Skin Deep: On Projected Paintmaps and SubSurface Scattering

It seems I spend a great deal of time thanking Kent Trammell for his videos.
Perhaps I should do so more directly, I’m not sure he’s heard any of my thanks thus far, but I can’t see him on G+ so there we go. Gratitude extends only so far, and FB is far too far.

In my usual style of trying to pick and choose bits of learning from a larger programme, and thanks to a tip from +Thomas Sanjurjo, I took a look at Creating a Realistic Head in Blender over at CGCookie. Skipping over the sculpting part and getting straight into skin texturing, I fired up my Jon Stewart sculpt and plastered it initially with pixels from the low resolution photo I used as a source for the sculpt.


Plain diffuse material


Hackily projected diffuse map










Then moving onto the next video, I was instructed to ignore the delightful (snrk) colour-map that I’d created. So I put to one side the messy, low-res, hack-job of a texture and delved into the sub-surface shader world.

I had been aware of the wonders of SSS previously, but with my brain firmly planted in Cycles-land and being a thoroughly impatient fellow (CPU rendering!??! Whaaaaaa…), my first dealings with the concept were a frustrating chore.
In stark contrast to that experience I found working through the videoย to be a wonderfully methodical, comfortably predictable, easily understandable pleasure.

Single layer SSS (pretty shoddy, but there it is)

Single layer SSS (pretty shoddy, but there it is)

And at that point, my impatience kicked in once more and I jumped ahead, slinging textures on, inexpertly mixing them in the SSS settings, knowing not what I was doing. This was very silly of me, but I quite like the outcome.







Amazingly accurate image of Jon Stewart

Vaseline on the lens, but still, pretty nifty IMHO

I’m going to go back to the first couple of videos and either add detail to this model, fixing the various ways in which it’s wrong, or produce another one from scratch, because what I see thus far is quite promising, and it was easier than I thought it would be ๐Ÿ™‚

Thanks Kent!


Sculpting Continues


Hair is a bit of a let down on this one also.

I think I might be addicted to sculpting practice.
It’s not a major problem I suppose. I quite often find myself thinking about what to sculpt, how I’d do it, what it would look like, who might like it etc. and then I realise I was supposed to be doing something else. Fortunately, I don’t drive, so nobody’s in direct danger.
Anyhoo, a few more heads were attempted which you’ll find below.
I really ought to focus on materials, hair, lighting, backgrounds etc. but I burn through a lot of enthusiasm getting the face done, then I just want to see it look good.
Maybe I’ll spend some time setting up a little “head-studio”.


Hair and beard needs to be black for this to work…


Hooray! Bald! But I yawned 18,000 times while making this.


Oh my word, can someone do some free hair tutorials for me…


Another baldie thank goodness. Needs de-symmetrising in at least one spot

The Sculpture Update

Firstly, I want to say a big thanks to Kent Trammell over at CGCookie for (the first part of) The Art of Sculpting: Caricature.
I’d never tried caricature before (officially anyhow) and the way the subject was introduced and the techniques explained was exceptionally useful.

A few posts ago, I wrote about the loan of a graphics tablet from my Mum. In that post I added a picture of something I’d sculpted. It was a head. Well, it was supposed to be. Looking back on it now it looks absolutely dreadful.


Between then and now I bimbled around practicing here and there sculpting a lot of very uninteresting stuff. I got to the point where I could get something looking half decent, but was still fairly unsure of myself.

Hmm, not too bad…

Getting somewhere…

That’s the ticket.

Then I found the aforementioned tutorial. It was just around the time that I wanted to make something to show my appreciation to my brother, so I snagged one of his holiday snaps off FB and caricatured it up.

“Glad I’m not one of those fake fans that jumped ship and now can’t preorder their Xbone”

From that point, I kept on practicing, using some of the techniques showed in Kent’s tutorial and after a few tries, I think I managed to get some good likenesses without going too cartoony (not that there’s anything wrong with cartoony, I just wanted to dial the exaggeration back a bit).

“Sometimes the light at the end of the tunnel isn’t the promised land, it’s just New Jersey”
โ€œAgnostics are just atheists without balls.โ€

So I’m going to keep practicing, posting to G+, occasionally updating on this site, and hopefully I’ll see more improvement on the scale shown between the top and the bottom of this page.

