I‘ve spent countless hours learning Blender over the past few months (being completely oblivious of other 3D software to that point), mostly through many excellent Youtube videos like from BlenderGuru, and I think it is the most wicket piece of software I have tried to learn in decades. It makes learning the fundamentals of vi a piece of cake. It has hundreds of features, dozens of property dialogs, and most of all it requires a very peculiar interaction style that is very different from 2D editors like the ones found in Office/Google docs. The difference is similar to switching from a notepad/Word paradigm of being in edit mode by default to switching to the command mode in vim.
The most important learnings so far have been:
- Instead of creating objects from scratch, the paradigm is to modify shapes, starting from a number of primitives. That’s why you get the “default cube” when you start it. An important operation is extrude, which copies the selected parts of a mesh and connects the copies to their original. So if you want to draw a curve, you place a default curve and then use an operation like extrude or subdivide to add new points.
- There is a heavy interaction between mouse and keyboard. Depending on the task, it is vital to learn the keyboard shortcuts by heart. Those are often single keystrokes - e.g. tab for switching between object and mesh edit mode, 1/2/3 for switching between vertex/edge/face selection mode, g for moving the selected item, x/y/z to lock to an axis, x for delete, f for filling edges or faces into the mesh, e for extrude etc. while all of this is also possible to do with the mouse, there is a 10x productivity boost from learning the 10, 20 keystrokes that are used all the time.
- It is highly extensible through plugins, mostly written in Python, though, unfortunately, lower layer operations are only exposed in C++. All of the youtube videos use one of the many screencasting plugins, which display the recent mouse buttons or key presses
- it is a good idea to use a numpad, mainly to switch between different views
The benefit: even with minimal manipulations in the lighting settings and the materials of the created objects, the renderer can already create raytraces that are absolutely jaw-dropping.
But: Even after several months I still often ask myself “how do I tell it the computer”. I feel like I have barely scratched the surface of what the tool can do. Rendering pipelines? Sculpting mode? Probably 2022.
Curious about the hottest tips from the HN users :-)
Once you start learning 3D applications you realize most other UIs are toys in compassion. 3D tools are designed for professionals, who take the time necessary to learn how to be productive.
Sculpting mode is worth learning. Grant Abbitt has some great tutorials on that front. Though if you dive too far into that you may wanna look at getting a Wacom tablet or similar. Actually Grant has a lot of good tutorials in general on Blender.
I‘m more interested in geometric structures / architecture, so the sculpt mode isn‘t very useful as I realized today (after watching a video about it). This is all about taking a sphere with lots of vertices and pushing them around. I‘m looking for a way to fill gaps in meshes and deal with relative simple low-poly geometries. Ideally with semantics (a thing is placed on a floor). What should I use for that? Are there plugins thatbpush Blender towards CAD (which I know nothing about)?
You can sculpt stuff you made with geometry, you don't have to start from the sculpt sphere. But yes if you only want things to be fairly geometric sculpting is less useful, but if you want to make things like broken pillars it makes creating those effects easier.
Yeah that was why I said or similar. Maybe should have called it out more, but I got a cheap wacom (non-screen) to learn with though I haven't been doing sculpting lately, need to get back to that.
> It is highly extensible through plugins, mostly written in Python, though, unfortunately, lower layer operations are only exposed in C++.
Other than cycles all the plugins are written in python—there is a C++ api generated at the same time as the python one but afaict only cycles uses it. Haven’t done any blender stuff in quite a while so I could be (hope I’m) wrong.
Though, with a little gumption, it isn’t all that hard to expose the underlying functionality to python if nobody has gotten around to it yet. Most of my contributions were of that nature (honestly there were some big gaps and a lot of low lying fruit back then).
Interesting(?) side note: the original pie menu script was written because my laptop didn’t have a numpad and changing views is a PITA without one. People ran with it and it became really popular so they built pie menus into blender proper.
When looking (briefly) at the API documentation, I was mostly interested in mesh manipulation - and there it says "Currently, for more advanced features such as mesh modifiers, object types, or shader nodes, C/C++ must be used." (https://docs.blender.org/api/current/info_overview.html#inte...)
I've been using Blender for, ooo, I'd say about 15 years now?
I've got almost every hotkey down to a reflex, explored the software extensively, even written my own plugins for Blender. I've watched it grow over the years and man, has it come SO far.
And let me tell you... I can honestly say 15 years... I still feel like I'm barely scratching the surface.
Blender is just an infinite well of potential power and productivity. The more you put in to it, the more it will give you back.
And every few months the Blender devs put out a new update, making it even better than before. Absolutely amazing tool, would be easily worth thousands a year to me for my professional work and yet it's free. It's a 'must have' for any graphic designer/3D artist/etc in my opinion.
In my experience sculpting is much easier to learn than the vertex-twiddling side of blender. Start with a cube. switch to sculpt mode. Find the remesher and give it a poke. That'll subdivide the cube. try some of the sculpt tools, most are pretty self explanatory. Remember to poke the remesher whenever the mesh faces get too distorted.
The most important learnings so far have been:
- Instead of creating objects from scratch, the paradigm is to modify shapes, starting from a number of primitives. That’s why you get the “default cube” when you start it. An important operation is extrude, which copies the selected parts of a mesh and connects the copies to their original. So if you want to draw a curve, you place a default curve and then use an operation like extrude or subdivide to add new points.
- There is a heavy interaction between mouse and keyboard. Depending on the task, it is vital to learn the keyboard shortcuts by heart. Those are often single keystrokes - e.g. tab for switching between object and mesh edit mode, 1/2/3 for switching between vertex/edge/face selection mode, g for moving the selected item, x/y/z to lock to an axis, x for delete, f for filling edges or faces into the mesh, e for extrude etc. while all of this is also possible to do with the mouse, there is a 10x productivity boost from learning the 10, 20 keystrokes that are used all the time.
- It is highly extensible through plugins, mostly written in Python, though, unfortunately, lower layer operations are only exposed in C++. All of the youtube videos use one of the many screencasting plugins, which display the recent mouse buttons or key presses
- it is a good idea to use a numpad, mainly to switch between different views
The benefit: even with minimal manipulations in the lighting settings and the materials of the created objects, the renderer can already create raytraces that are absolutely jaw-dropping.
But: Even after several months I still often ask myself “how do I tell it the computer”. I feel like I have barely scratched the surface of what the tool can do. Rendering pipelines? Sculpting mode? Probably 2022.
Curious about the hottest tips from the HN users :-)