OK so there's multiple things going on here. What you have here is a cylinder. Call it a quasi-, compound-, or even pseudo-cylinder, but the same rules apply: If you want it to be cylindrical, you're not allowed to mess around with the edgeflow, or amount of edges connected to any vertex without consequence.
The blue and green arrows here show unproblematic edgeflow which is what you want. They criss-cross each other nicely and are evenly spaced where it matters so you get what you want. The red edge is your enemy. First it creates a 5-pole at it's forward vertex, which on a curved surface messes with the consistency of all neighbouring quads, and it creates a triangle highlighted in red and yellow. Two bad. Triangles are converted to quads with catmull-clarke, but the "mesh density" increases in the area which disrupts the curvature within and around it, creating a sort of pinch point. It also turns those consistancy-bearing and criss-crossy blue and green edgeflow arrows into an orange L shape in a curved area. This is bad.
Aaahh. Much better. After fixing that, let's take a look at edge sharpness consistency. It can take years and years of practice to notice inconsistencies in supporting edges, so I thought I'd share a method I use. I select the edge (blue) I want to keep consistent, and simply have a lil looksie-pie at the criss-crossing edges (green) on both sides. The blue is sort of like the body of a centipede, and the length of it's green legs decide how far the blue edge is allowed to bleed. There are three centipedes here, your perfect form in blue and green, an orange centipede with yellow legs, and a red centipede with pink legs: Make yellow and pink equal lengths somehow. Something like this, but better, IDK:
Do notice that I messed with the density in that compound curve, which is not good. It takes too long to draw this shit in MSPAINT so I'm defaulting to telling you not to do that. So you shouldn't do that. I usually avoid it by detatching the geometry I need to double, subdividing it, and welding it back on, modifying the surrounding geometry to fit. I do think the low forward-to-rear-mesh-density on the compound curve is the main reason you have difficulties.
I used that method to get the curve I needed on the hammer of this: http://skins.thanez.net/357/marmoset/ I had decided to model in every detail as a challenge, and didn't have enough/correct geo to carry the top curve, nor enough base geo to build the ridges, nor casting lines from. Uhh.. Bye.
Adding to what Thanez has explained: sometimes it's beneficial to reduce unnecessary mesh complexity by using the minimum amount of geometry required to hold the shapes when the mesh is subdivided. This mesh simplification makes it easier to solve loop routing issues earlier in the modeling process. Focus on creating clean, well organized loop paths around the larger shapes first and it will be a lot easier to add support loops later with a bevel / chamfer operation.
Below is an example of what this approach to subdivision modeling could look like: start blocking out the basic shapes by adjusting both the number and position of the segments on intersecting curved surfaces. This way the edge spacing along the curves will be relatively consistent and aligned with the edges of adjacent shapes. Which provides a lot more options for adjusting the basic loop flow and support loop routing.
In this example the shape support loops are only there to support the transition areas between the larger shapes and the edge support loops are only there to sharpen the hard edges of the adjacent shapes. It's worth noting that, on this particular type of shape, it's often possible to redirect or terminate a lot of edge support loops with diamond quads. Which prevents them from running out onto curved surfaces and disrupting the segment spacing of the curves.
Having solved most of the loop flow issues earlier in the modeling process: the edge sharpening support loops can be added non-destructively using a bevel / chamfer modifier and any minor smoothing issues can be cleaned once the entire slide is modeled. Which further simplifies and streamlines the modeling process. Sometimes overly dense meshes or complex manual loop routing doesn't make sense in terms of time spent Vs results obtained.
Although the mesh used in the example below is
fairly simple, it still strikes that balance between softness and
sharpness that allows for a good visual read while also maintaining
relatively accurate shapes and shape transitions. All while remaining easy to edit. Subdivision modeling
is all about figuring which tradeoffs in accuracy and simplicity provide
usable results while still keeping modeling times reasonable.