On Dead Graphics Cards and Awesome Brothers

A couple of weeks ago, my GTX 560 TI Twin Frozr II became a Single Frozr I.

Gfx card with one fan missing

Single Frozr I – *cries*

I’ve set my machine to fire up Open Hardware Monitor ( on boot so I can have some pretty graphs to look at, but these days those graphs are looking less pretty and more scary.
I attempted to fix the card by gluing the fan head back on where it had sheared off. This led to some stomach churning noises, strange response from the fan speed control and all out woe.

Over the following weeks the rig hobbled along, GPU under-clocked and one fan in operation. Occasional stability issues and full on crashes aside, the only impact was that I couldn’t play fancy games or do much GPU rendering, so you could say it was “character building” in that I now had some constraints that required me to think up creative solutions to questions like “How do I waste a bunch of time without fancy 3D games?” (answer: Don’t Starve (

This all changed yesterday when my little brother handed me one of his 560 TIs out of his rig. I felt a bit bad taking it. If I had two of these little monsters I’d hold on to them for dear life, so I was incredibly grateful.

That gratitude (plus an appropriately timed look at the Blender Cookie Art of Sculpting: Caricature intro video ( led to the following sculpt of my brother, grinning like the Cheshire cat, for which I used a sneakily pilfered holiday snap from FB as a reference.

Thanks Rob ๐Ÿ™‚

My bruvva

Who put a real live toad in the ‘ole

Oh, Christmas Tree

I can highly recommend a viewing of Andrew Price’s tutorial on making a Christmas tree in Blender.
I stopped watching when I became thoroughly engrossed in what I was making using the techniques shown, but I’ve since been back to check out the part where he makes tinsel, baubles, a star for on top etc. and there’s some great tips in there too.

I’m part way through creating a scene using the tree making technique as a starting point, and I did a couple of extra tricksy things that Andrew didn’t mention in his tutorial which are as follows:

1. Rather than trying to manually weight paint the tips of the branches to scale down/remove the particles, I created a tapered cylinder that was a similar shape to the outline profile of my tree, then I used that object as the source for a Vertex Weight Proximity modifier on the branches themselves. What this allows you to do is smoothly gradiate the weight from 0 to 1 from the tips of all the outermost branches to some distance away from the surface of the surrounding object. ย I then did a similar thing but used the trunk of the tree to prevent particle growth at the very start of the branches. I think this gives a nice effect. (I’ll post some pictures of the resulting weight group later)

2. Instead of manually placing the curve for my tinsel/lights, I added a mesh circle, extrude-scaled it to make it a ring of faces, ripped one edge and pulled it up with Connected Proportional edit turned on, so that I ended up with a sort of spiral, added an array modifier to “copy” it a few times vertically so I ended up with several loops of a spiral.
Then with this loopy spirally string of faces, I added a Simple Deform modifier to taper it toward the top to a similar profile as my branches and positioned it roughly where I wanted it to be.
The next step was to grab a copy of the branches and apply a Decimate modifier to them set to a ratio of somewhere around 0.005 (I think this means it’ll try to remove 99.5% of the faces, which is roughly what I wanted) to create a really low poly tree “proxy” which I then made into a collider. I set up a cloth sim on my loopy spiral and played it for fifty or so frames, letting it jiggle about and come to rest on my tree proxy, then I grabbed one edge loop of the spiral, separated it from the mesh and converted it to a curve for beveling and conversion back to mesh for the core of the tinsel (I could have used a skin modifier, but I wanted the curve for other things too).

Here’s a 25 sample test render of where I’m up to with the scene:

Christmas Tree Test Render
25 sample test render of my Christmas scene built in Blender

I’ll post more when I get more done ๐Ÿ˜€



On the Questionable Wisdom of Sculpting Someone You Know

Sooner or later I was going to have to get serious about sculpting.

Time and again I’ve watched videos of folks squishing and sliding verts around to amazing effect, producing fantastic sculpted models in Blender (among other packages).

