Enjoy! (Best appreciated in HD.)

Oh and check out the art of this concept artist that worked on it.

Check it out: http://CGjokes.com/

It all started when I wrote the very first joke (the one at the very bottom.) I wondered if I could think of more, and so I did, but then I challenged myself to come up with a one-liner every workday or so. (Quality varies as it’s hard to write brilliant ones all the time, but I try.)

I looked online for more material but only found a really old thread from CGTalk, and many of their jokes were worse than my worst. I figured, if there’s sites out there listing trigonometry or Physics jokes, why can’t we have one more just for CG-related jokes? And so it came to be.

I will eventually run out of puns so I’d like to invite you, yes you, to think of hilarious material and submit it on the site. I will review any and all that come in and publish those that make me smile or chuckle; with your name as the post author of course. (If you leave a link to your blog I’ll link you if you get posted.)

We humans work numbers in what’s known as the “Numeral system“, also known as the “base 10″ system (as we have that many fingers…) but hold on… the alphabet has 26 letters, not 10, so what do we call that system? It has a name: it’s the Hexavigesimal (or “base 26″) system.

As the ever-so-handy Wikipedia denotes,

Any number may be converted to base-26 by repeatedly dividing the number by 26.

Pretty easy stuff for Python.

This is how I’d do it:

 
def base26str(i, padding = ""):
	"""
	Turns a number to a base26 string, with optional padding support.
	"""
 
	alphabet = map(chr, range(65, 91))
	alphaStr = ""
 
	if i > 0:
		number = i
		while (number > 0):
			remainder = number % 26
			alphaStr = alphabet[remainder] + alphaStr
			number /= 26
	else:
		alphaStr = "A"
 
	# Pad to X digits?
	if str(padding).isdigit():
		alphaStr = alphaStr.zfill(int(padding)).replace("0",alphabet[0])
 
	return alphaStr
 
# ------------------------------------
 
# Check this out...
print base26str(4)		# "E"
print base26str(4, 3)		# "AAE" ("E" padded to 3 letters. A = 0)
print base26str(28)		# "BC"
 
#

Now let’s put it to use in a simple renamer example. This sample below is Softimage-oriented, but you could easily change it to PyMEL:

from math import ceil
xsi = Application
 
typeSuffix = {
	"crvlist":	"CRV",
	"polymsh":	"GEO",
	"surfmsh":	"GEO",
	"null":		"NULL",
	"root":		"RT",
	"bone":		"BN",
	"eff":		"EFF"
}
 
oSel = xsi.Selection
prefix = xsi.XSIInputBox("What would you like to prefix these objects?", "NAME PREFIX", oSel(0).Name )
 
 
# Find out the padding required:
padding = ""
if oSel.Count > 26:
	padding = int( ceil( oSel.Count / 26.0 ) )
 
for i, eachObj in enumerate(oSel):
 
	# Get suffix if it has been defined, else leave blank:
	suffix = ( "" if eachObj.Type not in typeSuffix.keys() else "_"+typeSuffix[eachObj.Type] )
 
	# Rename object:
	newname = prefix+"_"+ base26str(i) +suffix
	print eachObj.Name+" --RENAMED-TO--> "+newname
	eachObj.Name = newname
 
#

Then highlight the bit of code below and drag it to your bookmarks bar on your browser.

javascript:location.href='http://chart.apis.google.com/chart?chs=300x300&cht=qr&chl='+encodeURIComponent(location.href);

Next time you want to open a page you’re browsing in your phone, press it, a QR code shows up, scan it and the url will be on your phone.

This one involved geometry tracking (PFTrack) for full head replacement. FaceRobot was used in production for the first time. The original plan was to use the automatic lipsyncing feature (introduced in Soft 2011) to save us time. It did work, but the client wanted 1:1 lip performance with what the actor did so in the end we didn’t use much of the automatic lipsync. — See for yourself how close the final is to the source footage here. Hats off to the animators.
 

I rigged the popcorn/kernel guy and helped set up the rigidbody simulation of the M&M’s pouring at the end. — Also wrote a little Python script to do a fair randomization of all 6 colours. You’d run it after the sim was happy to make sure the colours were random but in equal amounts.
 

I did the ICE instances setup for the lights and the rig of the lumpy bottle at the end. — By the way, the environment is 100% CG. There was a ton of rotoscoping done to make it work. There’s something like 60 shots, almost all shot against blackness (because if it were green/blue-screen it would have shown up in the reflections.)
 

Hope you likey.

Here’s a fun trick for parsing an ICE string attribute, identifying the letters and typing out the text with particles:

Download the sample scene here and read more about it below. (Soft 2011+ required.)

Taking this knowledge we can also do variations on the concept like turning a string into an array of booleans, for example:

Let me know in the comments if you make anything cool with this technique.

The classical way of doing symmetrical shapekeys (a.k.a. “blendshapes” if you’re in Maya or morphs if you’re in Max) is to make a symmetrical one then add it twice and paint off its influence with two weight maps, one for each half. In Maya you’d use the Paint Blend Shape Weights tool, but it’s basically the same deal.

In ICE, shapekeys have a .positions ICEattribute in them. This is an array of positions (vectors) representing the relative offset from the original mesh. In other words, it means you need to add them to the existing point positions to make said shape occur. That said, we won’t be using .positions in this post but full PointPosition’s instead.

