Blender vs Catia and other cad programs

To introduce my self, I am a mechanical engineer from Amsterdam.

In my profession, I make technical drawings, 2d and 3d.

I have experience with multiple cad programs, both 2d and 3d.
I also once build maps for a game, Aliens vs Predator 2.

Some drum I made inCatia:

After watching some tutorial vids of blender I get confused.

Nobody in blender seems to care about the actual size of things ?

I also watched a vid of a guy making a coffee cup.

I totally don’t understand why he does not draw a cross section of that cup, and then use a commando like revolve, to make the 2d sketch a 3D object. ( he does this to make the handle of the cup, he send a circle along a path )

Then his way of connecting that handle to the cup. :confused:
Is that really the right way to do it in blender?

In a cad program I would have used a boolean operation, which I already have seen in blender.

From reading online I think I do understand that blender does not use what cad program calls solids.
I think I also understand why, nobody with blender is interested in stress calculations, so solids are overkill for a program like blender.

But did I pick the wrong tutorial (s) or, do I need to learn a new way of thinking and 3d modeling.

The drum in the Catia picture, can this be done in Blender with using actual dimensions ?

Like the drum is 2000 mm ( round ) and a height of 2000 mm, and the the cut out groove 20 mm in height ?

To end this positively. I tried blender before, and gave up rather soon, because that interface drove me mad.
My first impression of the newer blender is much better.

I installed blender on opensuse leap 42.2. (linux)

I managed to have Blender start with milliliters. Its a start.

There seem to be some plugins that give blender some more cad program like sketch options.
I will have to look deeper into those.

From what I read around the web, a few plugins might solve my problems.

I am also gonna have a look at the 3D max interface.
I have used that program, so it might be more familiar.

With 3d studio, I usual started with importing a 2D cad drawing, from Autocad.

Just a few remarks:
I don’t think it’s even remotely true that “nobody” models in Blender to scale. Many do - and should do, unless we’re talking about modeling microbes or star systems. While Blender is very much size agnostic when it comes to modeling, it isn’t when it comes to rendering (e. g. light falloffs and intensities). And if you want to combine assets later in a shared scene, having all objects at the same (I’ll admit: not necessarily real-world) scale is paramount.

You can’t compare NURBS booleans to polygon booleans. NURBS booleans are a major CAD workhorse, as they are so immensely powerful and more or less fire and forget. Booleans in polygon based 3D applications on the other hand are something nobody really likes to use, because they will usually create horrible geometry, which will then create shading artefacts in the render. So, using booleans in polygon modelers usually is an intermediary step that has to be thoroughly cleaned up manually afterwards.

The point is: NURBS modeling mostly frees you from even thinking about geometry issues, while polygon modeling is all about geometry and polygon flow. That’s two completey different modeling paradigms and mindsets - and following workflows.

Even with plugins Blender will never be a piece of CAD software and isn’t intended to be.

Blender is not a CAD software, it’s a polygonal modeler, and you’re doing it wrong. CAD modeling paradigms are for interfacing with the real world, polygonal modeling is for visual purposes. Unlike absolute precision in CAD programs, polygonal modeling is approximation and precision is relative.

In the visualization world of polygon/sub division modelling, if it looks right, it is right.

If you want a polygon modeller where it is easy to enter dimensions and sizes you could give sketchup a try.

Thank you guys, you’re answers are really helping me, and about the scale, that was the answer I was hoping for.

This means for me I will have think more like I did in dedit, and let go for some part my way of cad thinking.
( which is mostly about makeability, machining and in Catia even stress calculation. )

Dedit the map making program for Aliens vs Predator 2.

A map I ones made, were I imported textures from two startrek games into dedit.

Also the remark about booleans, is really helping me, I had not even realized they are not the same as the ones I am more familiar with.

I will look for better tutorials, I already found one about making a bearing.

This kinda answer all my other questions, about how to do things in blender.

I am looking forward to making some spaceships in blender, as requested by somebody, for a game. (only static models)
He asked me for advice on what program to use, and I pointed him to blender.

You have CAD background and that’s much worse than being completely new to 3D when switching to polygonal modeling. For one many with that aren’t open to an idea that they don’t actually know what they’re doing and can be quite arrogant. “I’ve used SolidWords for years so I know what I’m doing”. And second hardest thing for many has been to let go about the idea that everything needs to be precise.

Since you seem more open, I’ll try add a bit more because the tutorial you linked to isn’t actually that helpful. It just somehow gives you hope that you can stick to what you already know, with false comfort. It’s bad because it makes you focus on things that don’t actually matter, and take your time away from the things that do.

Most uses for polygon models are about something visual. The models are made for still images, animation, games, 3D printing, or other. The only time one cares about absolute precision is when the model is turned into a real object such as with 3D printing, and even then the tolerances are ridiculous compared to high precision machining from CAD side of things. The model structures for those targets can be very different, they can be different within those general cases, and is very important so the model works for the intended end use.

Polygonal modeling is approximation of curves and curved surfaces. Curves are described with straight edges, and curved surfaces with polygons. There goes the precision. But also the refinement algorithms that are used to get visually more precise model also use approximation schemes, so absolute precision is thrown out the window again just to make sure it was properly defenestrated.