I really love the sculpt tools in Blender. They’re great fun to use. More so if you have a pressure sensitive input device of some kind. I’ve been practicing here and there, mostly making humanoid, although not all of them are immediately recognisable as humanoid, especially in the early days.

It struck me the other night that my imagination is pretty poor at coming up with characters to sculpt, so I had in mind the idea that I might have a go at someone who actually exists.

If you want to see some amazing sculptures of people who actually exist (or have existed), you *really should* check out Doris Fiebig’s work over at It really is something else, and while inspiring, made me a little apprehensive to try my hand at creating a likeness.

So I considered a few options and came to the conclusion that if I did a sculpt of someone that I know personally, but that isn’t an instantly recognisable celebrity (is that harsh? I didn’t mean it to be love…) then I still have somewhere to hide when it comes to “Oi! That doesn’t look like such and such…” comments.

You may have guessed from the previous paragraph that my whizz-bang idea was to sculpt a likeness of my lovely missus.

It’s altogether possible that anyone reading this who has experience of painting/sculpting/photographing loved ones will at this point be saying, “well, that was a mistake” and it’s also possible that I’d agree, but not entirely.

You see, I’m fairly familiar with the subject’s face, so that helps, and I have access to the subject most of the time, so that helps too. I wonder if she minds being called “the subject”…

The difficulties begin when I want to show off work in progress renders. Now, I think that it’s pretty good work, but I also feel that it’s got to hit a certain level before I can show it to anyone, because of my relationship with the subject, and the possibility that I might get strangled in my sleep if I’ve done a poor job of it.

I’m really, really, really enjoying it though. It’s *great fun*. Even when I screw it up and have to smudge the crap out of it to get it back to a workable state, it’s all fantastically enjoyable.

So there’re ups and downs, but I’m glad I’ve had a go at this, and, with the subjects permission, will post some renders soon.


Thanks Mum!

About a year or so ago now I’d guess, I started getting (back) into Blender in a fairly big way.
I followed tutorials and build some fairly naff stuff, but I got better I think.

One of the things that helped immeasurably was that my Mum had a Wacom Intuos 3 that she wasn’t using a great deal and, to my eternal gratitude, passed on to me.

I’ve been very thankful for that ever since, but may not have expressed it as much as I perhaps should.

What got me thinking about this was that I’ve been practicing sculpting recently and for a change of pace, decided to try doing the base mesh by hand to as great a level of detail as I could manage, with no reference images or sculpt tools allowed.

So using only the mouse and keyboard I got this far:


Then checked out what it looked like subdivided:


And it was at that point I said “right, enough is enough, lets get those sculpt tools out”, grabbed the stylus and got cracking.

About two minutes later I had this:


So, thanks Mum! ๐Ÿ˜€
I guess I’ll never know how far I’d have progressed in my work/play/practice/learning in this field without the tablet, but I can tell you with some certainty, I would very likely have given up.


Glowing Stuff!

I like glowing things. Little floaty lights make me smile.
With that in mind, I started having a play with some particle effects that I could “glow up” in the compositor.

In short, I created a plane to use as a dynamic paint canvas, set to vertex and wave mode so that the particles made ripples in the surface when they connected with it, then I made another plane above it and set it up as a dynamic paint brush, set to particle mode, and added a particle system.

I then created a small icosphere and used it as the particle duplication object, lessened the effect of gravity and cranked the random initial velocity.

I set up an emissive material for the particles and a glossy/glass mix for the canvas surface, gave the particles lots of frames to fall down, rendered it out and composited in some blurring on the emission pass and a few other bits to get the look I was after.

The result is below ๐Ÿ™‚

So after that, I was reading some bits and pieces and was reminded about the Particle Instance modifier. I had a quick play and used that, and hair mode on the emitter and ended up with some pretty nice tendrils which worked well with my materials and compositor setup in individual frames, an example of which is shown below.


What they didn’t work very well with was the dynamic paint, which I’m putting down to my settings, or the fact that most of the interpenetration of the tendrils and the canvas is edge-face rather than vertex-face. (Further info on that if requested.)

The result was underwhelming, but posted below for completeness.

How Important *IS* Hardware

Ay up.

I was listening this evening to Mr. Price’s latest podcast entitled “How Important is Hardware?” and reached the unofficial end of the piece struggling a little to maintain my normally unshakably placid demeanour .

