Manager. Use groups to manage details and provide a high-level view. If the model gets complex, take advantage of Layers (and Layer colors affecting objects!) to orthogonally (with respect to the aforementioned groups) reorder, (re)group, and organize the parts of your model. Take advantage of Python to help automate repetitive tasks. This is probably the most important bullet point of them all, and for many the most difficult due to the lack of good examples of actual workflows (regardless of what MAXON's website seems to claim/imply is available on the web). Use a template Layer to store and organize often used objects, of different shapes, sizes, and with various numbers of segments you find useful and/or frequently encounter in your modeling workflow. When starting a new parametric shape, duplicate (or import from another scene) one of them instead of starting from scratch with a default parametric built-in that you have to readjust over and over again to become the one thing you need out of it. Don't just limit yourself to Bevel, Extrude, and Inner Extrude. Very often, there are other ways of accomplishing the same task that are far less dependent on (and affected by) neighboring geometry, especially when you want to perform said functions independent of neighboring geometry. The details on this one would require an entire two to three hour long instructional video which I hope to make at some point hopefully in the near future. Those are just a few of mine and I'll add more to this thread in the coming days as I think of and/or encounter them yet again while modeling. I'd love to read your ideas/tips, as well.
  3. You made some very good points in your post about C4Ds differences from a normal Windows app. Thank you!
  4. Has anyone found a setting in R21 that allows for the Alt key to be used to open menus like in other windows programs. This would allow, for example, default Windows Application behavior like "Alt-F" to open the file menu and then "S" to save. Thanks, Michael
  5. I guess for a parametric spline, there really is no way to zero out those angles until it is first made editable.
  6. Thanks guys, I am aware of this for editable splines and although I explicitly said that the spline is editable in my original question (definitely a mistake!), my intention was to be able to do this with a parametric spline object, such as a circle or rectangle for example, where using Axis Mode to modify values is not an option, since Axis mode does not work for parametric objects. Sorry for the confusion.
  7. Updated: Crossed out the editable portion and changed it to non-editable, since with an editable spline, Axis modification makes this trivial. I start by creating a planar spline shape (let's pretend it's a circle). The spline lies along the XY plane - you could say it is vertical if Y points up (its Z size is therefore zero). The spline is editable not editable, but that probably doesn't matter, except to say that any re-orienting will be done via the rotation tool (or by modifying its HPB Values via the Coords attributes panel). OK, so the goal is this: I want the spline to lie on the XZ plane - making it horizontal. "OK, no problem, " you say. Just do a 90 degree rotation in Pitch (i.e., around the X axis). Now, after said rotation, the spline lies along the XZ plane, as desired and its Y size is now zero. But, the kicker, I want to reset its HPB rotation values of the spline in this new orientation to zero - my goal is as simple as that! Resetting the PSR at this point, to accomplish the goal, will just make the circle spline vertical again. I guess this is expected functionality, since resetting the PSR also changes the position, scale, and rotation of the object being reset as part of said reset. Playing tricks such as creating a counter-rotated Null object, placing the spline inside of it, and trying to reset the PSR of the spline and/or the Null, and pulling the spline back out, also gets us nowhere - either the circle becomes vertical again with a zero Pitch angle, or horizontal, but with the dreaded -90/90 degree pitch angle. I finally came up with a solution which went along the lines of placing the horizontal spline in a counter-rotated Null, putting a protection tag on the Null, clearing the PSR on the circle (or maybe I rotated it by hand), and then pulling it out, and this finally got the result I wanted - a horizontal circle spline with zeros for all of its rotation Coords in its new horizontal orientation. I should also note here that doing the opposite, that is putting the spline inside a counter-rotated Null, so that the spline's HPB values once inside the Null get zeroed out, then putting a protection tag on the spline in the hopes that one can then do whatever one wants with its parent Null without modifying the Spline's HPB values or its orientation, does not work! There has to be an easier way to do this for an arbitrary spline! That is, to rotate it into some orientation and then reset its rotation Coords back to zero in this new orientation. I would love any tips on an easy way to accomplish this, without having to go through the Null/Protection tag method I used above, which seems overly onerous for something this simple.
  8. I am sure many of you use the Commander (Shift-C) to find a particular command. The Commander is great for searching for a command and its shortcut. Is there anything in C4D that can help identify, given a command, where it can be found in the new "more intuitive" rearranged R21 menu system? It sure would have been nice if the Commander supplied this info along with the shortcut - the location of the command in the Menu Hierarchy. Thanks for any tips
  9. Another idea: Use a volume effector to hide intersecting cloned objects using its Visibility parameter. I will have to try this and it may solve my problem without having to create a scaled-down version of the "e".
  10. Wow, good one!! It preserves the detail of the original and all. I will have to try it out. My Idea #1 was: Use Boole to carve the "e" out of a sphere (i.e., sphere minus "e"). Then put the cloner cubes constrained to the "e" object inside that hollowed out space. Of course some of them would protrude outside the "e" and into the "sphere minus e" volume. Then, use dynamics. Since those cloners that protrude from the volume of the "e" (which is now a hollow part of the sphere) would overlap the Boole volume (i.e., sphere minus e), dynamics would push them inside the hole carved out by the "e" in the sphere. This should result in all cubes being inside the "e" volume after some number of steps. Then remove the Boole or hide it, since it is no longer needed. Now all cloner cubes should no longer protrude from the "e" volume. Those that did would have migrated inside being pushed in by dynamics. Idea #2 was without dynamics is to use the "Push apart" effector to hide those cubes that intersect the "sphere minus e" Boole created in Idea #1. This would also have the benefit of spacing out the cubes more evenly and avoiding overlapping cubes inside the "e" volume - an almost perfect solution to the whole thing, with the caveat that many cubes would get removed (due to overlap with each other or overlap with the Boole). I know, it sounds far more complicated than your solution, but it is a general solution that would work for all reasonable meshes including ones that do not start off as splines.
  11. I am not sure how I can do that. It's not just a matter of simple scaling, since for this to work correctly it has to be done from the surface of the letter e inward towards an imaginary spline running through the middle of the e's volume (i.e., the hole in the "e" expands, while the outer shell contracts, making the "e" in effect, thinner since we're scaling from the surface inward). I hope that description got the point across. Otherwise, I will have to draw a picture to describe it. How in the world do you auto-scale like that?
  12. When using a Cloner with a Volume distribution constrained to an object (let's call it the enclosing object), is it possible to constrain the objects being cloned to be totally inside the volume of the enclosing object without poking out? It seems that only the cloned objects' centers get constrained and so the meshes of clones near the surface of the enclosing object tend to poke out through that surface. I would the entire meshes of the clones to be inside the enclosing object's volume, so as not to protrude out of that volume with any of their respective parts. Attached is an example image of the problem. Some of the cloned white cubes that are supposed to fill the volume of the "lower-case e" object are protruding through its surface. Their centers are properly constrained, but the rest of their mesh is not:
  13. I think this is similar to the halo seen when Unsharp Mask is applied in Photoshop.
  14. Is this possible with AMD ProRender as the renderer?
  15. I would like to know if AMD ProRender supports full spectrally correct dispersion of light beams going through glass (e.g., through a prism) in a similar vein to other spectrally correct renderers such as Octane/Maxwell/Indigo/etc... aberration in renders based on Abbe value among other things (e.g., thickness of object, its shape and especially its curvature).

Latest Topics

Latest Comments


  • Create New...