2.3 Real-World Scale
2.3.1 Real-world scale within 1-3% – Model can use any units to achieve real-world scale. If the model does not have an exact real-world counterpart (such as a human character or an unbranded car), the model must use the size/scale of comparable objects in real life.

2.3.2   Exception  for exceedingly large/small models – Models of objects that have a  real-world scale at a microscopic or astronomical level, such as amoebas  and solar systems, are excepted from having real-world scale.

Final models are in real world scale, so they’re the correct size relative and in proportion to other models in other scenes, and would work accordingly with things that need a dimension. That’s within 1-3% when there’s certification requirements for a model like above. Things like refraction with physically based rendering needs thickness on a glass for example. No one cares if the glass is 1mm thicker than in a real car, but it matters it’s in the right ballpark to look like glass.

The precision with polygonal models is relative because of visual reasons. Relative to

  • Distance from camera. If things are further away, things don’t have to be as accurate as on the foreground
  • Resolution. Amount of pixels for a detail, especially with textures, can be a limiting factor
  • Other objects, proportion. You can see something is off when you have something to compare to, so if a 10,00 cm high can is 10,50 cm, it doesn’t matter unless you place a 10,00 cm can next to it. The same as having a measuring device that is off, couldn’t tell until comparing it to the actual size. Unless the ruler claims its units are in km, could smell something fishy there.

There’s so much to learn. Requirements from the end use and from the pipeline stages before the model is ready

have general understanding of the available tools so you can make things your own and learn more as you go. Getting to understand topology which is about structures related to forms and proportions, and/or functions like deformation, which in turn can lead to study of anatomy.

Interpreting forms and proportions, understanding required structures, and knowing about a subset of tools to get that structure adds up to a workflow that you plan ahead, and then start modeling. Won’t be like that for a while, it comes with experience. In the end, everyone models differently, even though the end result is similar or same as with another modeler.

See this thread for CAD tools added in an SVN but still a WIP

happy bl

Thank you guys, and yes I am open to having to let go of my cad ways.

Its a bit scary, I am used to selecting a function, and then the cad program wants me to type a value.
Opening blender and messing a bit around feels beyond weird for me, there is much more freedom.

You also answered a question, I had not even asked about 3D printing.
I was already wondering about that.

I am a bit lost on the bearing tutorial reply.
If going from a 2d sketch is not the blender way, although possible, then what would be the blender way to make that bearing ?

Or are you referring to the fact that for games and such this much detail ( precision ) is over kill ?

I have a friend who is a mechanical engineer like me, but he also used to make technical 3D (perspective) illustrations by hand.
I remember him explaining, that you needed to draw things sometimes a bit to big, because that looked better, instead of doing to a precise scale.

I have not watched the full you tube vids yet, but I will.
Looks like exactly what I need to get starting.

Then I might just for exercise have a go at that drum made in catia.

That looks interesting, but for now I first want to learn the standard Blender.

Apart from actually using Blender tools, there’s no Blender way. There are many ways to do things, not one right way, or a Blender way. People also ask about best ways to do things, there isn’t one, not with the amount of information they know to provide. There are requirements and then options to fulfil those requirements.

The tutorial doesn’t specify why it shows one way of using Blender to achieve “a degree of precision”.

If it’s for 3D printing, yes, that’s one way of approaching it. But I suspect the actual reason for the tutorial is a theoretical exercise. The tutor has a technical background and has an access to CAD tools, even if he didn’t have an access to online tools, didn’t want to use open source CAD offerings, and maybe even has a CAM software that can intepret meshes for hobby CNC machining.

Even so, another way is to use a vector graphics program to draw the profiles and import them in Blender, or just use curves in Blender to make them

That gives the advantage of having a separate control over the resolution along and around with a slider, and put that extra resolution on the curved bits where it matters.

And third way is to use primitives and booleans. There’s even a handy addon called BoolTools to help manage boolean operations, making the workflow fast. 3D print models don’t care about pretty mesh structures, they only need to be manifold. There are other considerations though, like tolerances, overhangs, which are the same regardless of the way the thing is modeled.

If it’s for rendering or other visual purpose, you don’t need to model to a degree of precision as it were. It only has to look right, which can mean handling some dimensions to have the proportions right, but being very exact only slows you down. It’s much more important to focus on the density of the mesh so the silhouette looks right, and not do something if it’s not needed. For example if the bearing is rendered with the cover and clip on, you don’t actually need the bearings nor interior profiles. It could be more important to have much simpler mesh that can be UV unwarapped easily and textured, so the exterior has the right looking markings on.

Or as another example. If you for some reason are using Blender to make a 2D blueprint, the things you see on the blueprint are important, not the exact dimensions of the model. If you have a cross section of a part with outside dimensions of 5 x 15 mm, it doesn’t matter if the model is 5.030 x 14.800 meters, since it will look proportional and the texts 5mm and 15mm on the blueprint itself is what makes it exact.

There’s a saying “if you’re not faking it, you’re doing it wrong”. While modeling for visual purposes is a lot about smoke and mirrors, and absolute precision doesn’t have much value, it doesn’t mean you can be sloppy. It means solving visual problems the smart way and knowing when there’s the need to be exact.

Thank you, now it is clear to me.