Here’s what we’re gonna be working with in my example:

Normally you would make the 2 weightmaps, select one then Shape->Modulate Shape Key with Weight Map. The map looks kinda like this:

This involves making sure you have 100% of one half and 50% of the middle. You want 50% because you want to be able to add the shapes for both sides and get 100% of the deformation in the middle, else you’d get 200% and that’s no good.

Let’s see how we can use ICE to skip that step altogether by only deforming points in one half instead of relying on a map to define it, as pictured at the beginning of this post.

So as you can see we read the PointPosition, split out to scalars with the “3D Vector to Scalar“, then feed the X to different nodes to see if it’s bigger or equal to / bigger / smaller or equal to / smaller than zero. With the “Select Case” we use the correct one as defined from our dropdown menu input attribute “Type“. (If you didn’t know: to define a dropdown menu for an attribute in a compound just rightclick it and go to Properties, then fill in “Combo string” and “Combo value” then click “Add combo” as desired.)

The If node before the Linear Interpolate makes sure that if we’re not dealing with the correct half, we just let the positions pass through untouched. Otherwise, we check if the “Type” (mode) is set to one that includes the middle line, and if X is 0 it means we have a point in the perfect middle of the mesh, in which case we only affect it by 50%, else we let it go through. At the end we do one more Linear Interpolate between the affected result and the original mesh, so we can blend it gradually with a slider.

Lastly, if anyone’s curious what the word “Epsilon” means in regards to an “Equals” node, it’s basically the margin of error permitted in the comparison since vector values aren’t exact numbers most of the time in 3D. Thus an epsilon of 0.001 in this case means that a vector’s posX can be off from 0 by 0.001, so it could anywhere between -0.001 and 0.001 and still be considered to equal 0. You can increase the epsilon if you need to.

For those unaware, the Apple keyboard has no PrintScreen key. (On another note, they have F13 to F19 keys where normal PC keyboards stop at F12.) You could download a screenshot program of course, but they’re always convoluted. Too many options when all you truly want is the simple key.

So if you’re in Windows with an Apple keyboard, or know a friend who is, download this and pass it on.

It’s just 2 lines of AHK:

; F15 = PrintScreen tool / http://darkvertex.com
#NoEnv
#NoTrayIcon
#Persistent
#SingleInstance ignore
SendMode Input
 
~F15::
Send {PrintScreen}

There’s no interface, install or anything. You run it and it’ll sit idle in the background. Whenever you hit F15 on your shiny white Apple keyboard it will trigger the PrintScreen key. Simple as that! (To stop it, kill the process manually.)

“One Artist A Day”

As the name implies, I intend on mentioning 1 (likely awesome :p) musician or band I genuinely enjoy, every single day for the forthcoming future.

It’s been almost a month so far and it’s been a fun challenge to mention artists while trying to stay away from mainstream names. After all, I want to show you music you haven’t heard before, not what’s pumping on MTV or some lousy Top40s’-like radio show. (I miss when MTV had music in it; don’t you?)

So… check out my new blog! I will of course continue to keep this one and as soon as production slows a little where I’m working at I will be posting a backlog of educational content I have been meaning to submit for a while. (I currently have 9 drafts of coolness in my WordPress waiting for you.)

Last but not least, I welcome any music suggestions if anyone’s reading this and want to leave a comment down below. I promise to give you credit and link to your blog if I use your suggestion.

Here’s my take on said function for both Softimage and Maya…

First of all, we need to know the math behind it. Knowing global position vectors A and B, the formula for the distance (d) between them would be:

d = √  Ax-Bx² + Ay-By² + Az-Bz²  

I’ve decided to borrow sqrt() and pow() functions from Python’s math library, by the way.

First let’s see how to do this with Python in Softimage:

xsi = Application
lm = xsi.LogMessage
 
def ctr_dist( objA, objB ):
	from math import sqrt,pow
 
	Ax, Ay, Az = objA.Kinematics.Global.Transform.GetTranslationValues2()
	Bx, By, Bz = objB.Kinematics.Global.Transform.GetTranslationValues2()
 
	return sqrt(  pow(Ax-Bx,2) + pow(Ay-By,2) + pow(Az-Bz,2)  )
# -------------------------
 
 
# Try it out with 2 selected objects:
fromObj, toObj = xsi.Selection(0), xsi.Selection(1)
 
lm(
"Distance between <"+fromObj.FullName+"> to <"+toObj.FullName+"> is: "
+str( ctr_dist(fromObj, toObj) )
)

Now let’s repurpose it for PyMEL in Maya, without using distance nodes:

from pymel.core import *
 
def ctr_dist( objA, objB ):
	from math import sqrt,pow
 
	Ax, Ay, Az = objA.getTranslation(space="world")
	Bx, By, Bz = objB.getTranslation(space="world")
 
	return sqrt(  pow(Ax-Bx,2) + pow(Ay-By,2) + pow(Az-Bz,2)  )
# -------------------------
 
 
# Try it out with 2 selected objects:
sel = ls(sl=1)
print "Distance between <"+sel[0]+"> to <"+sel[1]+"> is: "+str( ctr_dist(sel[0],sel[1]) )

By the way, note the awesomeness of Python for letting you assign multiple variables from an array using a single line. That’s pretty nice.

Hope somebody finds this useful! You can use it a measuring tool or for distance-dependent rigging logic.

If you do use it for rigging, please remember that if you want the distance to stay the same while a character is scaled, you must divide the distance by the scaling value.