Only kidding, I was raging, as always, without fail.
Ranty opinion piece follows.

Throughout those first ten or so minutes, while not gratefully receiving the wisdom not to walk into any old Asian massage parlour, I actually spluttered to myself “but… but… what about…”, and then held my tongue and continued listening, only for the same thing to occur moments later.

Don’t get me wrong, the overall message is one of hope, and that’s lovely. To say to all the “budding artists” that it doesn’t matter if you can’t afford high-end hardware is, for the most part, a noble platitude (or cynical coaxing? …nah…) but for one thing; it’s just not true.
The power of your rig will influence how and what you learn and how and what you produce, more or less radically, depending on the circumstances.

The *most* important thing, literally, the *most* important thing is to be comfortable using your own machine. To know the limits of your rig is to prevent disaster (lost work from crashes), to ease frustration (understand why that operation takes so *damned long*), to optimise workflow (easy on the verts there tiger) and have a happier experience all round.
It’s simply not true to say that hardware isn’t that important. It absolutely *IS* that important, just not in the way the headline might lead you to think. You don’t need the fastest clocked CPU, the largest, zippiest RAM money can buy, or even a graphics card that keeps your local power plant in business. What you need, what you absolutely must have, is hardware you’re comfortable with, and an understanding that a beefier graphics card won’t magically make you better at art (which is probably what Andrew was getting at, but isn’t the whole story).

I’m not supremely artistic.
I can’t paint worth a damn, and the best thing I’ve ever sculpted in real life is a blu-tac teddy bear. I know my limitations, and for that reason (among others) I haven’t gone out to buy oil paints. I might, one day, if I have the time to spare, but I don’t expect great things so I’m not going to invest just yet.

What this past year using Blender has taught me is that I do have a little aptitude for some things and if I practice, I might get better at them.

What I feel I have a better than average aptitude for though, is knowing what my machine and I are capable of. It doesn’t expect me to be able to spread paint on a canvas in a pleasing formation and I don’t expect it to be able to handle forty million vertices with any kind of fluidity(see footnote). It doesn’t expect me to be able to chisel a new Venus de Milo and I don’t expect it to be able to load twenty 8k textures without straining a little at the time (numbers mostly pulled from somewhere sunless, but you get the gist).

The point is, that I’ve come to find my preferred group of tools and techniques based primarily on what we two are able to do together; what I can work with, and what my machine can manage.

So the importance is this; your hardware could, and probably will, shape your usage of any software.
If you’re lucky enough to have CUDA cores out the wazoo, you’ll do certain things that someone with fewer resources will not do. This can be good or bad on either side of the equation. Only the other day I saw a tutorial showing how to model a length of rope in a way that ended up spawning tens of thousands of vertices where a simple repeating normal map (or even a shader level displacement rather than the mesh level one they used) would have achieved much the same result with a tiny fraction of the memory/CPU footprint. The same person who is able to handle scenes using this technique, adding verts willy nilly, will undoubtedly also be able to do some very fine sculpting work on their rig, but if they never learn to work within the constraints of lower spec hardware, they may well miss out on some approaches that can be helpful in other situations.

That last section, as I read it back to myself, sounds a bit like “Worry not poor penniless Blenderhead, you’ll learn more than those rich folks!” which wasn’t my intention. I’m not trying to make anyone *feel* better, the poor about the experience they may gain or the rich about the freedoms they may enjoy. I simply want to say that while it’s not exactly the hardware itself that’s important, the hardware, and your relationship with it will be a major part of what will ultimately shape how you learn, what you learn, how you produce and what you produce.

Saying otherwise is to ignore some pretty obvious facts and is more than a little misleading.


Embarrassing footnote: Yesterday evening I joined together three meshes, one of which had over 30k verts, and only realised with horror in the moments thereafter that one of the meshes I’d joined had five or so levels of multires subdivision on it, and the selection order had been such that it was now attempting to subdivide my 30k+ vert mesh five times. This doesn’t invalidate my point though, because once I realised what I’d done, I had no illusions about why my machine had just ground to a halt, so there ๐Ÿ˜›

1 2 3 4