Blender Journals

April 2018 retrospective

So that was April 2018. When a Toronto cop showed the world how it’s done. When the Koreas finally started talking about ending their war. When the Australian financial sector’s dodgy behaviour was finally exposed for all to see. When.. uh… something of significance happened in Perth probably, not sure what.

I got a couple of new synthesiser bits this month, including the venerable complex LFO TINRS Wobbler and a portable enclosure with a built-in power supply. The week that the enclosure began its journey over from Melbourne, my car decided once and for all that my modular synthesiser was getting far too much attention. There soon came an ominous and expensive rattling noise from the engine…

If my chances of making it to Blender Conference 2018 were slim before, my car’s repair bills have now nailed that particular coffin shut, set it ablaze and launched it into the sun. Foo.

Blendery things

The Blender Institute began their super duper code sprint up in Amsterdam and the daily builds of Blender 2.8 are getting super duper interesting as a result! I’ll hold off from making any remarks because it’s all very much a work in progress, but there’s some promising developments.. as well as a reignition of the old LMB/RMB select debate.. ah well, may as well take the good with the bad!

Animation Bootcamp

I finished up CG Cookie’s Animation Bootcamp this month. This was my first submission for Exercise 11, the first walk cycle of the course!

I got some feedback from Instructor Wayne on this submission and had another crack at it.

Not great, true, but not awful either. The second full-body walk cycle was a little better.

It needs some polish but my peers at CG Cookie decided it was good enough to pass as is. I’ll come back to both of these another day.

I went on to the next CG Cookie animation course, Animation Workflow and Body Mechanics, and now that I’ve got a better eye for when something looks off it’s taking much longer to get my exercises to the point where I’m happy to submit them. Knowing when something’s off is easy, knowing how it’s off and how to fix it efficiently is what I’ve yet to learn. Better stick with it!

A pig

Out of the blue one Friday, I decided to challenge myself to realise a character design because it’s been a while since I tried. I did up this orthographic reference in Krita.

A flat pig

Then I tried to sculpt it. It came out looking different but so much better:

Oh hi!

Here, Mr Pig is an unretopologised sculpture. The animation is done by shape-keying a deforming lattice (head and cheeks) and the sculpted mesh (eyebrows). The freckle texture is done with vertex painting. The animation is extremely limited – he can’t open his mouth or blink yet, for instance. Still, cute pig!

The miracle of lattices!

I’m retopologising the sculpture to a mesh so that I can use it to practice UV unwrapping and surfacing, specifically texture painting. And maybe a little bit of animation too. We’ll see!

That’s all for April. If you’ve got any comments or questions or just want to rap, you can reply to this post’s toots/tweets on Mastodon or Twitter.


One of the people helping me get through making this film works in feature animation. After confessing that I’d lost perspective on whether what I was making was any good, he told me that I should stop trying to perfect things at one stage and push something forward into blocking – just so I didn’t feel like I was perpetually stuck in script rewrites, animatic tweaking, etc. There’s one sequence that’s been around for months which isn’t going anywhere, so that’s the one I’m going to try to block out.

Before I can do a layout/blocking pass, I need to do some modelling and rigging. The Pointy rig’s been a glitchy off-model piece of crap for many months so I decided to redo him from scratch. I also need some low def environment models like grass, rocks, that kind of thing.

It occurred to me that I can greasepencil certain character bits instead of building them as rigs, at least for blocking. It may turn out that a hybrid greasepencil/3D character blocking technique is more trouble than it’s worth but I’m optimistic. At any rate, we’ll all find out by tomorrow evening…


Gronky is still somewhat lumpen but he’s getting better every day. 🙂


I’m posting this image as a permanent reminder that I have a deceptive amount of free time at 6pm on a Sunday evening, even when I think I don’t.

So! Here’s Gronky’s modelling sheet with a big cheesy grin for good measure. It looks pleasingly like he is tapping himself on the shoulder. As you can see, he is now a large vaguely egg-shaped doofus as opposed to a large polony-shaped doofus. The hunch pokes his head forward and gives him this involuntary look of enthusiasm.

The basic design is a few weeks old now, from the first week of July even, but I’ve sat down tonight and refined it a bit. It might prove impossible to model, and even if it can be modelled it’s might prove impossible to animate, but it’s just a guide ultimately.

This week 21 – 27 July was not a hugely creative week. I was mostly either playing the highly addictive Mexican space sim Kerbal Space Program so that I can teach my nephew about astronautics or taking in a couple of movies.

This week’s movies were Orson Welles’s feverish 1958 film noir Touch of Evil and the 1967 school drama To Sir, With Love starring Sidney Poitier. ToE was great – almost proto-Coen Brothers with respect to certain scenes and characters. TS,WL was good in its own way, an enjoyable way to pass a Saturday afternoon.

Most of what I got up to this week I don’t want to show off – it’s either too unpolished and ugly or, in the case of what I did at this year’s “do what you like but get it done within 24 hours” marathon at work, exercising discretion about my employer’s identity. 🙂

It’s become something of a tradition for me to present some animation for this particular annual event. I made the company mascot in Blender last year, so this year I fixed the rig up and made a ten second video of it busting a move to some music we had playing. The shot was fully motion tracked off some video from a Samsung Galaxy S3.

I had a bit of trouble getting rid of the large tracking reference I put on the carpet and shadow catcher geometry seemed more important to sell the CG element, so the mascot ended up dancing on one of the team signs. Also the lighting wasn’t matched, the reconstruct was just good enough but the perspective was munted.. but it only had to go for ten seconds and get a laugh.

So I rendered and composited to a PNG sequence overnight, for there was no time to do comp/render separately without sacrificing sleep – I decided if it sucked, I just wouldn’t show it.

It didn’t suck. It was pleasantly well received, got a laugh at the project presentations, and a few people didn’t even tweak that the mascot was done in CG – they thought it was Peter Dinklage in a pint-sized mascot suit or something. I said we couldn’t get Peter Dinklage so we rebuilt that corner of the office at double scale instead and used a normal actor. Point being: basic motion matching and shadows go a long way to sell crappy everything else.

Incidentally, the Lens Distortion values for the S3 now that I’ve had time to figure them out appear to be in the neighbourhood of K1: 0.099; K2: -0.144 – or thereabouts. 🙂

So that was my week. This coming week I’ll have that modelling sheet to work off and I can start getting Gronky into virtual form.


I was sculpting Ktish but the head was coming out all.. woobly and not good. What to do? Lattice that stuff back into submission at once!

Lattice deform is the process of applying a lattice to a mesh in order to warp it into a different shape. The lattice is a grid of points; the points in their default configuration represent “start” positions, and wherever you move those points to is a “destination” position. When applied to the sculpt via a lattice modifier, mesh around any given “start” position warps towards the “destination position”, and this way you can non-destructively deform the sculpture.

It’s great for coarse form experimenting, correcting sculpt shapes that aren’t right, and any task where you want to move around a heap of polys at once (e.g. in a sculpted mesh where dynamic topology has been used) without having to hit up Edit Mode and faff around with proportional editing.

Related are Mesh Deform (instead of a square lattice you apply an arbitrary “cage” mesh over the top of something – slower to set up but arguably finer control) and the new Laplacian Deform which comes from hooks and science.

Just mentioned this technique on Twitter so I thought I’d show it as pictures too. 🙂