One of the downsides to both subdivision modeling and high to low poly normal baking is that certain types of sharp edge to curve transitions will have a broader highlight than what's found in real life. This is just the nature of the medium and it's often necessary to accept that certain types of shape transitions cannot be both perfectly sharp and visually readable at a distance. Which is why it's generally considered best practice to view the mesh from multiple angles both up close and far away to get a sense of what's an acceptable level of softness for the model's specific use case.
The loop routing around this tighter triangular edge of the shape transition could be further simplified by dissolving one edge of the loops being redirect by the diamond quad. While this approach does produce an n-gon it's not a significant visual issue because it's constrained by the adjacent loops and will be resolved back to quads once the subdivision is applied.
Not only does this alternate approach to loop routing simplify the mesh but it also produces a more even edge highlight along the adjacent surface where the redirected edge loop was removed. This is because the edge width of the support loop is now consistent across the larger underlying surface.
Unless there's a visible smoothing error, surface quality issue, or valid technical reason for avoiding them: it's generally acceptable to use a few triangles or n-gons to simplify the mesh. Especially when they're constrained by adjacent support loops or on flat surfaces and the mesh still smooths well when subdivision is applied.
Below is an example of what the simplified mesh looks like at 2x subdivision level.
There's a number of variations on the fillet path along the front of this slide. Some are more complex than others and how much geometry is required for the basic curvature depends entirely on the complexity of the shapes in the references. For a basic fillet and radiused corner, that doesn't have any intersecting rail cutouts, a simpler mesh and loop routing strategy will probably make modeling this shape a lot easier than throwing more geometry and support loops at it. Having that flat side wall between the corner and the round over at the bottom of the slide also makes things a lot easier.
Recap:
-Simplify the mesh by using the minimum amount of geometry required to support accurate shapes and clean surface transitions.
-Focus on solving the topology flow issues around the larger shapes first and try to use the existing geometry of adjacent shapes are part of the support loop routing strategy. This will help with editability and make it a lot easier to sharpen the edges without distorting the underlying shapes.
-Whenever possible, redirect or terminate support loops before they run out into the curvature of adjacent shapes. This will help prevent unintended surface deformation and pinching smoothing artifacts.
Hi. how do i model this crystal basin pattern ? i have searched everywhere for a few patterns to build it with heightmap or displace. but the outcome is not even close. need help.
Look into voronoi meshing structures, might help get closer to that semi-random cellular structure.
isn't it easier to design a heightmap in substance designer ? i tried to do a part and make it tile but its very hard to keep it this level undetected. its so hard to find a repetitive pattern. seems like fractal Voronoi.
I got kind of close. This is with the Generate Topology tool in 3ds Max, which makes cellular kind of shapes.
I think the inside surface of the sink is actually smooth, and the outsides are where all the gemstone-like cuts are, only makes sense if it's wash sink. I didn't bother with that part though.
Hello guys, I am struggling with this train front part, especially because its sort of organic hard surface with cutout. I have tried block modeling, subdivisions, splines, which produced insane hight ammount of face, lines, alas, with no success. It might seem trivial, but I just cant get my head around that shape. I almost got that shape, but cant connect the roof with the front side, and the cotout is a lot problematic for me.
Posting my trials, what I am aming for, as well as link to webpage with more images for refference.
Hi. Is there a short cut to convert the stepy-edge's to a circular shape in edit poly mod ? i started with a cylinder and then i focused on a quarter of it. modeled the first quarter and then using symmetric modifier i got it to this level. since i have to continue the inside part and its flat, i have to convert the last edge to this round shape. is there a short-cut to make it soft ? i think im also able to use "shapes" to make the insite part. (yes) but couldn't be better to model this all in one part ? also let me ask, is there an easier approach to the out part ?
I had to cheat a bit with an Edit Poly, to get a top bevel, separate the drain area, and fix a bad vertex. I hate using Edit Poly because it breaks parametric editing... Edit Poly is vertex-count-dependent so if you edit stuff below it you're more likely to break things.
Anyhow, you can go through the modifiers one-by-one to see what I did. Just turn off "Show End Result" to see the steps.
thank you so much for the info and the trying it ! i got it. i used an eclipse, duplicating and scaling it to match the final shape, then i created a surface, then i customized only a part and using symmetry it just fit really nice together. love this section. very informative post. well done everyone.
Replies
Mario Elementza - YouTube
What you have here is a cylinder. Call it a quasi-, compound-, or even pseudo-cylinder, but the same rules apply: If you want it to be cylindrical, you're not allowed to mess around with the edgeflow, or amount of edges connected to any vertex without consequence.
The blue and green arrows here show unproblematic edgeflow which is what you want.
They criss-cross each other nicely and are evenly spaced where it matters so you get what you want.
The red edge is your enemy. First it creates a 5-pole at it's forward vertex, which on a curved surface messes with the consistency of all neighbouring quads, and it creates a triangle highlighted in red and yellow. Two bad.
Triangles are converted to quads with catmull-clarke, but the "mesh density" increases in the area which disrupts the curvature within and around it, creating a sort of pinch point.
It also turns those consistancy-bearing and criss-crossy blue and green edgeflow arrows into an orange L shape in a curved area. This is bad.
Aaahh. Much better. After fixing that, let's take a look at edge sharpness consistency. It can take years and years of practice to notice inconsistencies in supporting edges, so I thought I'd share a method I use.
I select the edge (blue) I want to keep consistent, and simply have a lil looksie-pie at the criss-crossing edges (green) on both sides.
The blue is sort of like the body of a centipede, and the length of it's green legs decide how far the blue edge is allowed to bleed.
There are three centipedes here, your perfect form in blue and green, an orange centipede with yellow legs, and a red centipede with pink legs:
Make yellow and pink equal lengths somehow.
Something like this, but better, IDK:
Do notice that I messed with the density in that compound curve, which is not good.
It takes too long to draw this shit in MSPAINT so I'm defaulting to telling you not to do that.
So you shouldn't do that. I usually avoid it by detatching the geometry I need to double, subdividing it, and welding it back on, modifying the surrounding geometry to fit. I do think the low forward-to-rear-mesh-density on the compound curve is the main reason you have difficulties.
I used that method to get the curve I needed on the hammer of this: http://skins.thanez.net/357/marmoset/
I had decided to model in every detail as a challenge, and didn't have enough/correct geo to carry the top curve, nor enough base geo to build the ridges, nor casting lines from.
Uhh.. Bye.
need help.
but there seems to be some additional crisscros (i tried in on direction)
Maybe it's even more complex. Or.. some opticial illusion.. The baseplate is a mirror !!... but i'm too lazy to look at this after i tried all this
I think the inside surface of the sink is actually smooth, and the outsides are where all the gemstone-like cuts are, only makes sense if it's wash sink. I didn't bother with that part though.
Posting my trials, what I am aming for, as well as link to webpage with more images for refference.
http://www.drah-servis.cz/detail.php?typ=dokumentace_histcr_skoda_s699
Is there a short cut to convert the stepy-edge's to a circular shape in edit poly mod ?
i started with a cylinder and then i focused on a quarter of it. modeled the first quarter and then using symmetric modifier i got it to this level.
since i have to continue the inside part and its flat, i have to convert the last edge to this round shape. is there a short-cut to make it soft ?
i think im also able to use "shapes" to make the insite part. (yes) but couldn't be better to model this all in one part ?
also let me ask, is there an easier approach to the out part ?
There’s also a tool in the Graphite ribbon, to make the edges evenly-spaced, maybe under Loop tools?
is there an easier approach ? like making using world path modifier ?
I had to cheat a bit with an Edit Poly, to get a top bevel, separate the drain area, and fix a bad vertex. I hate using Edit Poly because it breaks parametric editing... Edit Poly is vertex-count-dependent so if you edit stuff below it you're more likely to break things.
Anyhow, you can go through the modifiers one-by-one to see what I did. Just turn off "Show End Result" to see the steps.
i got it.
i used an eclipse, duplicating and scaling it to match the final shape, then i created a surface, then i customized only a part and using symmetry it just fit really nice together. love this section. very informative post. well done everyone.